(ssa) from a consumer behaviour perspective

37 downloads 289655 Views 3MB Size Report
May 5, 2010 - through the use of the Sponsored Search Advertising (SSA) approach. ...... The Internet has three main tools: Email, Usenet and the World Wide ...
AN EXPLORATION OF CONSUMER RESPONSE TOWARDS SPONSORED SEARCH ADVERTISING (SSA) FROM A CONSUMER BEHAVIOUR PERSPECTIVE

MOHAMMAD AL KHASAWNEH B.BUS, MBA (COVENTRY UNIVERSITY)

DEPARTMENT OF MARKETING GRIFFITH BUSINESS SCHOOL GRIFFITH UNIVERSITY

Submitted in fulfilment of the requirements of the degree of Doctor of Philosophy August, 2009

ABSTRACT The Web has become a major source for information in peoples’ everyday lives. However, finding useful and relevant online information has become an increasingly complicated task for many online users, especially as information on the Web has increased exponentially, and millions of new Web pages are introduced daily into the Web environment. Consequently, Web search engines are now the starting point and major informational sources for online users in facilitating Web search activities related to both commercial and non-commercial Web sites. As such, businesses have recognised the importance and value of having their website highly ranked and visible in search engine result pages, which may be achieved through the use of the Sponsored Search Advertising (SSA) approach. SSA represents an advertising technique in which a fee is paid for specific keywords to guarantee priority placement on search engine results pages. The increased market demand for more consumeroriented, targeted and non-intrusive Web advertising formats has led to the development of SSA. Such advertising offers Web advertisers the ability to more precisely target and direct Web users to their websites. Accordingly, SSA has become a critical component of companies’ marketing campaigns, with a global annual growth rate predicted to be 37 percent, and scaled rate to be more than US$33 billion in 2010 (Ghose and Yang, 2007). In addition to becoming the most prominent format of Web advertising in terms of revenue, SSA is also the major source of revenue for Web search engines, for example, 67 percent of Google’s 2008 revenue came from SSA (Google, 2008). However, the success of search engines in generating revenue using SSA depends mainly on the number of users (consumers) who click on Sponsored Search Advertisements (as advertisers are only required to pay when users click). For this reason, it is essential that online users notice and pay attention to SSA, to ensure that the desired communication goal of such advertising (click through behaviour) is achieved. Therefore, the key aspect to whether SSA is a viable business model depends, to a large extent, on the consumer and how he or she responds to this type of Web advertising. More specifically, as suggested by Jansen and Spink (2007a), SSA may be improved and become more effective if a greater understanding is gained of consumer behaviour in such advertising contexts. Interestingly, despite the phenomenal growth of SSA, scant research attention has been given to how consumers process and respond to this form of Web advertising and, more specifically, the particular factors that drive consumers to pay attention to, and respond positively, towards SSA. Additionally, most previous SSA research has generally been undertaken from the advertisers’ perspective (Bradlow and Schmittlein, 2000; Feng et al., 2003; Telang et al., 2004). At the same time, existing anecdotal studies indicate that, generally, consumers have little awareness of the practice of SSA and, consequently, most tend to ignore them or do not use them. This lack of awareness, and use, has highlighted the need for further and more extensive research. Thus, an exploration of the processes surrounding consumer response to SSA is warranted. To advance our current understanding of SSA from the consumers’ perspective, the focus here is on examining a range of consumer behaviour variables that may determine how they respond to SSA. Therefore, it is proposed that such an examination will advance the understanding of consumer behaviour within the context of SSA. As such, the primary focus of the current research is on the consumer and those associated consumer related factors that i

determine their responses to SSA. Therefore, the following research objective was considered worthy of investigation: To explore the impact of consumer related factors on consumer response towards Sponsored Search Advertising. Based on an examination of the extant literature, a preliminary conceptual Model of Consumer Response Towards SSA was presented. This proposed model incorporates a number of relationships, including a representation of consumer related factors, along with consumer attention towards SSA, attitude towards SSA and intention to click on Sponsored Search Advertisements. The current research methodology combines both qualitative (phase one) and quantitative (phase two) approaches, with semi-structured interviewing and online surveying being used. As little is known about how consumers respond to SSA (with most research focusing on practitioners’ views), an exploratory qualitative approach was initially taken in phase one of the data collection. As such, eight semi-structured interviews were conducted with Australian residents who had Internet access and who had purchased any product or service online using a Web search engine. The purpose of this phase was to explore the research objective and obtain greater understanding of consumer response towards SSA. The findings of phase one were used to clarify and confirm the appropriateness of the preliminary conceptual Model of Consumer Response Towards SSA. Based on these findings, a revised conceptual is presented. The second phase (the quantitative research) was based on the development of an online survey which allowed for the measurement of the factors surrounding consumer response towards SSA. The development of the survey followed a logical two-step process which involved utilising and adapting existing measures and then pre-testing these measures and, thus, ensuring that every effort was made to develop a psychometrically sound survey instrument. The final survey used ‘Travel Tickets’ as advertising stimulus for the study. Data collection resulted in 325 usable surveys for subsequent analysis. Analysis of the data was conducted via correlation analysis, exploratory factor analysis, reliability analysis and Partial Least Squares (PLS) regression analysis. Overall, the findings provided support for the Consumer Response Towards SSA Model. All the hypothesised paths included within the inner model were supported, and all components of the measurement model were found to be statistically significant. The findings provide a number of theoretical and practical implications for research. From a theoretical perspective, we believe this study is among the first studies to examine the impact of consumer related factors on consumer attentional processing, and attitudinal and behavioural responses toward SSA. Therefore, the study adds to and expands our knowledge of the factors that influence consumer attention towards SSA and determining how attention to Sponsored Search Advertisements, along with perceptions of SSA credibility, influence attitude and behavioural intentions towards those advertisements. Thus, the current study has applied proven theories and constructs from marketing, advertising and consumer behaviour research and has extended and validated the theoretical relationships among consumer related factors and attention, as well as the consequence of such attention.

ii

The practical implications of these findings are that they add to the understanding of SSA from a consumer behaviour perspective and, therefore, act as a valuable base for SSA practitioners. Specifically, the current study provides practitioners with insights into consumers and the factors that influence their intention to click on Sponsored Search Advertisements when they use Web search engines. Importantly, the findings identify which type of consumer (according to their experiences, subjective knowledge, familiarity with brands (or websites), and perceptions of credibility and relevancy) is more likely to attend and process SSA, and then links these to the outcomes of consumer response towards SSA, that is, attitude to SSA and intention to click on Sponsored Search Advertisements.

iii

Declaration I, Mohammad Al Khasawneh, declare that this work has not previously been submitted for a degree or diploma in any university. To the best of my knowledge and belief, the thesis contains no material previously published or written by another person except where dur reference is made in the thesis itself.

Mohammad Al Khasawneh September, 2009

iv

Table of Contents Abstract

i

Declaration

iv

Table of Contents

v

Table of Tables

x

Table of Figures

xii

Acknowledgements

xiii

Chapter One: Introduction………………………………………………….………….…..…1 1.1

Background………………………………………………………….………..….….1

1.2

Research Objectives………………………………………………….………….….4

1.3

Definitions and Terms…………………………………………….….……………..7

1.4

Justification and Importance of the Study……………………………..………....10

1.5

Research Methodology……………………………………………………..……....12

1.6

Delimitations…………………………………………………………………..…….14

1.7

Thesis Outline…………………………………………………………..…...……....14

1.8

Conclusion………………………………………………………………….....……..16

Chapter Two: Literature Review & General Propositions Development………..………...17 2.1

Introduction……………….………………………………………………………...17

2.2

The Internet, The Web, and Web Advertising…………………………………....19

2.2.1

The Internet………………………………………………………………......19

2.2.2

The World Wide Web (Web)…………………………………………….…..20

2.3

Web Advertising……………………………………………………..……………...21

2.3.1

Characteristics of Web Advertising…………………………………...……..22

2.3.2

Web Advertising Forms…………………………………………...…………24

2.3.3

Keyword Search Advertising (KSA)………………………………………...25

2.3.4

Sponsored Search Advertising (SSA)………………………………...……...27

2.3.5

Section Summary……………………………………………………….…....28

2.4

Web Advertising from a Consumer Behaviour Perspective………………….….29

2.4.1

An Overview of Consumer Behaviour in Terms of Web Advertising………29

2.4.2

Consumer Behaviour in the context of Sponsored Search Advertising……...31 v

2.4.3

Consumer Response to Advertising………………………….……….……..32

Consumer Attention towards SSA……………………………………...…………..35

2.5 2.5.1

Prior Experience with Sponsored Search Advertising (SSA)….......................38

2.5.2

Subjective Knowledge about Sponsored Search Advertising (SSA)….……...40

2.5.3

Familiarity of Brands (or Websites)…………………………………...........…43

2.5.4

Attitude towards SSA………………………………………………….……...45

2.5.5

Behavioural Intentions (Intention to Click on SSA)…………………..……....48 Theoretical Model and Propositions Development……………………...……….....49

2.6 2.6.1

Determinants of Attention towards Sponsored Search Advertising (SSA)…...51

2.6.1.1

The Relationship between Prior Experience and Attention……………..........51

2.6.1.2

The Relationship between Subjective Knowledge and Attention………….....52

2.6.1.3

The Relationship between Familiarity of Brands and Attention………………53

2.6.2

The Relationship between Attention and Attitude…………………….……….54

2.6.3

The Relationship between Attitude and Intention to Click………….…….......55

2.6.4

Section Summary………………………………………………….……………56

2.7

Conclusion……………………………………………………………………………..57

Chapter Three: Research Design and Qualitative Methodology For Phase One…..…..........59 3.1

Introduction………………………………………………………..…………..….…..59

3.2

Research Plan………………………………………………………….…………...…61

3.3

Paradigms of Research Inquiry…………………………………………….…...…...61

3.3.1

Types of Paradigms……………………………………………………...…......62

3.4

Research Design…………………………………………………………….…….......65

3.5

Choice of Semi-Structured Interviewing……………………………………..……...69

3.6

Reliability and Validity of the Interviews………………………………….………..74

3.7

Planning and Implementation of Semi-Structured Interviewing Approach……...76

3.7.1

Identifying the Information Required…………………………………………76

3.7.2

Identifying the Sampling Aspects……………………………………………..76

3.7.3

Determining the Interview Process……………………………………………78

3.8

Data Interview Analysis and Interpretation…………………………..…………....80

3.8.1

Findings from the Semi-Structured Interviews……………………………......81

3.8.2

SSA from the Consumer Behaviour Perspective-Preliminary Comments.........84

3.8.3

Determinants of Consumer Attention toward SSA and Subsequent Outcomes……………………………………………………...….86 vi

3.9

Conclusion……………………………………………………………………….….97

Chapter Four: Research Design and Quantitative Methodology For Phase Two…..…….99 4.1

Introduction……………………………………………………………………..….99

4.2

Justification of Survey Methodology for the Second Phase………………...…...100

4.3

Survey Design and Development………………………………………………….101

4.3.1

Identify the Information Needed and Operationalisation…………………..103

4.3.2

Specifying the Survey Method………………………………………………110

4.3.3

Determine the Question Wording, Content, and Structure……………….....113

4.3.4

Determine the Scale and Response Format……………………………….…114

4.3.5

Pre-Testing and Revising the Questionnaire………………………………...116

4.3.6

Reliability and Validity Considerations…………………………………….118

4.3.7

Administering the Questionnaire Online……………………………………120

4.4

Sampling Strategy…………………………………………………………...……..122

4.5

Anticipated Data Analysis…………………………………………………………122

4.6

Ethical Considerations…………………………………………………….……….124

4.7

Conclusion………………………………………………………………………….125

Chapter Five:

Analysis and Results of Survey Data…………………………………..…126

5.1

Introduction………………………………………………………………………..126

5.2

Data Preparation……………………………………………………………….….128

5.2.1

Editing the Data……………………………………………………….……128

5.2.2

Data Coding………………………………………………………………...128

5.2.3

Transcribing the Data……………………………………………………….128

5.3

Sample Profiles……………………………………………………………………..129

5.4

Preliminary Data Analysis……………………………………………………..….132

5.5

Preliminary Results……………………………………………………………......134

5.5.1

Preliminary Data Analysis – Prior Experience…………………..………….135

5.5.2

Preliminary Data Analysis – Subjective Knowledge……………………….136

5.5.3

Preliminary Data Analysis – Familiarity of Brands (or Website)………......137

5.5.4

Preliminary Data Analysis – Sponsored Search Advertising Relevancy…...138

5.5.5

Preliminary Data Analysis – Sponsored Search Advertising Credibility…..139

5.5.6

Preliminary Data Analysis – Attention to Sponsored Search Advertising…139

5.5.7

Preliminary Data Analysis – Attitude towards Sponsored vii

Search Advertising………………………………………………….……140 5.5.8

Preliminary Data Analysis – Convergent Validity…………………………141

5.5.9

Preliminary Data Analysis – Discriminant Validity………………………..142

5.5.10

Preliminary Data Analysis – Common Method Variance………………….143

5.5.11

Summary of Preliminary Analysis………………………………………….144 Data Analysis Using Partial Least Squares (PLS)……………………………….145

5.6 5.6.1

Structural Equation Modelling (SEM)………………………………….......145

5.6.2

Partial Least Squares (PLS)…………………………………………………146

5.6.3

Overall Model Results………………………………………………………150

5.6.3.1

Measurement Model Results……………………………………………….150

5.6.3.2

Structural (Inner) Model Results…………………………………………...153

5.6.3.3

Summary of Results – H1 to H10………………………...………………...157

5.7

Conclusion…………………………………………………………………………159

Chapter Six: Discussion………………………………………………………………………..160 6.1

Introduction………………………………………………………………………..160

6.2

Conclusions about the Hypotheses………………………………………………..164

6.2.1

Results of Hypothesis One (The Effect of Prior Experience with SSA On attention Towards SSA)………………………………………………..167

6.2.2

Results of Hypothesis Two (The Effect of Subjective Knowledge of SSA On Attention to SSA)……………………………………………………...168

6.2.3

Results of Hypothesis Three (The Effect of Familiarity of Brands or Websites on Attention towards SSA)……………………………………...170

6.2.4

Results of Hypothesis Four (The Effect of SSA Relevancy on Attention Towards SSA)……………………………………………………………...171

6.2.5

Results of Hypothesis Five (The Effect of Familiarity of Brands or Websites on SSA Relevancy)……………………………………………....172

6.2.6

Results of Hypothesis Six (The Effect of Prior Experience on SSA Credibility………………………………………………………………….173

6.2.7

Results of Hypothesis Seven (The Effect of Attention Towards SSA on SSA Credibility……………………………………….................................175

6.2.8

Results of Hypothesis Eight (The Effect of SSA Credibility on Attitudes Towards SSA)……………………………………………………………...175

6.2.9

Results of Hypothesis Nine (The effect of Attention to SSA on viii

Attitude Towards SSA)……………………………….......................……177 6.2.10

Results of Hypothesis Ten (The Effect of Attitude towards SSA on Intention to Click on SSA)……………………………………………...…178

6.3

Implications………………………………………………………………………..180

6.3.1

Theoretical Implications…………………………………………………....180

6.3.2

Practical Implications…………………………………………………...….185

6.4

Limitations…………………………………………………………………...……189

6.5

Future Research……………………………………………………………..…....190

6.6

Conclusion…………………………………………………………………..…….193

Appendix A…………………………………………………………………………….…195 Appendix B……………………………………………………………………………….196 Appendix C………………………………………………………………………………..199 Appendix D..........................................................................................................................202 References…………………………………………………………………………………....204

ix

Table of Tables Chapter Two: Literature Review and General Propositions Development 2.1

Research Propositions…………………………………………………….…58

Chapter Three: Research Design and Methodology Phase One 3.1 3.2 3.3 3.4 3.5

Principles of the Research Paradigms………………………………………62 Comparison between Qualitative and Quantitative Methods……..………67 Comparison between Unstructured, Semi-structured and Structured Interviews………………………………………………………………..…...74 Summary of Results from Semi-structured Interviews………….….…..…82 Research Hypotheses.......................................................................................97

Chapter Four: Quantitative Methodology 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11

The information needed for this Research including Research Questions & Related Constructs………………………………..……........103 Adoption of Existing Measures for the Research Constructs….…..........104 Examples of Prior Experience with SSA Scale…………………....…..….106 Examples of Subjective Knowledge of SSA Scale……………...…….…..107 Examples of Familiarity of Brands (or Websites) included in the SSA Scale………………………................................................…………...107 Examples of SSA Credibility Scale…………………….....…………….....108 Examples of SSA Relevancy Scale…………………………………….......108 Examples of Attention towards SSA Scale …………………...………….109 Examples of Attitude towards SSA Scale…………………………...…... 109 A Comparison of Survey Methods……………………………………......111 Changes Made to Draft Survey Items for Final Survey…………………118

Chapter Five: Analysis and Survey Results 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10 5.11 5.12 5.13 5.14 5.15

Demographic Profile of the Sample………………………....……………129 Web Usage Profile……………………………………………………........130 Search Engine Usage Profile………………………………………….......131 Preliminary Data Analysis-Prior Experience with SSA…………….......136 Preliminary Data Analysis- Subjective Knowledge of SSA……….....…137 Preliminary Data Analysis- Familiarity of Brands (or Websites) included in the SSA………………………….............................................138 Preliminary Data Analysis- SSA Relevancy…………………………......138 Preliminary Data Analysis- SSA Credibility…………………………….139 Preliminary Data Analysis- Attention towards SSA………………........140 Preliminary Data Analysis- Attitude towards SSA………………….….140 Average Variance Explained & Reliability Estimates………………......142 Component Loadings (Reflective) for the Measurement Model…..…...151 Correlations among Constructs & Scores ……………………………...153 Partial Least Squares Results for the Theoretical Model ……………...155 Results of Hypotheses Testing…………………………………………....157

x

Chapter Six: Discussion 6.1 Results of Hypotheses Testing…………………………………………...165

xi

Table of Figures Chapter One: Introduction 1.1

SSA Example when ‘travel’ is the Search Term in Google…………..……8

Chapter Two: Literature Review and General Propositions Development 2.1 2.2 2.3

Outline of Chapter Two……………………………………………….…….18 Preliminary Conceptual Model of Consumer Response Towards SSA..50 Preliminary Conceptual Model of Consumer Response Towards SSA.....57

Chapter Three: Research Design and Methodology Phase One 3.1 3.2

Outline of Chapter Three……………………………………………..……60 Revised Conceptual Model of Consumer Response towards Sponsored Search Advertising (SSA) Following the Semi-Structured Interviews...........................................................................98

Chapter Four: Quantitative Methodology 4.1 4.2

Outline of Chapter Four…………………………………………….….…100 Questionnaire Design and Development Process……….…………..........102

Chapter Five: Analysis and Survey Results 5.1 5.2

Outline of Chapter Five…………………………………….………….….127 Proposed Model with Results of Analysis…………………..…………....158

Chapter Six: Discussion 6.1 6.2

Outline of Chapter Six…………………………………….……………....163 Model of Consumer Response Towards Sponsored Search Advertising(SSA).……..............................................................……...........167

xii

Acknowledgements I have been inspired on my learning journey by a number of people who contributed meeting and overcoming this challenging research experience. My thesis is the end product of all those endeavours.

First of all, I had a great fortune to study under the supervision of Dr. Deborah Griffin and Dr. Scott Weaven. My sincerest appreciation is given to my primary supervisor, Dr Deborah Griffin, for her encouragement, advice, direction, support, availability and patience during the doctoral process. She has been a true source of inspiration and motivation for me over the past few years and I truly value her help. For these reasons, she has earned my greatest respect and I will remain forever grateful.

My sincerest appreciation also goes to my other primary supervisor, Dr. Scott Weaven for his confidence in my choice of the topic and his ongoing and insightful comments. His profound level of knowledge and experience, along with his amiable personality has enabled me to make the significant progress required to complete this thesis. I am also very grateful for his friendly support and enthusiasm.

I am also grateful for the service and support I received from other staff in the Department of Marketing at Griffith University. Many thanks go to Dr. Arthur Sweeny for his guidance and advice in the initial stage of my doctoral thesis. My sincerest appreciation goes to Dr. Debra Grace, for her assistance and support in the last stages of this thesis research. I would like also to express my gratitude to the departmental secretary, Ms. Caroline Frost, for her help and support over the past few years. I also thank Ms Carmel Wild for her editorial support.

This thesis has also benefited greatly from the valuable input given by many fellow postgraduate students. So many have provided support and assistance over the past years that I cannot thank all of them by name. However, I would like to give special thanks to Dr. Mitchell Ross, Dr. Kelly Bodey, Dr. Rahim Hussain, and Sam Al-naimi; thank you all for your support, encouragement and friendships.

xiii

My most special appreciation and thanks go to my family members. Their commitment to my education, and support for my life and career choices, has made prosper and develop into the person I am today. I am deeply indebted to my father, Hamdi Al Khasawneh, for inspiring me in my quest for knowledge and for supporting me financially and emotionally; I am also thankful for my mother’s compassion and prayers. Mom and Dad, I am truly grateful for your investment in my education and personal development, I could not have done it without you... I love you. I also would like to thank all my brothers and sisters, nieces and nephews, who have believed in my ability and patiently waited for me.

Most of all, I thank God for giving me the health, stamina, and ability to complete this project.

I dedicate this thesis to my father, who has always inspired me to challenge myself and who has taught me that education equals a better future.

xiv

CHAPTER ONE INTRODUCTION 1.1

Background

The Internet represents one of the most successful technological revolutions in the twenty first century and has changed consumers’ (users) and businesses’ patterns of communicating, purchasing and advertising (Berners-Lee, 2006; Danaher and Mullarkey, 2003; Eighmey, 1997; Gallagher, Foster, and Parsons, 2001; Palanisamy and Wong, 2003; Yao and Lin, 2006; Yoon and Lee, 2007). On the one hand, the Internet’s global and versatile nature has attracted many users from around the globe. For example, according to the Internet World Statistics (2009) estimate, there are currently almost 1.6 billion online users with access to the Internet worldwide. On the other hand, the growth in the number of online users has created greater opportunities for businesses to improve their positions in the marketplace (Berners-Lee, 2006; Ghosh, 1998; Lazer and Shaw, 2000; Piers and Aisbett, 2001) with global e-commerce revenues are expected to reach US$267 billion by 2010 (Knight, 2008). More specifically, the technological capabilities of commerce via the Internet has provided businesses with an ability and potential to disseminate information, and promote and sell products and services to a wider target market at lower costs (Berners-Lee, 2006; Palmer and Griffith, 1998).

The Internet has three main tools: Email, Usenet and the World Wide Web (Web). The World Wide Web (hereafter Web), in particular, is a medium with unique multimedia and hypertext capability features which are accessible through millions of networked computers around the globe (Eighmey, 1997; Gallagher et al., 2001; Hoffman and Novak, 1996; Palanisamy and Wong, 2003; Yao and Lin, 2006; Yaveroglu and Donthu, 2008). Given the Web’s increasing technological capabilities and extensive geographical reach, many organisations consider the Web to have more marketing potential than other Internet tools (that is, Email and Usenet), especially in relation to promoting products and services at minimal cost and with precise targeting (Chatterjee, Hoffman, and Novak, 2003; Ghose and Dou, 1998; Leong, Huang, and Stanners, 1998; Palanisamy and Wong, 2003; Ravi, 2005; Singh and Dalal, 1999; Yaveroglu and Donthu, 2008). As such, it is the Web and its adaptation for commercial and marketing activities that has provided the impetus for this study.

1

As the Web has evolved as a mainstream commercial medium, Web advertising revenues have increased exponentially, reaching US$23.4 billion in 2008 in the United States of America, representing a more than a 50 percent increase from the 2005 revenues (IAB, 2009), with revenues are expected to reach US$81.1 billion worldwide by 2011 (EConsultancy, 2007; Jaffray, 2007). This dramatic growth in Web advertising has increased the need for a greater understanding of Web advertising and its effectiveness in achieving the desired communication goals.

Web advertising began in 1994 with the presentation of banner advertisements on the Hotwired Website (Barnes, 2003; Kaye and Medoff, 2001; Robinson, Wysocka, and Hand, 2007). Since then, advertisers have considered the Web as one of the most important marketing communication channels because of its distinctive features of interactivity, flexibility, customisation, accessibility and tracking capabilities (Bush, Bush, and Harris, 1998; Chatterjee et al., 2003; Li and Leckenby, 2004; Palanisamy and Wong, 2003; Richard and Chandra, 2005; Robinson et al., 2007; Yaveroglu and Donthu, 2008). The term Web advertising covers many types of advertising from electronic advertisements, that are similar to traditional advertisements (e.g., billboards, banner advertisements), to formats that are distinct from traditional advertisements (e.g., pop-up advertising) (Schlosser, Shavitt, and Kanfer, 1999). In addition, Web advertising formats include the original banner advertising, skyscrapers, pop-up advertising, interstitials, online video advertising, rich media, corporate websites, and keyword search advertising (Briggs, 1999; Chandon, Chtourou, and Fortin, 2003; Cho and Cheon, 2004; Choi and Rifon, 2002; Flores, 2001; Li and Leckenby, 2004; Robinson et al., 2007; Yao and Lin, 2006).

Although banner advertisements have been the dominant format of Web advertising for several years, the revenue from banner advertisements has declined steadily since the advent of other Web advertising formats, such as Keyword Search Advertising (KSA) (Li and Leckenby, 2004). Recently, KSA has proven to be the main, and fastest growing, form within the Web advertising market, growing from one percent in 2002 to 41 percent in 2006 (Maddox, 2006). Simply put, KSA is a process by which search engines place advertising in the search results pages of certain keywords and advertising companies pay fees to the search engine providers when users click on a link to their web sites (Burns, 2005; Schmidt and Patel, 2004).

2

Of the various forms of KSA, Sponsored Search Advertising (SSA) has been identified as the most effective form (Delaney, 2004; Harrison, 2005). SSA represents an advertising technique in which a fee is paid for specific keywords to guarantee priority placement on search engine results pages (Feng et al., 2003; Lu and Chau, 2006; Overture, 2005). The increased market demand for more consumer-oriented and targeted advertising and nonintrusive Web advertising formats (Weidlich, 2002) has led to the development of SSA because it offers Web advertisers the ability to more precisely target and direct Web users to their websites (Dou, Linn, and Yang, 2001; Ghose and Yang, 2007; Jansen and Mullen, 2008). Accordingly, SSA has become a critical component of companies’ marketing campaigns, with a global annual growth rate predicted to be 37 percent, and scaled at more than US$33 billion in 2010 (Ghose and Yang, 2007).

For the online user, SSA offers highly relevant search results (Kiritchenko and Jiline, 2008), which are based on the consumer’s own queries and, thus, they are considered less intrusive than banner advertisements or pop-ups advertising (Ghose and Yang, 2007). In addition, SSA reduces online user search costs (Jansen and Mullen, 2008) and increases the accessibility to useful information within a limited time frame (Lu and Chau, 2006). Consequently, SSA has become an important element of online users browsing and information searching experiences on the Web (Richardson, Dominowska, and Ragno, 2007).

Previous research relating to SSA has focused primarily on the advertisers’ perspective. For example, SSA and search engine performance (Bradlow and Schmittlein, 2000; Telang, Boatwright, and Mukhopadhyay, 2004), quality uncertainty and adverse selection in SSA (Animesh, Ramachandran, and Viswanathan, 2006), SSA ranking strategies (Feng, Bhargava, and Pennock, 2003), and the implementation of SSA strategies (Weber and Zheng, 2003). Another area of SSA research has focused on SSA auction mechanisms, from the economic perspective (Edelman et al., 2007; Edelman and Ostrovsky, 2007), for example, the best bidding strategies for advertising companies to maximise profits (Animesh et al., 2006; Chakrabarty, Zhou, and Lukose, 2007; Coy, 2006; Feng, 2008; Feng et al., 2003; Kitts and LeBlanc, 2004; Mehta et al., 2005;Varian, 2006), as well as decentralising and finding alternative mechanisms for SSA auctions (Khare and Sittler, 2005). However, recently, academicians have called for further research into specific aspects of SSA, especially research related to the consumers’ point of view (Animesh et al., 2006; Li and Leckenby, 2004). Moreover, this need to establish a new direction for future SSA research was 3

identified by Ghose and Yang (2007) and Li and Leckenby (2004), who suggest that little understanding currently exists on, how consumers process and respond to SSA.

Therefore, understanding the factors that influence consumers to attend to and/or respond to Sponsored Search Advertising would seem an important and necessary step in any future research. This is because existing research suggests that SSA should be noticed and attended in order to motivate users to click on such advertisements (Lu and Chau, 2006). Further, Lu and Chau (2006) noted that, “as sponsored link advertising (SSA) has become one of the most important businesses for leading e-commerce companies, it is crucial to determine the effectiveness of sponsored links (SSA) on attracting users’ attention” (p.2). In addition, Li and Leckenby (2004) have questioned the outcomes of consumers’ attention towards SSA.

The above discussion has focused on the gap in the literature pertaining to the factors driving consumer response towards SSA. More specifically, an exploration of SSA, from a consumer behaviour perspective (Faber et al., 2004), has not been examined previously in the context of Web advertising and, therefore, further research in this area is warranted.

1.2

Research Objective

The background discussion provides a brief overview of Sponsored Search Advertising (SSA) and highlights the need to further our understanding of this particular area of Web advertising in the context of consumer behaviour. Therefore, the following research questions were considered worthy of investigation, with the broader primary research objective, addressed in this study, being:

To explore the impact of consumer related factors on consumer responses towards Sponsored Search Advertising. To advance our current understanding of SSA from the consumers’ perspective, the focus here is on examining a range of consumer behaviour variables that may determine how they respond to SSA. Therefore, it is proposed that such an examination will advance the understanding of consumer behaviour within the context of SSA. As such, the primary focus of this research is on the consumer and those associated consumer related factors that determine their responses to SSA. Extending Rodgers and Thorson’s (2000) contentions, consumer responses to SSA may be viewed as comprising three elements: determinants, 4

process and outcomes. Determinants are those factors that may determine consumer’s attention towards SSA. Process element stands for the consumer’s initial processing of the SSA (Webb, 1979) which is represented in this research by consumer attention to SSA. The final element, consumer responses, involves outcomes, namely those factors that result from consumer attending to SSA. These factors may be characterised as the consumer’s attitude towards SSA and intention to click on Sponsored Search Advertisements. Therefore, the broader research objective is further elaborated into a number of research questions which are presented below:

RQ1: What are the consumer related factors that account for consumer attention towards SSA and to what extent do they influence attention towards SSA? RQ2: To what extent does consumer attention towards SSA influence consumer attitude towards SSA? RQ3: To what extent does consumer attitude towards SSA influence consumer intentions to click on SSA? In addition, another research question was developed based on the results of the qualitative phase of data collection conducted in this research (Chapter Three):

RQ4: To what extent does credibility of SSA influence consumer attitude towards SSA? To address these research questions, relevant areas in the extant literature, within the domains of Web advertising and consumer behaviour, were examined and are discussed in Chapter Two. Particular focus was given to the Sponsored Search Advertising literature within the consumer behaviour context. As discussed previously, this area of research is pertinent as there has been little empirical research on the context of SSA from a consumer behaviour perspective (Ghose and Yang, 2007; Li and Leckenby, 2004). In addition, although recent studies have highlighted the phenomenal growth of SSA (for example, Animesh et al., 2006; Ghose and Yang, 2007; Jansen and Mullen, 2008), research in this area has generally been from the advertisers’ perspective (Bradlow and Schmittlein, 2000; Feng et al., 2003; Telang et al., 2004). However, there appears to be recognition of the need to understand Web advertising effects from a consumer behaviour perspective given that the resultant findings may provide important contributions to the Web advertising literature (Faber et al., 2004). In addition, existing anecdotal studies indicate that, generally, consumers have little awareness of the practice of SSA and, consequently, most tend to avoid them (Fallows, 2005; 5

Greenspan, 2004; Jansen and Resnick, 2005; Jansen and Resnick, 2006). In view of this lack of awareness, and use, the need for further research is obvious. Thus, an exploration of the processes surrounding consumer response to SSA is warranted.

As little is known about how consumers respond to SSA (with most research focusing on practitioners’ views), an exploratory approach was initially taken to identify the core issues involved with the phenomenon under investigation (Neuman, 2003). The exploratory approach is also an effective method of discovering variables not detailed in the extant literature, thus adding new dimensions to the study (Hannabuss, 1996; Jarratt, 1996; Patton, 1990). In addition, the scant consumer behaviour literature relating to SSA has neglected many important factors that may influence processing and behavioural responses towards such advertising (Animesh et al., 2006; Ghose and Yang, 2007; Jansen and Resnick, 2006). Therefore, the aim of this research is to make a contribution to the body of knowledge regarding consumer responses towards SSA through an in-depth qualitative and quantitative analysis that not only recognises the importance of this prominent Web advertising format, but also enables theoretical understanding of how consumers respond and process SSA.

Accordingly, a combination of qualitative and quantitative approaches was used in this research. An examination of the literature assisted in the development of a preliminary conceptual model and a general set of propositions explaining consumer behaviour associated with Sponsored Search Advertising. This was followed with phase one of the data collection process in which exploratory semi-structured interviews were conducted with Australian online users (who had searched online for products and services) to gather insights about how consumers respond and process SSA. The phase one findings were used to clarify and confirm the appropriateness of the conceptual Model of Consumer Response Towards SSA. Based on these findings, a revised conceptual Model of Consumer Response Towards SSA was presented.

In the second phase of the data collection, a sample of the Australian population (who have online access and who have searched or purchased online) were surveyed using an online self-administered survey. The resultant data were analysed to test the revised Model of Consumer Response Towards SSA. The results build upon the qualitative stage of the research, providing empirical evidence of consumer response in the context of SSA.

6

In brief, qualitative and quantitative methodologies were critical to advancing our understanding of consumer behaviour associated with SSA and, in particular, the determinants of consumer attention towards SSA (including prior experience with SSA, subjective knowledge of SSA, familiarity of brands (or websites) included in the SSA, SSA relevancy and credibility) and the outcomes of consumer responses, which are characterised by consumer’s attitude towards SSA and their intention to click on Sponsored Search Advertisements.

1.3

Definitions and Terms

To establish an understanding of the foundation of the constructs used in this research, and to reconcile any differences in definitions, fifteen key terms of this research are defined below.

The Internet. The Internet is a worldwide, publicly accessible network of interconnected computer networks and individual networks (Bickerton, Bickerton, and Pardesi, 1998; Forrest, 1999). It transmits and communicates data using the standard Internet Protocol (IP) (Forrest, 1999) through email, data file transfer, forums, and many other applications. The Internet has three main tools: Email, Usenet and the Web (Forrest, 1999).

World Wide Web (Web). The Web is an abbreviation of the World Wide Web and refers to a hypertext medium and represents a “… user friendly graphics-capable component of the Internet” (Palanisamy and Wong, 2003, p.15). Moreover, the Web is considered the fastest growing subset of the Internet, with the greatest potential for new marketing opportunities by providing information to Web users in various ways, including text, audio, images video, and sound (Eighmey, 1997; Gallagher et al., 2001; Palanisamy and Wong, 2003; Yaveroglu and Donthu, 2008).

Search Engine. A Web search engine is an information retrieval system that is designed to search for information on the Web in which the search results are usually presented as two listing categories: paid listings and natural free listings. These search results may consist of Web pages, images, documents, information or other types of online files (Bar-llan, 2005; Jansen and Mullen, 2008).

7

Online / Web Users. This term refers to all individuals with access to, or who make use of, the Web for any purpose. They may or may not make a shopping purchase on the Web (Jansen and Mullen, 2008).

Web Advertising. Web advertising is Web content paid for by an identifiable advertiser with persuasive intent to achieve marketing communication goals (Cho and Leckenby, 1999; McMillan, 2004; Yoo and Stout, 2000). The advertising includes many forms of commercial content available on the Internet, from electronic advertisements, that are similar to traditional advertisements, to formats that are distinct from traditional advertisements (Robinson et al., 2007; Schlosser et al., 1999; Yao and Lin, 2006). Keyword Search Advertising (KSA). KSA represents “… an advertising technique on the Web which uses search systems to target advertisements to the appropriate audience” (Schmidt and Patel, 2004, p.1). Specifically, KSA displays a firm’s text advertising by matching a site’s content with the advertiser’s keyword[s] (Burns, 2005; Cheek, Kunz, and Osborne, 2001; Vaughan-Nicholas, 2003).

Sponsored Search Advertising (SSA). SSA is a KSA form in which a fee is paid for specific keywords to guarantee priority placement on search engine result pages (Feng et al., 2003; Overture, 2005). As the fee is paid when the users click on the links provided, it is also referred to as pay-per-click advertising (Feng et al., 2003). Also, SSA listings are usually identified by search engines as sponsored links which are located at the top (in shaded colours) of the natural free results or vertically down the right side of the search results pages. Figure 1.1 shows a SSA example from Google search engine.

Figure 1.1

SSA Example When ‘Travel Tickets’ is the Search Term in Google

8

Prior Experience with SSA. Prior experience with SSA refers to information learned and conclusions drawn from prior experiences with SSA that functions both as an important feedback mechanism as well as determining how likely is it that the consumer will engage with the advertising in the future (Cho and Cheon, 2004). Subjective Knowledge of SSA. Subjective knowledge of SSA is defined as the consumer’s perception of the amount of knowledge of SSA that they have stored within their memory (Flynn and Goldsmith, 1999).

Familiarity of Brands (or Websites) included in the SSA. Familiarity of brands (or websites) in the SSA is defined as the level of importance attached by consumers to familiar brands (or websites) included in the SSA (Simonin and Ruth, 1998).

Sponsored Search Advertising Credibility. SSA credibility represents online users’ perceptions of the truthfulness and believability of the Sponsored Search Advertising in general (Mackenzie and Lutz, 1989).

Sponsored Search Advertising Relevancy. SSA relevancy concerns the degree to which the SSA and its keywords are pertinent, applicable and related to a consumer’s search needs (Lastovicka, 1983).

Consumer Attention towards SSA. Attention towards SSA refers to the extent of cognitive resources a person devotes to the Sponsored Search Advertisement (Muehling, Stoltman, and Grossbart, 1990; Laczniak and Muehling, 1993; Laczniak, Muehling, and Grossbart, 1989).

Consumer Attitude towards SSA. Attitude towards SSA is defined as a learned predisposition to respond in a consistently favourable or unfavourable manner, towards SSA during a particular search session (Holbrook and Batra, 1987; Pollay and Mittal, 1993).

Intention to Click on Sponsored Search Advertisements. Intention to click on Sponsored Search Advertisements is defined as the likelihood that a consumer will engage in a clicking behaviour on Sponsored Search Advertisements (Shamdasani et al., 2001).

9

1.4

Justification and Importance of the Study

Research regarding the way in which consumers process and respond to Sponsored Search Advertising is an area worthy of academic research. A number of practical and theoretical grounds emphasise the importance of this research, justify its focus and highlight its potential contribution to the domains of SSA and consumer behaviour. Importantly, the Web has become a major source for information in peoples’ everyday lives (Bar-llan, 2005; Jansen and Spink, 2006). However, as information on the Web has increased exponentially, and as millions of new Web pages are introduced daily on the Web environment (Lorigo et al., 2006), finding useful and relevant online information has become an increasingly complicated task for many online users. Consequently, Web search engines have become the starting point and major informational sources for online users in facilitating Web search activities related to commercial and non-commercial Web sites (Jansen and Molina, 2006; Spink et al., 2006; Sullivan, 2003). For example, 94 percent of online users rely on search engines to find information and locate Websites (Nielsen-Net Rating, 2006). Moreover, Web searching has become the second most popular online activity after sending and receiving emails (Pew Internet and American Life Project, 2005). This being the case, businesses have recognised the importance and value of having their website highly ranked and visible in search engine result pages (Jansen and Mullen, 2008), which may be achieved through the use of the SSA approach. More specifically, SSA was developed to accommodate the new needs of consumers, advertisers and Web search engines.

Consequently, Sponsored Search Advertising had become the most prominent format of Web advertising by 2003 in terms of revenues (IAB, 2004), as well as the major source of revenues for Web search engines (Jansen and Resnick, 2006). For example, 67 percent of Google’s 2008 revenue came from SSA (Google, 2008). However, the success of search engines in generating revenues using SSA depends mainly on the number of users (consumers) who click on Sponsored Search Advertisements (as advertisers are only required to pay when users click) (Vragov, 2009). For this reason, online users need to notice and pay attention to SSA, so that the desired communication goal of such advertising (click through behaviour) may be achieved (Lu and Chau, 2006). Therefore, the key aspect to whether SSA is a viable business model depends, to a large extent, on the consumer and how he or she responds to this type of Web advertising. More specifically, as suggested by Jansen and

10

Spink (2007a), SSA may be improved and become more effective by gaining a greater understanding of consumer behaviour in such advertising contexts.

Interestingly, despite the phenomenal growth of SSA, scant attention has been given to how consumers process and respond to this form of Web advertising and, more specifically, the particular factors that drive consumers to pay attention to and respond positively, towards SSA. Therefore, given that most previous SSA research has examined consumers’ avoidance and its associated determinants, it appears that extending our understanding of what factors determine consumers’ attention towards SSA is worthy of further investigation. In addition, as noted by Elpers, Wedel and Pieters (2003), consumers need to be exposed to, and pay attention to an advertisement before further responses can take place. Thus, this research will provide better insights into the outcomes of consumer attention towards SSA and how that attention may influence their responses.

The previous discussion (highlighting the importance of research in Web advertising and consumer behaviour domains) provides preliminary justification for the current research, especially as it explores SSA in the context of consumer behaviour. The following statement of Li and Leckenby (2004) strongly supports the need for further research in the context of Sponsored Search Advertising: “Keyword search (Sponsored Search Advertising) is superior over other internet advertising formats in that it delivers relevant commercial information at the moment when users need it. As a result, users are less likely to consider it intrusive. Unfortunately, little research is available about the effectiveness of this format of internet advertising at present” (Li and Leckenby, 2004, p.19).

The current study is not only justified on the basis that research in the areas of SSA and consumer behaviour is important to both practitioners and consumers, it is also justified by the paucity of consumer research that currently exists in the area of SSA. As previously discussed, most SSA research has been conducted from the advertiser’s perspective. As such, the empirical examination of the conceptual Model of Consumer Response Towards SSA will enhance our understanding of the ways in which consumers’ process and respond to SSA. Accordingly, there is strong justification for the current study because of the theoretical and practical contributions it will make to extending knowledge in the context of SSA, from a consumer behaviour perspective. 11

From a theoretical perspective, the knowledge obtained through this research will enhance our understanding of the factors influencing consumers’ attention towards SSA as well as the outcomes resulting from such attention. From a practical perspective, the findings will assist in providing a greater understanding of consumer related factors that are important for determining consumer attentive processing of SSA; the outcomes of such attention along with perceptions of SSA credibility are expected to include reasons for positive attitudes towards SSA and for higher intention to click on Sponsored Search Advertisements. Thus, the knowledge gained from this research will provide a clearer understanding of the determinants and outcomes of consumer response towards SSA from a consumer behaviour perspective.

1.5

Research Methodology

As previously discussed, qualitative and quantitative methodologies incorporating semistructured interviews and an online survey were used to collect data for the current research. Specifically, following a review of the extant literature, a series of qualitative semi-structured interviews were conducted to identify and organise observable facts and explore the under researched topic of interest. Next, a quantitative (online survey) approach was used to collect quantitative data, which was subsequently analysed, via correlation analysis, exploratory factor analysis, reliability analysis and Partial Least Squares (PLS) regression analysis. These methodologies are introduced briefly in this section, then detailed more comprehensively in Chapters Three and Four.

Qualitative Methodology (Phase One). Exploratory research should be first undertaken when little is known about the research issue (Zikmund, 2003). Therefore, in the current research, the exploratory phase was used to assist in providing background information about the research objective, to explore patterns and ideas that could be tested later (Burns and Bush, 2003), and to validate the general propositions (Malhorta, 1999). This phase of data collection involved a series of semi-structured in-depth interviews in which data was collected and analysed to explore the research objective. The flexibility of the semi-structured interview method allowed the researcher to obtain in-depth information (Partington, 2001) which was used to provide a greater understanding of the variables related to the research (Nair and Riege, 1995).

Eight semi-structured interviews were undertaken with a sample of Australian residents who had Internet access and who had previously purchased products and/or services online (in the 12

last 12 months) using a Web search engine. That is, respondents were identified on the basis of their involvement and experience with online searching and purchasing. New aspects of interest were identified from the eight exploratory interviews that were audio-taped, transcribed and analysed manually. From these findings, a revised conceptual Model of Consumer Responses Towards SSA was presented.

Quantitative Methodology (Phase Two). The second phase of the research was based on the development of an online survey questionnaire which allowed for the measurement of the focal constructs. These included determinants of consumer attention towards Sponsored Search Advertising (prior experience with SSA; subjective knowledge of SSA; familiarity of brands (or websites) in the SSA; SSA credibility and SSA relevancy), and associated outcomes (attitude towards SSA and intention to click on Sponsored Search Advertisements). The data collection method was chosen because of its ability to measure the focal constructs as well as its suitability for Web advertising and consumer behaviour studies (for example, Bell and Tang, 1998; Burns and Lutz, 2006; Gordon and Lima-Turner, 1997; Shamdasani et al., 2001; Shim et al., 2001).

The development of the survey followed a logical two-step process. In the first step, items were generated from the extant literature relating to offline traditional and online advertising and consumer behaviour. The second step entailed conducting focus groups to assess the content and layout of the survey as part of the pre-testing of the draft survey. By following this procedure, every effort was made to develop a psychometrically sound survey instrument to address the hypotheses presented in the current research.

The administration of the survey followed a Web-based self-administered method. The data collection involved the researcher employing an Australian market research company (The Final Prospect Company) to send an email, on behalf of the researcher, to Australian online users inviting them to participate in the study. The email included a link to a unique Web site location to access the survey. In addition, the email informed potential respondents about the purpose of the research, the length of the survey, and provided assurance that the collected information would be treated confidentially and would be used only for the purposes of the research. Analysis of the data was conducted via correlation analysis, exploratory factor analysis, reliability analysis and Partial Least Squares (PLS) regression analysis.

13

1.6

Delimitations

The nature of the research meant that some delimitations of the scope had to be set. Firstly, the study was limited by the choice of the advertising stimulus of “Travel Tickets”, the search term of reference for which respondents registered their responses. However, “Travel Tickets”, as a stimulus for this research, was chosen following careful examination of the most popular product and service category online (Google AdWords Tool, 2008), as well as by surveying individuals who represented the population of interest regarding the most highly searched and purchased product or service online. Nevertheless, the application of the findings to other product or service categories (that share common characteristics with the chosen stimulus) should be attempted with care.

Secondly, as is the case with all data collected via a convenience sample, the extrapolation of the findings to a larger population also should be attempted with care. Thirdly, employing only eight interviewees in the first data collection phase may affect the validity of the qualitative findings. For this reason, larger qualitative sample may be employed in future research to enhance the validity of the qualitative results. Fourthly, the study is delimited by the particular variables under investigation, which were found in the existing literature and results from the qualitative stage of this research. However, although these variables may not have been exhaustive, they provide a starting point to explore the domain of SSA from a consumer behaviour perspective. Any further research should perhaps include other consumer related variables or incorporate advertiser related factors that may influence consumer response towards SSA. Finally, the use of self-reported behavioural measures may be less valid than actual behavioural measures, as the respondents were asked to report their behavioural intention to click on Sponsored Search Advertisements instead of recording their actual clicking behaviour. Nevertheless, this study reported acceptable validity results, which indicated that the use of the self-reported measure of clicking intention was not problematic. While delimitations exist in this study, they do not, in any way, mar or render the significance of the findings of this research.

1.7

Thesis Outline

This research used a six chapter structure adapted from Perry (1998). This chapter provides an overview of the research by identifying the topic of interest, discussing the research objective, providing a justification for the research, describing the proposed methodological approaches and presenting the delimitations of the study. 14

Chapter Two builds a theoretical foundation for the study by reviewing the relevant literature, including offline traditional and online advertising, as well as consumer behaviour literature, with the particular focus being on consumer responses towards advertising studies. The key issues, and a range of variables associated with understanding consumer behaviour in the SSA context, are identified and discussed. From this basis, a preliminary conceptual Model of Consumer Responses Towards SSA and the general propositions are introduced.

Chapter Three presents the methodological approaches and research design adopted for the study. More specifically, it justifies the choice of the realism research paradigm. Then the qualitative methodology (first phase of data collection in this research), involving exploratory semi-structured interviews, are discussed. Following this, the process involved in planning, managing and implementing the semi-structured interviews is detailed. Based on the findings of the eight semi-structured interviews, the proposed hypotheses and revised conceptual Model of Consumer Response Towards SSA are detailed.

Chapter Four details the methodology of the quantitative methodology (phase two of data collection) and the survey development process. More specifically, the reasons justifying the choice of an online survey, as the most appropriate method of data collection to obtain the required information and to test the research hypotheses, are presented. In addition, a discussion of the survey design and the development process, the sampling selection procedures, the anticipated data analysis and the ethical consideration is provided.

Chapter Five presents the results of the data analysis for phase two. Initially, a profile of the respondents is presented, followed by the results of the preliminary analysis, including correlation analysis, exploratory factor analysis and reliability analysis. Having established the validity and reliability of the measures, the results addressing the hypotheses are presented.

Chapter Six provides a detailed discussion of the important research findings, in terms of the Model of Consumer Response Towards SSA. Then, a number of theoretical and practical implications are identified, and finally, the limitations of the study and the implications for future research are presented.

15

1.8

Conclusion

This chapter has detailed the foundations of the current research. Firstly, a brief overview of the research topic was presented and the key issues within the Web advertising and consumer behaviour literatures were outlined. This led to the presentation of the research objective, followed by the importance and justification for the current research. Key definitions of the focal constructs of the research were then presented, and the proposed methodology adopted was briefly discussed. Finally, the structure of the thesis was outlined prior to identifying the delimitations and boundaries of the research.

16

CHAPTER TWO LITERATURE REVIEW AND GENERAL PROPOSITIONS DEVELOPMENT 2.1

Introduction

Chapter One identified the need for the current research in filling a visible gap in the Web advertising literature; it also clearly outlined the research objective and associated research questions. Chapter Two reviews the relevant literature on Web advertising and consumer behaviour so as to provide a foundation for the theoretical framework developed in this research to assist in gaining an understanding of the process involved the consumers‟ responses towards Sponsored Search Advertising (SSA). Therefore, this chapter is designed to provide a context for understanding consumer behaviour within the SSA arena.

This chapter is divided into five sections, as shown in Figure 2.1. The first section provides some background about the Internet, the World Wide Web (Web) in general, and introduces the discussion into Web advertising in particular. Next, the relevant literature of Web advertising, especially, keyword Search Advertising, is examined in section 2.2. The third fourth, and fifth sections explore the consumer behaviour domain with a particular focus on a range of variables (for example, consumer attention and attitudes towards SSA, and their intention to click on SSA), that may be relevant in the context of consumer response towards SSA. In section, 2.6, a number of general propositions are developed, detailing consumer response towards SSA, from the consumer behaviour perspective. The final section provides a summary of this chapter (section 2.7).

17

Figure 2.1

Outline of Chapter Two

2.1

Introduction

2.2

The Internet and the Web

2.3

Web Advertising

2.4

Web Advertising from a Consumer Behaviour Perspective

2.5

Theoretical Model and General Propositions Development

2.6

Conclusion

Source: Developed for this research

18

2.2

The Internet, The Web, and Web Advertising

To understand and provide a comprehensive foundation for the current research, the literature regarding the Internet and the World Wide Web was reviewed with a particular focus on Web advertising research. This section is comprised of five parts. The first part gives a brief overview of the Internet, its development, and importance. Next, the World Wide Web (hereafter “Web”) and its marketing implications are reviewed. The third part identifies and discusses the importance and characteristics of the Web within the advertising context. More specifically, it presents a discussion of the evolution of Web advertising as a source of revenue, as well as serving the needs of consumers. The final part provides some background of SSA through a review of the existing literature.

2.2.1 The Internet The Internet was developed in the early 1970s by the Advanced Research Project Agency (ARPANET) (Schneider, 2006). ARPANET was a research and defence network created in the United States (US) and its aim was to develop a communication tool that would be robust on the occasion of military attacks or sabotage (Schneider and Perry, 2001). Later, the Internet was used predominantly as an academic and research tool in the fields of science and education whereby it was designed to connect universities and research centres (Cailliau, 1995). The Internet has grown rapidly since its evolution in the early 1990s to become a worldwide, publicly accessible network of interconnected computer networks and individual networks (Bickerton, Bickerton, and Pardesi, 1998; Forrest, 1999); it transmits and communicates data using the standard Internet Protocol (IP) (Forrest, 1999). This latter innovation has allowed individuals to use the Internet for maintaining communication through email, data file transfer, forums, and many other applications. It has also functioned as a new business environment where companies and consumers may browse, sell, and buy different products and services (Laudon and Traver, 2004; Yao and Lin, 2006). Thus, the Internet has become an important part of everyday life for millions of people around the world (Ko, Cho, and Roberts, 2005; Shamdasani et al., 2001; Yoon and Lee, 2007), and it has attracted businesses due to significant opportunities offered within the digital realm.

The Internet has three main tools: Email, Usenet and the Web (Forrest, 1999). The original component of the Internet is Email which refers to the process of exchanging electronic messages and computer files between computers that are connected to the Internet (Forrest, 1999). With Email, users gain a low cost and fast delivery method of national and 19

international communication (Godin, 1995). Alternatively, the first tool of the Internet to unify like-minded users was the Usenet which is a worldwide network of discussion groups allowing users to share and exchange ideas about various topics (Forrest, 1999; Settles, 1995). The third tool of the Internet emerged in 1992, called the Web (Berners-Lee, 2006; Hoffman and Novak, 1996); it is a hypertext media and represents a “… user friendly graphics-capable component of the Internet” (Palanisamy and Wong, 2003, p.15). Moreover, the Web is considered the fastest growing subset of the Internet and provides the greatest potential for new marketing opportunities by providing information in various ways including text, audio, images video, and sound (Berners-Lee, 2006; Eighmey, 1997; Gallagher et al., 2001; Palanisamy and Wong, 2003). This dramatic growth of the Web provides clear justification for research into this emerging arena. Due to its importance, the Web is discussed in greater detail in the next section.

2.2.2 The World Wide Web (Web) The application of the Web, due to its unique features and multimedia and hypertext capabilities, has enabled the Internet to be customised and has enhanced the way for commercial activity to be undertaken on the Internet (Berners-Lee, 2006; Eighmey, 1997; Gallagher et al., 2001). Given these increasing technological capabilities, many organisations consider the Web to have more marketing potential than other Internet tools, especially in promoting products and services because of its minimal costs and precise targeting capabilities (Leong et al., 1998; Palanisamy and Wong, 2003; Yaveroglu and Donthu, 2008). In addition, the Web is recognised as an effective advertising channel that may be used to inform and persuade consumers, create brand awareness, improve consumer attitudes and generate revenues (Ghose and Dou, 1998).

However, the Web presents organisations with still-to-be-met opportunities and challenges and, consequently, there have been calls for more empirical research in this area (Ducoffe, 1996; Faber et al., 2004; Li and Leckenby, 2004; Schlosser et al., 1999). For example, in relation to consumer behaviour research, understanding how users perceive the Web as a source of advertising is essential for advertisers to effectively communicate with current or potential consumers (Cho and Cheon, 2004; Ravi, 2005). In addition, there is an opportunity for researchers and practitioners to understand how advertising in this new medium can effectively serve the needs of consumers. Therefore, it is important to examine Web advertising from the consumer behaviour perspective as this medium has changed, and will 20

continue to change, the way advertisers communicate with consumers as well as the way consumers respond to advertising. Accordingly, the following literature review examines the disciplines of Web advertising and consumer behaviour. Firstly, the domain of Web advertising is discussed in the following section.

2.3

Web Advertising

On October 27, 1994, the Web found its commercial potential when Hotwired Website signed up for 14 banner advertisements to go online (Barnes, 2003; Kaye and Medoff, 2001; Robinson, Wysocka, and Hand, 2007). Since then, advertisers have considered the Web as one of the important marketing communication channels through which advertising or promotional messages are delivered to target markets (Bush et al., 1998; Chatterjee et al., 2003; Li and Leckenby, 2004; Yaveroglu and Donthu, 2008). More specifically, the Web has been viewed as a new means for reaching consumers (Ho, 1997). Due to the low cost and ease of developing a website, many commercial organisations have established their presence on the Web to provide services to customers, promote their products (Cheung, 1998), conduct transactions (Hsieh and Lin, 1998), streamline internal processes (Trappey and Trappey, 1998) and, thus, create competitive advantages in the marketplace (Bloch and Segev, 1996). To promote such sites, companies usually run advertisements in various popular portals and, consequently, Web pages have become dominated by advertisements similar to those viewed on television programs (Novak, Hoffman, and Duhachek, 2003).

In line with the growth of Web advertising expenditure (which is expected to reach US$81.1 billion worldwide by 2011) (E-Consultancy, 2007; Jaffray, 2007), there has also been a remarkable growth in the online consumer population with 1.6 billion online users in 2009 (Internet World Statistics, 2009). With this being the case, the increasing accessibility of the Internet appears to justify the investment in Web advertising. Therefore, given the significant marketing implications of Web advertising, it is important to review the various definitions of Web advertising that are presented in the relevant literature.

As previous research has used terms including Internet advertising, online advertising, and Web advertising to refer to advertising on the Web medium, the researcher conducted searches of databases using the keywords „Internet Advertising‟ , „Web Advertising‟, and „Online Advertising‟ in order to better define Web advertising. The search addressed articles dating from 1997 to 2008 in journals such as the following: Journal of Advertising Research, 21

Journal of Advertising, Journal of Interactive Advertising, Journal of International Marketing Review, Journal of Marketing, Journal of Internet Research, Journal of Interactive Marketing, and International Journal of Market Research. This review provided important insights into the development of scholarly research relating to Web advertising.

Web or online advertising is broadly defined, from a marketing perspective, as any paid form of presentation for ideas, goods, or services on the Web (Cho and Leckenby, 1999; Yoo and Stout, 2000). A more recent definition viewed Web advertising as an evolving and multifaceted form of advertising that is not limited by space or time, and that has the capability to communicate and interact with consumers in a different capacity, that is both involving and engaging (McMillan, 2004; McMillan, Hwang and Lee, 2003). For the purpose of this research, Web advertising is defined as Web content paid for by an identifiable advertiser, with persuasive intent to achieve marketing communication goals. More specifically, Web advertising includes the many forms of commercial content available on the Internet, from electronic advertisements that are similar to traditional advertisements (for example, billboards and banner advertisements), to formats that are distinct from traditional advertisements (for example, pop-up advertising) (Robinson et al., 2007; Schlosser et al., 1999; Yao and Lin, 2006). Accordingly, the characteristics of Web advertising are discussed in the following section.

2.3.1 Characteristics of Web Advertising The Web environment has unique characteristics that have created new shopping experiences for consumers, as well as methods for searching product information that is not possible in traditional media contexts (Richard and Chandra, 2005). It is important to reiterate here that while the characteristics of the Web mostly apply to the electronic medium itself, by association, advertisements that appear in the Web medium also possess similar characteristics.

The research suggests that the Web has four major characteristics: interactivity, tracking, accessibility and flexibility. Interactivity is the feature that most distinguishes the Web from offline traditional advertising media and relates to its ability to facilitate interaction within the communication process (Lai and Yang, 1998; Lee, Tansey, and Franwick, 1998; Yang, 2004). Additionally, interactivity refers to the interaction between a website and its user (Merrilees and Fry, 2003) or the two-way communication between site and user (Merrilees 22

and Miller, 2001); it is predicted to change not only the way of designing advertising but also the process in which it impacts consumers‟ behaviour (Chandon et al., 2003; Yang, 2004). For example, consumers can click on the advertisement for additional information and can then buy the product or service online in the same session. Simultaneously the advertisers may be provided with immediate responses and feedback from consumers (Wolak, 1999; Zeff and Aronson, 1999).

Tracking capabilities on the Web provide advertisers with the ability to identify online consumers activities, including online searching actions, the number of Web pages visited, and the length of the time spent at each website, as well as detailed information relating to the content viewed (Kanso and Nelson, 2004; Van Doren, Fechner, and Green-Adelsberger, 2000). In addition, Web advertisers can also track how consumers interact with their brands and accurately assess their responses to advertisements through using Web tracking software (Wolak, 1999; Zeff and Aronson, 1999). Moreover, the ability to track consumers‟ online activities has enabled advertisers to gather precise consumer information on the Web and identify the target market more accurately. As a result, advertisers are able to customize and personalize advertising content on an individual basis (McMillan, 2004; Palanisamy and Wong, 2003; Wolak, 1999; Zeff and Aaronson, 1999). For example, advertisers on the Web often deliver advertisements based on consumers‟ personal preferences and actual behaviour (Wolak, 1999; Zeff and Aronson, 1999).

Accessibility gives consumers control over when and where they choose to access Web information (Gallagher et al., 2001; Palanisamy and Wong, 2003; Yaveroglu and Donthu, 2008) and, therefore, has the advantage of making information immediately accessible at the (right) time when consumers need that product information (Lee et al., 1998). Further, consumers have the ability to close out advertisements they do not want to see or revisit advertisements that are of interest. This provides consumers with the opportunity to view advertisements more than once, based on their interests and desires (Vakratsas and Ambler 1999). Also, consumers can arrange advertisements to their liking and in accordance with their preferred scheduling. These observations are consistent with Sterne (1995) and Lee et al.‟s (1998) findings that the Web is a pull, and not a push, medium (pulling consumers to the messages instead of pushing messages to the consumers). Furthermore, the ease of access to the Web has created a business environment for small, medium and large sized companies

23

that have the ability to reach and communicate with current and potential consumers, with affordable costs and global reach (Quelch and Klen, 1996; Siegel, 2006).

Finally, the Web is flexible in that it can present information in numerous ways, including text, images, video and sound. This multimedia aspect of the Web has enhanced the consumer Web experience to be more fun and stimulating, often resulting in higher consumer attention (Ghose and Dou 1998; Yaveroglu and Donthu, 2008). Web advertising is also said to be more flexible in launching, updating, or cancelling information and delivering messages at any time (Wolak, 1999; Zeff and Aronson, 1999). However, it can come in many different forms, which may work individually or together to affect consumer behaviour. The dominant forms of Web advertising are discussed in the next section.

2.3.2 Web Advertising Forms Broadly, Web advertising includes corporate logos, banners, pop-up messages, and text-based hyperlinks to official websites (Briggs and Hollis, 1997; Korgaonkar and Wolin, 2002; Robinson et al., 2007). Specifically, four kinds of Web advertising have been identified in the Web advertising literature (Briggs, 1999; Chandon et al., 2003; Cho and Cheon, 2004; Choi and Rifon, 2002; Flores, 2000; Li and Leckenby, 2004; Robinson et al., 2007; Yao and Lin, 2006). They are:

1. Advertisements placed within website content. These include banners, rich media advertising, pop-up advertising, interstitial, buttons, Keyword Search Advertising (KSA), and animated cursors; 2. Sponsored elements within a website (e.g., paid product placements, or content designed around a sponsor); 3. E-mail advertising; and 4. Corporate or product specific company sites.

Recently, there has been a dramatic change in the relative importance of the various forms of Web advertising. Banner advertisements used to be one of the most popular Web advertising tools which are generally placed on high-traffic websites as display advertisements or hyperlinks that transfer users to the advertiser‟s website (Dou et al., 2001; English and Pearce, 1999; Hussain and Fenech, 2004; Hussain and Sweeney, 2005; Raman and Leckenby, 1998; Ryu et al., 2007). However, the effectiveness of banner advertisements has been 24

questioned, due to a noticeable decline in click through rates and reported negative perceptions of consumers towards banner advertisements (Dreze and Hussherr, 2003; Robinson et al., 2007; Ryu et al., 2007; Yaveroglu and Donthu, 2008; Yoon and Lee, 2007). While the revenues of banner advertisements have decreased from 56 percent in 1998 to 21 percent in 2003, conversely, rich media advertisement revenues have increased from two percent in 2002 to eight percent in 2003. In addition, Keyword Search Advertising (KSA) has grown from one percent in 2002 to 35 percent in 2003 (IAB, 2004; Li and Leckenby, 2004), and 41 percent in 2006 (Maddox, 2006). Such increases indicate that KSA represents the major and fastest growing form of the Web advertising formats. However, to date, within the literature, there has been limited research on KSA.

2.3.3 Keyword Search Advertising (KSA) This section reviews both the academic and practitioner literature regarding KSA. Firstly, various definitions of KSA are reviewed to identify a suitable definition for this research. Then, various types of KSA (with particular focus on SSA) are discussed, and the effectiveness of KSA reviewed with reference to the available Web advertising literature. However, there appears to be a paucity of prior research on the topic of KSA within the academic field and, therefore, many more studies in the trade literature were identified and reviewed.

KSA emerged from the need to create new, more effective and non-intrusive Web advertising formats (Emerson, 2005), as evidence suggests that consumers are becoming more oriented towards advertising and products that are of interest to them (Weidlich, 2002), rather than accepting traditional mass marketing approaches. Overall, KSA is said to provide an effective means of presenting commercial information relevant to user needs at the right time (Li and Leckenby, 2004). Different terms have been used to define KSA, such as „Search Engine Advertising‟ (Scott, 2003), „Search Advertising‟ (Chuang and Chong, 2004; Schmidt and Patel, 2004), or „Keyword Advertising‟ (Arnold, 2003; Cheek et al., 2001). However, despite its various labels, KSA represents “… an advertising technique on the Web which uses search systems to target advertisements to the appropriate audience” (Schmidt and Patel, 2004, p.1). Specifically, KSA displays a firm‟s text advertising on the Web by matching a site‟s content with the advertiser‟s keyword[s] (Burns, 2005; Cheek et al., 2001; Vaughan-Nicholas, 2003). 25

Importantly, KSA comprises two main types, namely, natural and paid search results. The former are free or natural search results, which means that the site‟s ranking is based only on content and keyword relevancy (Precision Marketing, 2005). These results appear in the main body of a search engine‟s listings. However, as advertisers cannot guarantee ranking positions or a listing on the first page of the search results, „natural‟ KSA is not the focus of this research. Rather, the focus here is on paid KSA that involves the payment of money to improve the ranking of a particular website on a search results page. Specifically, three different types of paid search results have been identified within the Web advertising literature: Paid inclusion advertising, Pay-Per-Call advertising and, of particular interest for this study, Sponsored Search Advertising (SSA).

Paid Inclusion Advertising Paid inclusion advertising requires firms to pay a fixed rate to search engines, which allows their advertising to become part of the natural search results. However, as there is no guarantee regarding the ranking of links by keywords (for example, there appearance on page one or page fifty) (Nicholson et al., 2005; Van Couvering, 2004), this form may not be considered viable for many advertisers. In particular, this form of advertising: (1) does not guarantee high placement on a search engine‟s natural results (Mummert, 2005); and (2) is not utilised by all search engines. This type of advertising is predominantly used by MSN and Lycos. Although one of the benefits of paid inclusion is that it is the least expensive of the KSA forms, it is also considered less effective (Mummert, 2005).

Pay-Per-Call or Click-to-Call-Advertising With Pay-Per-Call search advertising (Blyth, 2006; Krol, 2005) or Click-to-Call-advertising (Marketing Week, 2006), businesses buy advertisements that appear in search results alongside SSAs. The goal of Pay-Per-Call advertising is to drive online users to a phone number rather than a website (Laratro, 2006). This method is less expensive in comparison to traditional yellow pages advertising and it has the added advantage of giving a basic Web presence to companies (Marketing Week, 2006; Laratro, 2006). In addition, one distinction between Pay-Per-Call and other KSA types is the use of categories rather than keywords. Advertisers choose a business category in which they want their advertisements to appear and bid a certain price per category (Krol, 2005). While this type of KSA is prominently used by particular search engines, such as „search.com‟, it is not commonly used within popular search engines and, therefore, will not be investigated in this research. 26

Sponsored Search Advertising (SSA) Sponsored Search Advertising (SSA) is a KSA form in which a fee is paid for specific keywords to guarantee priority placement on search engine result pages (Feng et al., 2003; Lu and Chau, 2006; Overture, 2005). As the fee is paid when users click on links, it is also referred to as Pay-Per-Click advertising (Feng et al., 2003). Also, SSA listings are usually identified by search engines as sponsored links, separated from the natural free results. For example, Google‟s revenue is mostly dependent on Sponsored Search Advertisements that are located along the top, or at the right side of the search results page (Delaney, 2004). Advertisers concentrate on bidding for particular keywords to obtain a high placement on the sponsored links section as there is evidence that users are more attracted to top ranked results (Harrison, 2005). For this reason, SSA is the focus of the current research, as it has been identified as the most effective form of KSA (Delaney, 2004; Harrison, 2005). Therefore, the following section examines SSA in more detail.

2.3.4 Sponsored Search Advertising (SSA) SSA was developed as a feasible alternative to natural free search results, as well as offline traditional advertising and has become a topic of interest for many researchers and practitioners (for example, Animesh, Ramachandran, and Viswanathan, 2007; Ghose and Yang, 2007; Jansen and Resnick, 2006; Marble, 2003). Indeed SSA is beneficial for all involved parties: the advertisers, search engines, and online users (Jansen and Resnick, 2006). For example, Ghose and Yang (2007) note that advertisers see the advantage of being included in the search engine results using the SSA system as it shifts the advertising experience from mass marketing to a more targeted advertising approach. Another benefit offered by SSA is that it provides an opportunity to reach Web users and, therefore, increase traffic flow to websites (Li and Leckenby, 2004).

Further, SSA has proved to be the major source of revenue for Web search engines (Ghose and Yang, 2007); 94 percent of online users use search engines to locate information on the Web (Nielsen-Net Ratings, 2006). It is no wonder that search engines are considered valuable sources of online information because they help Web advertisers to direct consumers to their websites through selling SSA (Ghose and Yang, 2007). Accordingly, SSA has become a critical component of companies‟ marketing campaigns, with a global annual growth rate predicted to be 37 percent and a revenue of more than US$33 billion in 2010 (Ghose and Yang, 2007). 27

Additionally, SSA has proved to be beneficial for online users (consumers) because they are presented with highly relevant search results (Kiritchenko and Jiline, 2008). Accordingly, SSA is an important element of the Web browsing and information searching experience (Richardson et al., 2007). Importantly, as SSA is based on the consumer‟s own queries and, it is thus considered less intrusive and more effective than banner advertisements or pop-up advertising (Ghose and Yang, 2007).

As SSA can increase the effectiveness of Web advertising, by allowing advertisers to more precisely target Web users, advertisements have become more relevant and meaningful to them, and less intrusive to the customers (Dou et al, 2001). A number of Web advertising effectiveness measures have been recommended, such as brand awareness, advertising recall (Dreze and Hussher, 2003), exposure, attention, click-through, interactivity, sales outcomes (Novak and Hoffman, 1997; Bhat, Bevans, and Sengupta, 2002), conversion rate, number of transactions (Dreze and Zufryden, 1998) and behavioural intentions (Shamdasani et al., 2001). However, the main measure for SSA effectiveness is the click-through rate (CTR), which is a behavioural indicator of a consumer‟s intention to receive more information on a specific topic or to purchase a product or service (Cyberatlas, 2001; Jin and Jun, 2007). This effectiveness is also measured by the number of matches between particular keywords and the search results content. Thus, if a search engine produces search results that are relevant to the search keywords, then it is likely to be effective in getting the consumers‟ attention, enhancing positive attitudes towards advertising and, consequently, increasing the possibility of clicking on the advertisements (Dou et al., 2001; Yus, 2005).

2.3.5 Section summary The preceding section has provided an overview to the Internet and the Web, as well as a presentation of a discussion on the development, characteristics and various forms of Web advertising. A review of the relevant literature revealed that dramatic structural changes are taking place within the rapidly growing Web advertising sector; the most notable being the rise in prominence of KSA in general, and SSA in particular. The online advertising literature suggests that SSA may increase the effectiveness of Web advertising through enabling online advertisers to more precisely target Web users. However, it seems that many studies have focused on investigating SSA from the advertiser‟s perspective rather than a consumer‟s

28

perspective. Therefore, the following section discusses Web advertising, and more specifically SSA, from a consumer behaviour perspective.

2.4

Web Advertising from a Consumer Behaviour Perspective

This research investigates factors influencing consumers‟ responses to SSA, with particular focus being given to the Web advertising literature associated with consumer behaviour. Thus, the section reviews the consumer behaviour literature, which is separated into three parts. The first part relates to research on consumer behaviour associated with Web advertising and SSA. The second part provides an overview of the theories associated with consumer behaviour towards offline and Web advertising. The third part reviews and presents a discussion on the literature regarding a range of consumer behaviour variables in relation to SSA. While a broad range of consumer behaviour areas are covered, the focus remains on identifying and exploring the factors that influence consumers to attend to, develop attitudes towards, and respond to, SSA.

2.4.1 An Overview of Consumer Behaviour in Terms of Web Advertising The dramatic increase in the number of Web users, and the increase in broadband technology use, has created a consumer population that is diverse in relation to its experience, knowledge, attitudes and behaviours (Pew Internet and American Life Project, 2005). Although it appears self-evident that understanding consumer behaviour is the key to successful marketing communications, companies and advertisers are not fully exploring consumer online behaviour (Dittmar, Long and Meek, 2004; Goldsmith, 2002; Palanisamy, 2005). Further, most previous advertising research has examined advertising effectiveness from the advertisers‟ perspectives (Faber et al., 2004). For this reason, Faber et al. (2004) suggest the need to understand Web advertising effects from a consumer behaviour perspective as a way of expanding and providing important contributions to the Web advertising literature. The current research, therefore, will fill that gap by exploring consumer behaviour on the Web in terms of understanding the factors surrounding consumer response towards Web advertisements.

Past consumer studies on consumer response to Web advertising have indicated that consumers differ on a number of dimensions, such as attention, attitudes, perceptions, and behavioural responses (Briggs and Hollis, 1997; Cho, 1999, 2003; Eighmey, 1997; Yoon and Lee, 2007). Early research has primarily focused on identifying the idiosyncratic properties of 29

Web advertising (Hoffman and Novak, 1996) and profiling online users and their behavioural responses toward Web advertising (Eighmey, 1997). In contrast, the more recent research has examined the impact of Web advertising characteristics (e.g., colour, complexity, and type) on consumer online behaviour and responses toward Web advertising (Briggs and Hollis, 1997; Cho, 2003; Ryu et al., 2007; Yoo and Kim, 2005)

Other studies have focused on investigating consumer response to different types of Web advertising, for example, banner, sponsorship and pop-up advertising (Coyle and Thorson, 2001; Ducoffe, 1996; Li, Edwards, and Lee, 2002; Ryu et al., 2007). However, there does not seem to be a consensus in the literature regarding the impact of Web advertising upon consumer behaviour. Nevertheless, the research does indicate that Web advertisements are effective in generating positive consumer behavioural responses (Calisir, 2003; Coyle and Thorson, 2001; Ducoffe, 1996; Schlosser et al., 1999), increasing brand awareness (Briggs and Hollis, 1997; Dahlen, 2001), and precipitating purchase decisions (Calisir, 2003; Schlosser et al., 1999). In addition, Web advertising is perceived by some consumers as valuable, informative and non- irritating (Ducoffe, 1996).

These sentiments are not, however, universal. Other studies suggest that Web advertisements are irritating and insulting to people‟s intelligence (Brackett and Carr, 2001; Omar, 2000), and are less effective in encouraging favourable behavioural responses (Gordon and De Lima, 1997). Indeed, some consumers are more likely to respond negatively toward Web advertising because they still consider the Web as a task-performing medium rather than an entertainment medium (for example, Cho and Cheon, 2004). More importantly, the Web medium has become so cluttered with advertisements that it is perceived as intrusive and annoying (Yaveroglu and Donthu, 2008) and, accordingly, Web users tend to avoid advertisements as they interfere with and interrupt, their tasks and interests (Cho and Cheon, 2004; Li et al., 2002; McCoy et al., 2008). Similarly, Edwards, Li, and Lee (2002) contend that the key reasons for ignoring Web advertisements are their intrusiveness and irritation factor, which in turn may generate negative effects on the perceived value of Web advertising (Omar, 2000). However, whilst Web users may respond negatively to Web advertising, it would seem that they are still more likely to attend and respond to targeted advertisements (Chandon et al., 2003; Chatterjee et al., 2003; Mehta and Sivadas, 1995). That is, consumers are less irritated by, and more attentive to, non-intrusive targeted Web advertisements, such

30

as SSA (Bloem, 2003). Thus, this area warrants further examination through investigating consumer behaviour associated with SSA.

2.4.2 Consumer Behaviour in the Context of Sponsored Search Advertising As an aim of the current research is to understand the factors that influence consumers‟ responses towards SSA, the literature related to SSA was reviewed. However, much of this research is largely based on practitioners‟ views and anecdotal studies with only a small number of scholarly studies being available. According to Vine (2004), consumers depend on Web search engines to find information on the Web prior to making online or offline purchase-decisions. However, despite the phenomenal growth of the SSA market, limited research has been undertaken to investigate the implication of such advertising upon consumer behaviour. As mentioned previously, SSA has, in the main, been investigated from advertisers‟ perspectives, which appears not to provide an encompassing understanding of their effectiveness (Feng et al., 2003; Weber and Zheng, 2003). Further, the consumer behaviour perspective has been largely overlooked (Yao and Male, 2008). In support of this view, Ghose and Yang (2007) indicate that there is a little understanding of how consumers respond to SSA.

Some research has identified: that online users respond more favourably to natural free results displayed on the search results page in comparison to SSA (Greenspan, 2004; Marble, 2003); that online users tend to be distrustful in relation to the viability of Sponsored Search Advertisements, which are perceived as less relevant than the natural free results (Greenspan, 2004; Fallows, 2005; Jansen and Resnick, 2007; Jansen and Resnick, 2006); that online users have a limited understanding of the nature of SSA, with few were able to recognise SSA (Marble, 2003). More specifically, Animesh (2005) suggests that the consumer knowledge of SSA mechanism may foster the click through rates for highly positioned search results. Support for this view is provided by anecdotal evidence suggesting that companies need to be positioned on the first page and within the top search results to match online users‟ behaviours (Miller, 2006; Newsome, 2006; Schwarz, 2005). Interestingly, Yao and Male (2008) have found that consumers who engage in more search and clicking behaviour on the Web tend to be more responsive to SSA than others.

In brief, despite the emerging stream of literature on SSA, consumer behaviours associated with this type of Web advertising are still not fully understood. Thus more research is 31

required to fill this gap. Most research on SSA has generally explored the topic from a practitioner‟s perspective or anecdotal studies. Further, there has been scant research within the consumer behaviour domain in relation to how consumers respond to SSA. As a consequence, examining consumer response towards advertising, based on offline traditional advertising literature, may assist in achieving a more cohesive understanding of how consumers process and respond to SSA. Thus, an examination of consumer response to advertising will be most helpful.

2.4.2 Consumer Response to Advertising A review of the offline traditional advertising literature has revealed that a range of variables have been examined in relation to consumer response towards advertising, for example, their attitude towards advertising and how it influences the way consumers respond to offline traditional advertising (Ajzen, 2005; MacKenzie, Lutz, and Blech, 1986; Mittal, 1994; Pollay and Mittal, 1993). Indeed, consumer attitudes have been conceptualised as a Tri-component of attitude evaluation consisting of cognitive, affective and behavioural components (Ajzen, 2005; Eagley and Chaiken, 1998; Thurstone, 1928; Zanna and Rempel, 1988). The Tricomponent model was considered as the cornerstone for most of the early attempts to develop systematic models of consumer behaviour (Holbrook and Batra, 1987). However, the essential limitation of this model lies in assuming that consumer response are mainly influenced by the cognition component (Hall, 2002).

Other consumer behaviour researchers have found that attention is an important factor that should not be overlooked. For this reason, attempts have been made to develop more comprehensive models by incorporating attention towards advertising; it was expected that such an approach would provide a greater understanding of how consumers respond to advertising. As a result, Lavidge and Steiner (1961) developed the hierarchy of effects model, which become the dominant and most applied conceptual model on advertising effectiveness (Hall, 2002; Vakratsas and Ambler, 1999), playing a major role in the development of advertising research. The model extended one of the earliest models offline traditional advertising studies (Attention-Interest-Desire-Action (AIDA) model) (Hall, 2002; Strong, 1925). These two models are based on a process which starts with cognition response, that translates into affective and, consequently, leads into conation responses (Hall, 2002). For example, the hierarchy of effects model postulated that attention (as a first necessary step of consumer response to advertising) leads to interest which in turn leads into conviction to 32

desire and consequently leads into action behaviour (Lavidge and Steiner, 1961). To provide further support to this view, advertising researchers and practitioners have highlighted the role of attention towards advertising in examining consumer response to advertising from a consumer behaviour perspective (Hall, 2002; Young, 1981). It appears that consumers need to be exposed to, and pay attention to, advertisements before any further responses occur (Wedel and Pieters, 2003).

Moving into the Web advertising context, research has identified that the Web offers a wider and more immediate scope for communication between consumers and advertisers than the offline traditional media (Pavlou and Stewart, 2000). However, given the similarities between the two media, it would seem that the criteria for examining offline traditional advertising effectiveness could be applied to the Web advertising context (Stewart and Pavlou, 2002). More specifically, Li and Leckenby (2004) propose that, although Web advertising effectiveness may be measured using these three dimensions (cognitive, affective and behavioural), consumer selectivity of exposure to advertising on the Web should not be overlooked. Of particular importance is the notion that the challenges of generating effective Web advertising may be greater than for offline traditional media (Briggs and Hollis, 1997; Pavlou and Stewart, 2000; Rodgers and Thorson, 2000). Online users can decide whether to be exposed to advertising messages or not (Abd Aziz, Yasin, and Kadir, 2008), as well as limit or extend the time of exposure in comparison to offline traditional advertising (Pavlou and Stewart, 2000). While Dreze and Hussherr (1999) support this view that consumer behavioural responses to online advertising are similar to their responses to offline traditional advertising, they also warn that Web advertising is more prone to be ignored.

For these reasons, other researchers recommend that new measures should be used to examine consumer behavioural responses to Web advertising (Chen and Wells, 1999; Pavlou and Stewart, 2000; Yaveroglu and Donthu, 2008). For example, they postulate that measuring the effectiveness of Web advertising, based on offline traditional advertising processing models may result in an underestimating of the value of Web advertising effectiveness. This is an important aspect as the most important advantage of Web advertising, in comparison to offline traditional advertising, is its ability to induce direct consumer behavioural responses. Earlier Web advertising effectiveness research has measured direct responses, such as the consumer tendency to click on advertisements (Hoffman and Novak, 1996; Shmadasani et al., 2001; Li and Bukovac, 1999; Yoon and Lee, 2007). Recently, however, actual click-through 33

rates have witnessed a huge decline to below one percent (Cho, 2003; Ryu et al., 2007; Yaveroglu and Donthu, 2008; Yoon and Lee, 2007). This dramatic decline has raised criticisms in relation to focusing only on these direct response measures, which may undervalue the effectiveness of Web advertising (Briggs and Hollis, 1997; Dreze and Hussherr, 2003; Ryu et al., 2007). Viable alternatives to complement the measurement of consumer responses towards Web advertising could include consumers psychological responses, such as attention, attitude and behavioural responses (McCoy et al., 2007; Moore, Stammerjohan, and Coulter, 2005; Rodgers and Thorson, 2000; Ryu et al., 2007; Yaveroglu and Donthu, 2008).

For Rodger and Thorson (2000), Web advertising effectiveness can be examined through the distinction of two major aspects: process and outcomes. For example, process measures include consumer selectivity in paying attention to, and processing of, the advertisement, while outcomes are mainly concerned with the consequent responses of noticing and attending to the online advertising messages, measured by attitudes and intentions (Li and Leckenby, 2004). Moreover, both aspects of process and outcomes appear to be related, to some extent to, the recipient (the consumer) of the advertising message (Li and Leckenby, 2004). From this perspective, and extending on Rodger and Thorson‟s (2000) postulations, the process and outcomes aspects can be acknowledged as relying on consumer related factors and, therefore, provide the opportunity to achieve a greater understanding of how consumers respond to Web advertising (for example, SSA).

In summary, the above discussion has reviewed a number of studies related to consumer response to offline traditional advertising, as well as Web advertising. Based on Rodgers and Thorson‟s (2000) findings that Web advertising effectiveness relies on how consumers respond to such advertising, and that it may be examined using determinants, process and outcomes, it is apparent that process focuses on the consumer‟s tendency to pay attention to advertising and, moreover, this process could be determined by a range of variables related to the consumers. Further, the outcomes appear to reflect consumers‟ attitudes and behaviour upon attending to advertisements. As noted previously, consumers‟ tendency to pay attention to Web advertisements has witnessed a decline since it was first introduced due to such factors as their intrusiveness and irritation level (Li et al., 2002; Yaveroglu and Donthu, 2008). However, given that SSA has proved to be the most effective form of Web advertising (Mack, 2004; Mara, 2004), it would seem there is an opportunity to examine consumer 34

attention to SSA and the related determinants of such attention. Li and Leckenby (2004) asserts this research need as there has been few studies into understanding how consumers attend to, and respond to such advertising. Therefore, in the following section, a number of focal constructs relevant to this research are discussed, in terms of consumer attention towards SSA and the determinants of such attention (prior experience with SSA, subjective knowledge of SSA, and familiarity of brands (or websites) included in the SSA) and the outcomes associated with attention towards SSA (attitudes towards SSA and behavioural intentions).

2.5

Consumer Attention towards SSA

Although Web advertising has grown rapidly, industry practitioners are facing increasing challenges in attracting consumer attention to Web advertisements (Low, 2000; Ryu et al., 2007; Yaveroglu and Donthu, 2008). Compounding these challenges, the typical consumer is exposed to more than 3000 advertisements a day (Solomon, 2004). Such clutter in both offline and online environment results has resulted in two reactions from consumers: attending to selected stimuli or ignoring it (Grunert, 1996; Yaveroglu and Donthu, 2008). Attracting consumer attention is the first important objective of advertising; according to Davenport and Beck (2001, p.13), “… understanding and managing attention is now the single most important determinant of business success”. To help to gain an understanding of online users‟ attention towards SSA, a review of the literature on attention towards advertising can be seen as a worthwhile endeavour.

Attention has been defined as the amount of mental effort or cognitive capacity allocated to a task (Kahneman, 1973). In general, attention refers to the selectivity of processing (Eysenck and Keane, 1995), and the concentration effort on a stimulus available to the cognitive system (Ashcraft, 1998). These definitions suggest that when consumers give attention to one stimulus, they tend to ignore other surrounding stimuli or distractions. For the purpose of this research, attention is viewed as the extent of cognitive resources a person devotes to an advertisement (SSA) (Laczniak and Muehling, 1993; Muehling et al., 1990).

Much of the previous research (anecdotal-based-studies) related to SSA indicate that online users tend to pay a different amount of attention to SSA displayed at various positions within a search results page (for example, Greenspan, 2004; Marble, 2003). In addition, most SSA studies have investigated consumer avoidance behaviour rather than focusing on consumer 35

attention (Hotchkiss, Garrison, and Jansen, 2004; Lu and Chau, 2006). For example, Hotchkiss et al. (2004) found that some online users tend to visually ignore or avoid engaging with SSA. Support for this view is provided by Lu and Chau (2006), who suggest that Sponsored Search Advertisements should be noticed and attended to by consumers in order that they acquire a clicking intention. In addition, some online users have biases against SSA; these biases influence the viability of attracting potential consumers to attend to the advertising (Langford, 2000). Therefore, given that SSA has become one of the most popular formats for Web advertisements by leading e-commerce companies (Vragov, 2009), it is important to examine consumer attention towards SSA, as well as the factors that influence consumers to pay attention to such advertising.

Other researchers have also emphasised the importance of attention to advertising (Cialdini, Petty, and Cacioppo, 1981; LaBerge, 1995; Wedel and Pieters, 2000). Indeed, Wedel and Pieters (2000) suggest that advertisements, which fail to gain consumers attention, may not be effective according to a pure cognitive orientation. Taking this point further, the capturing of consumer attention to advertisements facilitates higher cognitive functions that they work on more salient inputs (LaBerge, 1995). As discussed previously, the importance of attention in advertising is highlighted, by its inclusion in a number of advertising theories. For instance, one of the first advertising theories, AIDA, viewed attention as the first stage that individuals go through when exposed to advertising (Vakratsas and Ambler, 1999), while, the hierarchy of effects model (Lavidge and Steiner, 1961) suggests that attention is an essential step that should take place prior to the higher levels of consumer processing such as attitudes. Support for this contention is provided by Sacharin (2000, p.2), who suggests that “... marketers cannot persuade people unless they have their attention first... [therefore], attention is a prerequisite for all marketing efforts”.

Within the marketing literature, several factors have been identified as influencing the amount of attention given towards an advertisement. In particular, the attention allocated to advertisements has been assumed to be a function of the consumers‟ motivation, opportunity, and ability to attend advertisements (Celsi and Olson, 1988; Kahneman, 1973; MacInnis and Jaworski, 1989). In contrast, other studies have found that attention to advertising is influenced by the characteristics of the advertisement (Berdie, 1992; Janiszewski, 1998; Kelly and Hoel, 1991; Lohse, 1997). Thus, existing research related to factors influencing

36

consumers‟ attention towards advertisements can be classified into two streams: the impact of advertising message characteristics, and the impact of consumer related factors.

The first stream of research, the impact of advertising message characteristics on the amount of attention paid to advertising (Berdie, 1992; Janiszewski, 1998; Kelly and Hoel, 1991; Lohse, 1997; Pieters and Wedel, 2004; Singh et al., 2000), has generally been interested in discovering advertising attributes that significantly influence or grab consumer attention towards advertisements, such as headline, copy and size (Pieters and Wedel, 2004; Rossiter, 1982; Soley and Reid, 1983; Wolfe, 1998). However, as the research revolves around an understanding of consumer response to advertising, only consumer related factors are of interest rather than advertiser related factors (for example, advertising message characteristics). For this reason, the first stream of research regarding attention is not discussed in further details.

The second stream of research on attention to advertising deals with the impact of consumer related factors (Batra and Ray 1986; Celsi and Olson, 1988; Chun and Wolfe, 2001, Rayner et al., 2001; Rosbergen, Pieters and Wedel, 1997; Yantis, 2000). It appears that consumers' attention is strongly influenced by their motivations, abilities, and opportunities to process advertisements. For example, the ability to pay attention to advertising may be conceptualised by aspects related to experience with, and knowledge about, the related environment (Batra and Ray, 1986; Celsi and Olson, 1988; Hallahan, 2000; MacInnis, Moorman, and Jaworski, 1991). If this is so, then, aspects of prior experience with advertising (negative or positive) are likely to influence an individual‟s ability to attend to, and process, advertisements (Batra and Ray, 1986; Celsi and Olson, 1988; MacInnis et al., 1991). Therefore, prior experience with SSA was deemed pertinent to the current research as a determinant of consumer attention towards SSA. This position is consistent with the view that consumer attention is mainly dependent on prior experience and that prior experience is an important determinant of consumer behaviour (Fazio and Zanna, 1981; Jarvelainen, 2007).

A number of researchers suggest that the ability to attend to advertising messages may result from the amount and type of knowledge an individual possesses regarding the advertising, for example, subjective knowledge (Batra and Ray, 1986; Celsi and Olson, 1988; Hallahan, 2000; MacInnis et al., 1991). Therefore, subjective knowledge is another important determinant factor that influences consumer attention and processing behaviour towards SSA 37

(Bettman and Park, 1980; Claxton, Fry, and Portis, 1974; Brucks, 1986; Pingol and Miyazaki, 2005; Raju, Lionel, and Mangold, 1995). However, within the online environment, it appears that online users have limited knowledge of the approaches employed by search engines to rank, retrieve and prioritise Sponsored Search Advertisements (Greenspan, 2004; Hotchkiss, 2004; Marble, 2003). This being the case, it may be that consumer subjective knowledge of SSA could influence attention towards this type of Web advertising. Finally, the consumer‟s motivation to process the advertising message is largely determined by consumers findings cues in the advertisement that are recognisable (Batra and Ray, 1986; Celsi and Olson, 1988; Hallahan, 2000; MacInnis et al., 1991), for example, familiar brands. As such, familiarity of brands (or websites) included in the SSA was selected as an aspect of the current research, due to its importance in influencing advertising effectiveness (Campbell and Keller, 2003) and consumer attention toward advertising (Chun and Wolfe, 2001; Keller, 1991; MacKenzie and Spreng, 1992; Rayner et al., 2001; Rosbergen et al., 1997; Yantis, 2000).

As discussed above, consumer related factors, such as prior experience, subjective knowledge and familiarity of brands (or websites) deemed to be important determinants of consumers‟ attention towards advertisements and may operate in a similar manner in the online environment. As a consequence, these factors were deemed relevant to this study as the determinants of consumer attention towards SSA. These determinant factors are discussed in greater detail in the following sections.

2.5.1 Prior Experience with Sponsored Search Advertising (SSA) Based on the consumer behaviour and broader marketing literature, this section presents a discussion of the relationship between prior experiences with SSA and consumer attention towards such advertising. Prior experience, in the consumer behaviour literature, is viewed as information learned from consumers‟ actual usage behaviour with a particular stimulus (Dodd et al., 2005; Fazio and Zanna, 1981); and has a strong and direct influence on consumers behaviour (Fazio and Zanna, 1981) as consumers depend on and value more, information acquired from previous experiences (Cho and Cheon, 2004). More specifically, prior experience with advertising represents information learned, and conclusions drawn from prior experience with the advertising, thus functioning as both an important feedback factor as well as determining how likely it is that the consumer will engage with the advertising in 38

the future (Cho and Cheon, 2004). While consumers depend on advertising messages to obtain information about products, to assist them in making purchase decisions, prior experience is also considered as an important determinant within the decision-making process (Chang, 2004). These findings concur with previous studies suggesting that prior experience with a particular object lead to more specific consumer perceptions, while their judgment are based on more concrete criteria (Fazio and Zanna, 1981; Smith and Swinyard, 1983).

Other studies have examined the influence of prior experience on consumer attentive processing of different media content (Jarvelainen and Puhakainen, 2004; King and Xia, 1997; Rayburn, 1996; Rice, 1993). For example, Rayburn (1996) suggests that consumer expectations about a particular medium‟s performance are influenced by prior positive or negative experiences with the medium and its content. Similarly, King and Xia (1997) assert that an individual‟s perceptions of media appropriateness are associated with higher positive experiences with the media. These findings are particularly evident in the online context. Klein (1998) and Jarvelainen and Puhakainen (2004) contend that consumers, who had positive prior experiences with online purchases, were more likely to perceive the Web as the most appropriate source for their final purchasing decision (Jarvelainen and Puhakainen, 2004). Such findings are also consistent with research examining computer technology usage, whereby prior experience was found to influence user perceptions of the usefulness and ease of use of the Internet (Jiang et al., 2000; Koufaris, 2003).

There is also a general agreement in the marketing literature that prior experience influences consumer response towards advertising (Bruner and Kumar, 2000; Fazio and Zanna, 1981; Eagly and Chaiken, 1993; Taylor and Todd, 1996; Venkatesh and Davis, 1996). Specifically, prior experience has a major role in shaping consumer behaviour (Ehrenberg, 1974; Jones, 2002; Reed and Ewing, 2004), as well as in influencing consumer attention to the advertisement (Cho and Cheon, 2004; Eysenck and Keane, 1995; Jin, Morris, and Cho, 2005; Lodish et al., 1995; MacKenzie, 1986). For example, Cho and Cheon (2004) examined the impact of prior negative experience on consumer tendency to avoid or pay attention to, Web advertisements; such experiences were found to result in the consumer avoiding online advertising. The findings support the revised view of offline traditional advertising avoidance studies (Speck and Elliott, 1997; Elliott and Speck, 1998). Although Cho and Cheon‟s (2004) study provides a significant contribution to the understanding of online users‟

39

attention/avoidance behaviour, the research was limited to only one type of Web advertising, namely, banner advertising.

With regard to the SSA context, little research has been undertaken into exploring the impact of prior experience on consumer attention towards SSA. Moreover, the existing research associated with this topic has focused on the impact of user experience on information searching using Web search engines. The Pew Internet and American Life Project (2005) investigated how users interact with web search engines and concluded that the majority of experienced online users view search engines as fair and unbiased sources of information. In addition, the majority of online users indicated that they had positive and successful experiences with their previous online information searching; consequently, they felt more confident about their online searching skills and more satisfied with the search engines.

Given the above discussion, it would seem plausible that online consumers may be more likely to attend to and pay more attention to advertisements on the Web, if the first experience is perceived to be positive (Jarvelainen, 2007). Conversely, in the case of a negative first experience, the consumers may be hesitant to attend to the advertisements (e.g., SSA) to which they are exposed within the Web, or they may avoid them completely. Thus, it seems that the manner in which online users process SSA may be reliant on previous experiences with this type of Web advertising.

As discussed in section 2.4, subjective knowledge is viewed as another consumer related factor that influences consumer attention toward online advertisements. For this reason, the following section presents a discussion on this topic in more detail.

2.5.2 Subjective Knowledge of Sponsored Search Advertising (SSA) Generally, knowledge is defined as “… the information stored within memory” (Engel et al., 1990, p. 281), as well as the degree of consumers experience and familiarity with products prior to their conducting any external search (Brucks, 1985; Sujan, 1985). Consumer knowledge has received extensive examination in the consumer behaviour domain (for example, Brucks, 1986; Raju et al., 1995; Rao and Monroe, 1988; Sujan, 1985), due to its influence upon consumer behaviour and consumer searching behaviour (Brucks, 1986; Lehto, Kim, and Morrison, 2006; Pillai and Hofacker, 2007; Pingol and Miyazaki, 2005; Raju et al., 1995), along with their purchase-decisions (Brucks, 1985). For example, knowledgeable 40

consumers have a higher ability to process information when making decisions (MacInnis et al., 1991). In addition, knowledge is related positively to the efficiency of the information search (Putrevu and Ratchford, 1997). In particular, a knowledgeable consumer is viewed as having the ability to acquire the needed information with minimal effort (Putrevu, Tan, and Lord, 2004).

Consumer knowledge is distinguished in the related consumer behaviour literature into two distinct types: subjective knowledge and objective knowledge (for example, Alba and Hutchinson, 2000; Bearden, Hardesty and Rose, 2001). The general view is that subjective knowledge represents an individual‟s perceptions of the information stored in their memory (Brucks, 1985; Flynn and Goldsmith, 1999; Park, Motherbaugh, and Feick 1994) about an object; objective knowledge refers to the precise information stored in the individual‟s memory (Bettman and Park, 1980; Schmidt and Spreng, 1996).

Both subjective knowledge and objective knowledge have been described as related but distinct constructs with unique measures (Brucks, 1985) and different antecedents (Flynn and Goldsmith; 1999; Park et al., 1994; Raju et al., 1995; Schmidt and Spreng, 1996). Moreover, research suggests that those individuals who depend on subjective knowledge might go through the entire decision making process in a different manner than those who are dependent on objective knowledge (Raju et al., 1995). However, subjective knowledge was found to have a stronger impact upon consumers‟ behaviour in comparison to objective knowledge (Mitchell, 1982; Raju et al., 1995). Support this view, Fiske et al. (1994) found that subjective knowledge plays a stronger motivational role in terms of consumer behaviour than objective knowledge, thereby providing more impetus for including subjective knowledge in the current research.

Many researchers have focused on the relationship between subjective knowledge and consumer behaviour (Bang et al., 2000; Flynn and Goldsmith, 2001; Ward and Lee, 2000), identifying that subjective knowledge as an important determinant of consumer behaviour (Bang et al., 2000; Flynn and Goldsmith, 2001). For example, Flynn and Goldsmith (2001) concluded that the consumer‟s subjective knowledge of a particular object had an influence on their behavioural response toward that object. In addition, consumers with high subjective knowledge are more likely to respond positively towards the object (Alba et al., 1997; Bang et al., 2000; Ward and Lee, 2000). Conversely, Alba and Hutchinson (1987) argue that less 41

knowledgeable consumers are more likely to be inefficient in using their search time, which then leads to frustration and potentially negative behavioural responses towards that experience.

Subjective knowledge has also been found to influence consumer responses toward advertisements (Alba and Hutchinson, 1987; Berger and Smith, 1998; Flynn and Goldsmith, 1999; Olson, 1980; Rao and Monroe, 1988; Sohn and Leckenby, 2004; Sujan, 1985). For example, Cramphorn (2004) found that subjective knowledge shapes an individual‟s response to, and perceptions about, advertisements. Additionally, various levels of subjective knowledge may either constrain or enhance consumers‟ cognitive engagement with the advertisements (Petty, Cacioppo, and Schumann, 1983). Similarly, Meyers-Levy (1991) identified that the process of associating new information with subjective knowledge plays an integral part in the consumer‟s response toward advertisements. Limited information is also available on how online users‟ knowledge, regarding SSA influences their behavioural responses (for example, Consumer WebWatch, 2002; Marble, 2003). In those anecdotal studies that do exist, online users were found to have little understanding of the approaches employed by search engines to rank, retrieve and prioritise sponsored and natural free links within the search results page (Marble, 2003). Further, a survey conducted by Consumer WebWatch (2002) reported that 60 percent of online users were not aware of the SSA practises. A more recent anecdotal study, conducted by The Pew Internet and American Life Project (2005), reported that almost two thirds of the online users were not aware of the sponsored results [SSA] presented by the Web search engines.

In terms of knowledge of SSAs, Hansen (2002) found that the labels, used by the search engine to identify SSA (for example, recommended sites, search partners, featured listings, or start here), were perceived to be ambiguous and inadequate at informing users about the nature of SSA. This suggests that online users may be both insufficiently educated and/or suspicious about the viability of SSA, which in turn, could result in negative responses towards such advertisements (Fallows, 2005 Greenspan, 2004; Hansen, 2002; Hotchkiss, 2004). Building on these views, Animesh (2005) suggests that consumer behavioural responses toward SSA might be enhanced if the consumers become more educated in relation to the SSA practise (DoubleClick, 2006). Consequently, the greater the consumer‟s

42

knowledge of the SSA mechanism, the more likely the behavioural responses towards such advertising will be favourable (Animesh, 2005).

In summary, the anecdotal SSA research suggests that online user knowledge of SSA is limited. Given that there is little information about how consumers perceive SSA, the role and influence of consumer knowledge of SSA is unclear. As subjective knowledge is a better predictor of consumer behaviour in comparison to objective knowledge (Flynn and Goldsmith, 1999), and that subjective knowledge has a significant impact on consumer responses (Alba and Hutchinson, 1987; Berger and Smith, 1998; Flynn and Goldsmith, 2001; Rao and Monroe, 1988; Shon and Leckenby, 2004; Ward and Lee, 2000), subjective knowledge of SSA has been chosen as a determinant factor of consumer response towards SSA in the current study.

The preceding discussion has shown that prior experience and subjective knowledge are important factors that influence consumer processing of, and responses towards, advertising (Alba and Hutchison, 1987). In addition, brand familiarity is also viewed as an important determinant that influences consumer behavioural responses towards offline and online advertising (Bettman and Park, 1980; Campbell et al., 2003; Ha and Perks, 2005; Keller, 1991; Park and Lessig, 1981; Tam, 2008). Therefore, in terms of the current research, familiarity of brands is assessed as an important factor for understanding consumer response towards SSA. 2.5.3 Familiarity of Brands (or Websites) Included in the SSA Generally, brand familiarity has been identified, by Alba and Hutchinson (1987), Hoch and Deighton (1989), and Kent and Allen (1994), as the consumers‟ level of direct and indirect experience with products or brands. These direct and indirect experiences stem from factors such as advertising exposure, interactions with salespeople, word-of-mouth communication, and the usage and purchasing of brands (Alba and Hutchinson, 1987). More recently, Pieters, Warlop, and Wedel, (2002) identified brand familiarity as an individual‟s subjective experience. This definition was supported by Campbell et al. (2003), who suggest that brand familiarity captures the consumers‟ brand knowledge structures and brand associations that exist within consumers‟ minds. However, for the purpose of the current research, familiarity of brands (or websites) included in the advertising is viewed as the extent to which importance is attached by consumers to familiar brands (or websites) contained within the 43

advertising message (Simonin and Ruth, 1998). Such familiarity is directly related to the amount of time spent processing the SSA, regardless of the type or content of the processing involved (Baker et al., 1986).

The relevant marketing literature also identifies brand familiarity as influencing the process of consumers‟ decision-making (Bettman and Park, 1980; Coates, Butler, and Berry, 2006; Martinez, Polo, and Chernatony, 2008; Park and Lessig, 1981), advertising effectiveness (Campbell and Keller, 2003), and behavioural responses (Soderlund, 2002). Further, extensive research has demonstrated that the familiarity of brands included in the advertising, influences consumer engagement with, and processing of advertisements (Keller, 1991; MacKenzie and Spreng, 1992). For example, Keller (1991) found that the consumers‟ processing of advertising for familiar brands may be described as less extensive and more confirmatory in comparison to advertising for unfamiliar brands. Similarly, advertisements for familiar brands were found to be more effective in terms of achieving the desired communication goals compared to advertisements for unfamiliar brands (Alba and Hutchinson, 1987; Coates et al., 2006; Kent and Allen, 1994; Snyder, 1989). Holden and Vanhuele (1999) confirm these findings, noting that familiar brands tend to be favoured; because familiarity signals that the brands are tried and trusted. Additionally, familiar brands are more effective and noticeable within the advertising clutter (Dahlen, 2001). Indeed, Kent and Allen (1994) conclude that familiar brands are perceived as more important and relevant and, consequently, consumers spend more effort and time processing these advertisements.

Within the branding literature, researchers have identified a strong positive relationship between the familiarity of brands and consumers‟ attention towards advertising (Alba and Hutchinson, 1987; Chattopadhyay, 1998; Haider and French, 1999; Pechmann and Stewart, 1990). These findings are consistent with other studies‟ results that show consumers tend to pay more attention to advertising messages for familiar brands than for unfamiliar brands (Chattopadhyay, 1998; Kent and Allen, 1994; Park and Stoel, 2005). Indeed, Park and Stoel (2005), suggest that brand familiarity increases the consumers‟ level of confidence, leading them to perceive a lower degree of risk. Further, it appears that advertisements for familiar brands generate more favourable evaluations of the brands (Gregan-Paxton and John, 1997; Holden and Vanhuele, 1999; Janiszewski, 1993; Lee and Lee, 2007).

44

As discussed above, there has been extensive research on the relationship between brand familiarity and consumer behaviour in the offline advertising context (Alba and Hutchinson, 1987; Fazio, 1986 Keller, 1991; MacKenzie and Spreng, 1992). However, the role of brand familiarity seems to have been overlooked in the context of Web advertising in general, and SSA in particular. Consequently, it could also be that consumers are more likely to engage and pay greater attention to Sponsored Search Advertisements that include familiar brands rather than unfamiliar brands. The preceding sections identified the need for further exploration of particular determinants that may influence consumers‟ attention towards SSA. As such, three determinant variables were identified: prior experience, subjective knowledge and familiarity of brands (or websites) included in the SSA. In addition, a discussion supporting the choice of these particular variables was presented. Following on from this, and based on the previous discussion in section 2.4.2, that Web advertising effectiveness may be measured using determinants, process and outcomes (Rodgers and Thorson, 2000), the next research step is to examine the outcomes of attention towards SSA. Thus, these outcomes will be explored in terms of consumers‟ attitudes towards SSA and intention to click on SSA in the next section. More specifically, there is expectation that attention of online users towards SSA may lead to higher information processing and more positive attitudes (Cialdini, Petty, and Cacioppo, 1981), especially as higher attention makes attitudes more persistent and resistant to negative information (Haugtvedt and Strathman, 1990). However, as attitudes towards SSA are still unclear from the literature, the current study will investigate further attitudes toward offline and online advertising to better understand consumer attitudes toward SSA.

2.5.4 Attitude towards SSA Attitude, in the context of offline and online advertising, has received considerable attention within the literature. Research into the role of attitude towards advertising identifies the consumer‟s overall evaluation of the advertising as the first communication effect most likely to occur after consumers attend to, or are exposed, to advertising messages, namely, the consumer attitude towards advertising (Janiszewski, 1993; Rossiter and Bellman, 1999). Moreover, consumer attitude towards advertising is an important indicator of advertising effectiveness, and influences the consumer‟s response to the advertising (MacKenzie et al., 1986; Miniard, Bahtla, and Rose, 1990). However, to gain a better understanding of

45

consumer attitudes towards advertising and, more specifically, SSA, the consumer behaviour and broader marketing literature related to attitudes will be explored.

Within the consumer behaviour literature, the concept of attitude towards advertising seems to play an important role in predicting and understanding consumer behaviour (Jayawardhena, 2004; Leone, Perugini, and Ercolani, 1999; Priester et al., 2004), because it is formed as a result of attending to, and noticing, advertising (Park, 2002; Schiffman and Kanuk, 2000; Yoon and Lee, 2007). Although, there is some debate regarding a precise definition of attitude, there is a general agreement that attitude towards advertising can be viewed as “… a learned predisposition to respond in a consistently favourable or unfavourable manner to advertising in general” (Lutz, 1985, p. 53).

Over the past two decades, attitudes toward advertising have been studied more than any other concept in the marketing domain (Mittal, 1994; Pollay and Mittal, 1993; Shavitt, Lowrey, and Haefner, 1998; Wang et al., 2002). In particular, consumer attitude toward advertising has been largely examined because of its relation to consumer attention and responses towards advertisements (Schlosser et al., 1999) and its influence on attitude towards specific advertisements and subsequent behavioural intentions (Bruner and Kumar, 2000; Goldsmith and Lafferty, 2002; Lutz, 1985; McMillan et al., 2003; Mehta, 1994; Poh and Adam, 2002). Thus, the research reveals that a positive attitude towards advertising is vital to the long-term success of companies and the sustainability of their consumers.

Another area of attitudinal research is the relationship between consumer attitudes and advertising within the context of specific delivery mediums, for example, television (Alwitt and Prabhaker, 1992; Mittal, 1994), print (Yi, 1990), direct marketing (Korgaonkar, Karson, and Akaah, 1997), and the Web (Abd Aziz et al., 2008; Brackett and Carr, 2001; Ducoffe, 1995; Schlosser et al., 1999; Chen and Well, 1999; Wolin and Korgaonkar, 2003). Indeed, Chen and Wells (1999) suggest that investigating the impact of consumer attitudes toward Web advertising may enhance the evaluation of consumer responses toward Web advertising and, moreover, as the topic for the current research is concerned with Web advertising, an investigation of consumer attitudes toward Web advertising is warranted.

Although several studies have explored the underlying structure of consumer attitudes toward Web advertising (Abd Aziz et al., 2008; Chen and Wells, 1999; Ducoffe 1995; Eighmey 46

1997; Schlosser et al., 1999), there does not seem to be a definitive conclusion regarding the influence of Web advertising upon consumer attitudes. For example, a number of researchers found that consumers hold positive attitudes towards Web advertising (Calisir, 2003; Diaz, Hammond, and McWilliam, 1996; Ducoffe, 1996). However, it is important to note that, at the time of Ducoffe (1996), and Diaz et al.‟s (1996) studies, Web advertising was perceived as entertaining and informative as the intrusive pop-up advertisements had not, as yet, been introduced.

Conversely, other researchers found that consumers hold negative attitudes towards Web advertising (Brackett and Carr, 2001; Cho and Cheon, 2004; Mehta and Sivadas, 1995). For example, Mehta and Sivadas (1995) examined online users‟ attitudes toward email advertising; they identified that consumers tended to hold negative attitudes towards email advertising. In addition, Web advertising was perceived, by many, to be intrusive (Li et al., 2002), irritating and annoying (Brackett and Carr, 2001; McCoy et al., 2008; Reed, 1999). Moreover, due to such negative attitudes consumers tended to ignore Web advertisements (Cho and Cheon, 2004; Li et al. 2002; McCoy et al., 2008). However, Rust and Varki (1996) have warned that Web advertising should be less intrusive as forcing consumers to be exposed to the advertisements may lead to negative attitudes towards advertising. If this is so, then it can be postulated that consumers will pay more attention to, and hold more positive attitudes towards, unintrusive and targeted Web advertisements, such as SSA (Bloem, 2003). Indeed, Janiszewski (1993) suggests that attention towards advertising may enhance a consumer‟s attitude towards advertisements. Therefore, the current research extends this perspective by exploring the relationship between consumer attention and attitude towards advertising in the context of SSA.

Other perspectives of Web advertising research suggest that consumer attitude towards advertising is an important determinant of their responses and behaviours (Abd Aziz et al., 2008; Chen and Wells, 1999; Stevenson, Bruner, and Kumar, 2000; Wolin, Korgaonkar, and Lund 2002). For example, Stevenson et al. (2000) found that a negative attitude towards advertising was related with a negative consumer behavioural response towards the advertising, while Wolin et al. (2002) found that consumers, who held positive attitude towards Web advertising, were more likely to respond favourably towards the advertisements. Indeed, Mehta (2000) argued that consumers who had positive attitudes towards advertising were more likely to be persuaded by advertising. Support for this view is 47

provided by Korgaonkar and Wolin‟s (2002) finding that positive attitudes towards Web advertising were more likely to foster higher behavioural intentions. However, the relationship between attitude towards Web advertising and consumer behavioural intention is not conclusive, given that Karson et al. (2006) could not identify any significant relationship between attitudes and behavioural intentions across three different groups, all with different attitude levels. As such, the relationship between attitude and behavioural intention warrants further investigation.

2.5.5 Behavioural Intentions (Intention to Click on SSA) Behavioural intention refers to an individual‟s willingness to perform (Ajzen, 2002) or not to perform a specific future behaviour (Konerding, 1999); it has been considered as an important predictor of an individual‟s behaviour (Ajzen, 2002; Bagozzi, Wong, and Bergami, 2000; Castañeda Muñoz-Leiva, and Luque, 2007; Fishbein and Ajzen, 1975). Behavioural intention has often been examined within the framework of the Theory of Reasoned Action (TRA) (Fishbein and Ajzen, 1975). According to the TRA, the best predictor of behaviour is behavioural intention, which is largely influenced by consumers‟ attitudes (Fishbein and Ajzen, 1975) and, therefore, the attitude-to-behaviour relationship has received considerable attention in the marketing literature (Bagozzi et al., 2000; Castañeda et al., 2007; Chiang and Dholakia, 2003; Notani, 1997). For example, much of the advertising related literature has used behavioural intentions and their antecedents (including attitudes towards advertising) to measure advertising effectiveness and consumer responses towards advertising (for example, Brown and Stayman, 1992; Li and Bukovac, 1999; Reading et al., 2006; Thompson and Panayiotopoulos, 1999). Given that the consumers‟ intention to act upon advertisements is one of the most crucial concerns for advertisers, several offline traditional advertising studies have attempted to determine the range of consumers‟ behavioural intentions when being exposed to advertising stimuli. For example, Cronin and Taylor (1992) have used purchase intentions as a measure for behavioural intentions, while, Boulding et al. (1993) favour purchase intentions and willingness to recommend. Alternatively, Parasuraman, Zeithaml and Berry (1994) propose five dimensions to predict behavioural intentions: loyalty to company, propensity to switch, willingness to pay more, external response to a problem, and internal response to a problem. However, within the Web advertising context, intention to click has been used to capture consumers‟ behavioural intentions (Cho, 2003; Jin and Jun, 2007; Shamdasani et al., 2001; 48

Yoo et al., 2004). In particular, intention to click is defined as the likelihood that a consumer will engage in a clicking behaviour on the advertisement (Shamdasani et al., 2001). Within the Web advertising industry, one of the most common measures for consumer behavioural responses and advertising effectiveness is the click through (Bhat et al., 2002; Cho, 2003; Chandon et al., 2003; Manchanda et al., 2004; Yao and Chen, 2006; Yoo et al., 2004), which refers to voluntary behavioural action (Chandon et al., 2003; Hollis, 2005; Park, 2002) that measures responses to advertisements (Chandon et al., 2003; Collins, Enders, and Soto, 2001), or visits to websites (Manchanda et al., 2004), and shows the consumer‟s interest in Web advertisements (List, 2002). However, only a few studies have touched on consumer intention to click on Web advertisements in general (Cho, 2003; Rodgers, 2002; Shamdasani et al., 2000; Yoo, 2007; Yoo et al., 2004) and Sponsored Search Advertisements in particular. These studies suggest that consumer intention to click on Web advertisements is considered a behavioural response to Web advertisements for acquiring more detailed information about the advertised product or service (Yoo, 2007), and that it is influenced by consumer attitude towards Web advertising (Burns and Lutz, 2006; Cho, 2003; Yoo et al., 2004). Therefore, in the current research, the relationship between attitude towards SSA and consumer intention to click on SSA will be investigated to ascertain if the attitude-behaviour consistency (as discussed previously) holds in the SSA context.

2.6

Theoretical Model and Propositions Development

Based on the parent disciplines of Web advertising and consumer behaviour, discussed in the previous sections, this section presents a theoretical model that conceptualises the relationships amongst the focal constructs, that is, the impact of consumer related factors on consumer attention towards SSA and the outcomes of that attentional process which includes attitude towards SSA and intention to click on SSA (see Figure 2.2). The model extends Rodgers and Thorson‟s (2000) contention that consumer response to SSA comprising three elements: determinants, process and outcomes. The determinants are those factors that influence consumers‟ consumers‟ attention towards SSA; while process represents consumer initial processing of the SSA (Webb, 1979) which is represented in the current research by consumer attention to SSA. The final element of Consumer Response Towards SSA Model involves outcomes, namely, those factors that result from attending to SSA, and characterised as consumer attitudes towards, and intention to click on, Sponsored Search Advertisements.

49

Figure 2.2

Preliminary Conceptual Model of Consumer Response Towards SSA

Prior Experience of SSA

Subjective knowledge of SSA

Attention to SSA

Attitude toward SSA

Intention to click on SSA

Familiarity of Brands (or Websites)

Source: Developed for this research The following section is divided into four parts. The first section examines the determinants of consumer attention towards SSA: prior experience with SSA, subjective knowledge of SSA, and familiarity of brands (or websites) included in the SSA. The second part discusses the relationship between consumer attention towards, and consumer attitude towards, SSA. The third part addresses the impact of consumer attitude towards SSA on consumer intention to click on SSA. Finally, the fourth part concludes this section.

2.6.1 Determinants of Attention towards Sponsored Search Advertising (SSA) As discussed previously, one of the most crucial issues for advertising on the Web is the huge amount of Web advertisements and, yet, often there is limited attention paid by online users to those advertisements (Davenport and Beck, 2001; Dreze and Hussherr, 2003; Yoon and Lee, 2007). Indeed, according to Drèze and Hussherr (2003) less than half of the online users pay attention to Web advertisements. Thus, any failure to capture the consumers‟ attention reduces the effectiveness of the advertising and, consequently, minimises the advertiser‟s ability to achieve their long term communication and marketing goals (Pieters and Wedel, 2004). With a businesses‟ success resting on attracting consumer attention to Web advertisements, it has become essential to draw the eye of the consumer toward such advertisements.

50

Given that many consumers tend to pay little attention to Web advertisements during their online activities, and that most previous Web advertising studies (as discussed in section 2.5) have focused on consumer avoidance of Web advertisements, consumer attention towards Sponsored Search Advertisements is the focus of the current research. Additionally, as the review of the SSA literature suggests, that there has been little examination of the determinants of consumer attention towards SSA. For this reason, these determinants are investigated to identify how prior experience, subjective knowledge, and familiarity of brands in the SSA may influence consumer attention towards SSA.

2.6.1.1 The Relationship between Prior Experience and Attention Prior experience has been shown to be a significant predictor of consumer behaviour (Castañeda et al., 2007; Fazio and Zanna, 1981; Smith and Swinyard, 1983); information learned from prior experience is perceived to be valuable source in making decisions. Within the context of consumer behaviour, many researchers suggest that prior experience with advertisements plays a major role in consumer attentive processing towards traditional offline advertisements (Cronin and Menelly, 1992; Eysenck and Keane, 1995; Lee and Faber, 2007; Lodish et al., 1995; MacKenzie et al., 1986). For example, MacKenzie et al. (1986) found that prior experience was shown to influence the amount of attention a consumer gives to advertisements. More specifically, Cronin and Menelly (1992) noted that prior negative experience may lead consumers to avoid advertisements rather than attend to them.

Although there has been limited research focusing on the impact of prior experiences on consumer attention towards advertising on the Web, several Web advertising studies have examined the impact of prior experience on consumer avoidance of Web advertisements (Cho and Cheon, 2004; Jin, Morris, and Cho, 2005). For example, Cho and Cheon (2004), Guo, (2009), and Jin et al. (2005) determined that consumers tended to avoid Web advertisements as a result of prior negative experience with those advertisements. Given that prior experience tends to shape consumers‟ perceptions and judgements (Castañeda et al., 2007; Fazio and Zanna, 1981; Smith and Swinyard, 1983), Web users may also hold certain perceptions about SSA. Such preconceptions, formed by prior experiences with Web advertising, may result in either consumer attention or avoidance towards those advertisements (Ducoffe and Curlo, 2000; Guo, 2009). It is apparent; therefore, that consumer attention towards advertising is greatly influenced by prior experience with advertisements (Eysenck and Keane, 1995; Guo, 2009; Lee and Faber, 2007; Lodish et al., 1995; MacKenzie et al., 1986), and further that 51

prior experience with SSA may relate to consumer attention towards those advertisements. For this reason, the following proposition is postulated:

P1:

Prior experience with SSA is significantly positively associated with consumer attention towards SSA.

2.6.1.2 The Relationship between Subjective Knowledge and Attention As discussed previously, prior consumer behaviour research shows that subjective knowledge is related to consumer response and behaviour towards advertisements (Alba and Hutchinson, 1987; Berger and Smith, 1998; Flynn and Goldsmith, 1999; Rao and Monroe, 1988; Shon and Leckenby, 2004). According to Shon and Leckenby (2004), this occurs because it “… creates a psychological condition by which an individual‟s motivation is influenced and shaped” (p.45). In the context of offline traditional advertising, Smith and Wortzel (1997) identified subjective knowledge as having an impact on consumers‟ motivation to process advertisements. Similar findings have been reported by Cramphorn (2004), namely, that subjective

knowledge

shaped

consumer

responses

to,

and

perceptions

towards,

advertisements. More specifically, as outlined in section 2.5.2, consumer subjective knowledge may either constrain or enhance the amount of cognitive processing engaged in by the consumers (Petty et al., 1983).

Although the relationship between subjective knowledge and consumer attention towards advertisements has not been directly examined, previous consumer behaviour research (in the context of offline traditional advertising) has argued that subjective knowledge has an important role in facilitating consumer comprehension of advertising, and thus enables consumers to recognise and assess advertising cues more efficiently (MacInnis and Jaworski, 1989; Sujan, 1985). Additionally, such subjective knowledge appears to foster consumers‟ engagement at deeper levels of advertising processing (Olson, 1980). More specifically, consumers with a high subjective knowledge in a certain domain are more likely to engage in higher elaborative processing of the advertisements in comparison to those consumers with less subjective knowledge (Meeds, 2004; Park et al., 1994).

As discussed in section 2.5.2, few researchers have examined the impact of subjective knowledge of SSA on consumer behaviour, or how consumers‟ perceive them. While not referring to subjective knowledge, specifically, in the domain of SSA, some anecdotal studies 52

have reported that online users‟ respond negatively towards SSA when their knowledge regarding SSA was limited (for example, Consumer WebWatch, 2002; Fallows, 2005; Greenspan, 2004; Hansen, 2002; Hotchkiss, 2004; Marble, 2003). It seems reasonable, therefore, to assume that consumer behavioural responses toward SSA would be enhanced if consumers become more knowledgeable in relation to SSA practise (Animesh, 2005; Animesh et al., 2007; DoubleClick, 2006). Consequently, subjective knowledge about SSA may be related to consumer attentive processing of such advertising. Moreover, those online users who possess higher subjective knowledge about SSA are then expected to be more likely to give greater attention towards those advertisements. These arguments give rise to proposition two:

P2: Subjective knowledge about SSA is significantly positively associated with consumer attention towards SSA. 2.6.1.3

The Relationship between Familiarity of Brands (or Websites) and Attention

Brand familiarity has been shown (section 2.5.3) to have a positive relationship between familiarity of brands and attention towards advertising (Alba and Hutchinson, 1987; Haider and French, 1999; Pechmann and Stewart, 1990; Chattopadhyay, 1998). For example, advertisements that include familiar brands were found to be more effective in terms of encouraging consumers to pay more attention towards them (Chun and Wolfe, 2001, Rayner et al., 2001; Rosbergen et al., 1997; Yantis, 2000). Further, these familiar brand advertisements also generated more favourable evaluations towards such advertisements (Gregan-Paxton and John, 1997; Holden and Vanhuele, 1999; Janiszewski, 1993; Lee and Lee, 2007).

Although such a relationship has received considerable support in offline advertising studies, Web advertising and SSA literature has overlooked the imperative role of familiar brands (or websites) in the SSA context. However, based on the preceding discussion (section 2.5.3) and, given brand familiarity‟s significant role within the traditional offline advertising context, it is plausible that, within the Web context, SSA for familiar brands will be more effective in gaining consumer attention than unfamiliar brands. Therefore, it is posited that:

P3:

Familiarity of brands (or websites) included in the SSA is significantly positively associated with consumer attention towards SSA.

53

2.6.2 The Relationship between Attention and Attitude Within the conceptual model (Figure 2.2), consumer attention towards SSA is shown to be related to consumer attitudes towards SSA. Research focusing on the role of attitude toward advertising (section 2.5.4) suggests that the first communication effect that most likely to occur after consumers attend to advertising messages, is consumer attitude towards advertising (Janiszewski, 1993; Rossiter and Bellman, 1999). Although some consumer behaviour researchers have found no direct relationship between consumer attention towards advertisements and attitudes towards advertising (Thorson, Chi, and Leavitt, 1992; Thorson, Zao, and Friestad, 1988), many others support the existence of a relationship between attention and attitude in the context of offline traditional advertising, including television advertising (Chattopadhyay and Nedungadi, 1992; Janiszewski, 1993; Olney, Holbrook, and Batra, 1991) and print advertising (Babin and Burns, 1997). Earlier studies supporting this view were advocated by Cialdini et al. (1981), Petty and Cacioppo (1986) and Petty et al. (1983), namely, that consumer attention towards advertisements may generate more positive attitudes towards advertisements. Moreover, such consumer attention may keep them from mentally switching focus to an alternative activity, and resulting in greater information processing. Similarly, as noted by Haugtvedt and Strathman (1990), higher consumer attention towards advertising makes their attitudes more persistent and resistant to negative information.

Similarly, it is also argued that the purpose of attention is to allocate cognitive resources to process specific information from a very complex and diverse world. For this reason, consumer attention is given to advertisements from which they may attain positive or negative outcomes (Eysenck and Keane, 1995; Kahneman and Treisman, 1984). Thus, attitudes may serve as an orienting function, where attention is paid to advertisements that might produce beneficial outcomes (Eysenck and Keane, 1995; Kahneman and Treisman, 1984).

Although much offline traditional advertising literature has confirmed that a positive relationship exists between consumer attention and attitude towards advertising, such a relationship has received scant attention in the Web advertising context. More specifically, outcomes of consumers‟ attention in the Web context may operate in different directions, depending on the type of consumer exposure to Web advertising. For example, forcing online users to be exposed to Web advertising messages (such as the case with pop-up advertising) 54

may generate negative attitudes towards such advertising (Li et al., 2002; McCoy et al., 2008). This view is consistent with the earlier argument of Rust and Varki (1996), advocating that Web advertising should be less intrusive because forcing consumers to be exposed to advertisements may generate negative attitudes towards such advertising. Other Web advertising studies have also shown that consumers tend to ignore intrusive and annoying Web advertisements, often holding negative attitudes towards them (Brackett and Carr, 2001; Cho and Cheon, 2004; Li et al. 2002; McCoy et al., 2008; Reed, 1999). Alternatively, consumers are expected to pay more attention, and hold more positive attitudes, towards unintrusive and targeted Web advertisements such as SSA (Bloem, 2003; Mehta and Sivadas, 1995). Given the above discussion, it is argued that consumer attention towards SSA is related to consumer attitudes towards SSA. Therefore, it is proposed that:

P4: Consumer attention towards SSA is significantly positively associated with consumer attitudes towards SSA. 2.6.3 The Relationship between Attitude and Intention to Click The attitude-behaviour relationship in which behavioural intention is predicted by attitudes has been identified in various contexts and under various conditions, including the field of technology (Davis, 1986; Davis et al., 1989; Kim, Chun, and Song, 2009; Muthitacharien et al., 2006), mobile advertising (Chiou, 1998; Xu, 2007), online shopping behaviour (Shim et al., 2001; Van Der Heijden Verhagen and Creemers, 2003), and offline traditional advertising (Brown and Stayman, 1992; Mitchell, 1986; Shimp, 1981). This attitudinal and behavioural intention relationship (as discussed in section 2.5.5) was originally advocated by Fishbein and Ajzen (1975). More recent advertising research has validated this theory, namely, that the consumers‟ attitude toward the object (in this case the SSA) will influence their intention to use the object (intention to click on SSA) (MacKenzie et al., 1986). More specifically, this relationship has been found in some Web advertising studies which have reported a significant effect of attitude towards Web advertising on intention to click on Web advertisements (Burns and Lutz, 2006; Cho, 1999; Jin and Jun, 2007; Shamdasani et al., 2001; Yoo, 2008). However, it appears that few studies have examined such a relationship in the SSA context. Given the above discussion, it appears plausible that consumers‟ attitudes towards SSA are related to their behavioural intentions of clicking on Sponsored Search Advertisements. Therefore, the following proposition is presented:

55

P5:

Consumer attitude towards SSA is significantly positively associated with consumer intention to click on SSA.

2.6.4 Section Summary Based on the preceding literature review, a preliminary theoretical model has been provided to address the research objective and associated research questions, of the study. In doing so, the theoretical model conceptualises SSA from a consumer behaviour perspective. The Consumer Response Towards SSA Model also assists in the systematic development of five propositions (as presented above) that will direct the data collection required to explore the important issues contained in the current research. While the preliminary Model of Consumer Response Towards SSA, is effective in terms of conceptualising the theoretical relationships among the focal constructs (prior experience, subjective knowledge, familiarity of brands (or websites), consumer attention, consumer attitude and intention to click), the proposed model (Figure 2.3) also clearly identifies the propositions of this study. According to Figure 2.3, proposition One proposes that prior experience with SSA has a positive relationship with consumer attention towards SSA. Proposition Two postulates that subjective knowledge of SSA has a significant positive relationship with consumer attention towards SSA. Furthermore, proposition Three shows that familiarity of brands (or websites) included in the SSA and consumer attention are significantly positively related. Proposition Four proposes that there is a significant positive relationship between consumer attention towards SSA and consumer attitudes towards SSA, while proposition Five posits that consumers attitudes towards SSA has a significant positive relationship with consumers intention to click on Sponsored Search Advertisements. The theoretical model is expected to further our understanding of how do consumers process and respond to SSA and to determine the outcomes of consumer response to SSA.

56

Figure 2.3

Preliminary Conceptual Model of Consumer Response Towards SSA

Prior Experience of SSA

Subjective knowledge of SSA

Familiarity of Brands (or Websites)

P1

Attention to SSA

P2

P4

Attitude toward SSA

P5

Intention to click on SSA

P3

Source: Developed for this research The review presented in this chapter highlights a number of gaps in the literature, which has raised the following research questions:

RQ1: To what extent does consumer related factors (including prior experience with SSA, subjective knowledge of SSA, and familiarity of brands (websites) included in the SSA) influence consumer attention towards SSA? RQ2: To what extent does consumer attention towards SSA influence consumer attitude towards SSA? RQ3: To what extent does consumer attitude towards SSA influence consumer intention to click on SSA?

2.7

Conclusion

This chapter has reviewed the extant literature regarding Web advertising and consumer behaviour. The gaps in the literature have been identified in terms of the relationship among consumer related factors and consumer responses towards SSA. The preceding sections showed the need for further exploration of determinants that may influence consumers‟ attention towards SSA. Three determinant variables were identified (prior experience with SSA, subjective knowledge about SSA, and familiarity of brands (or websites) included in 57

the SSA). The reasons for choosing these variables have been outlined. That is, this Chapter detailed how consumer related factors may alter their attentional processing of SSA and associated behavioural outcomes. Of particular interest is how these factors may alter important outcome variables in consumer response towards advertising, including consumers‟ attitudes towards SSA and consumers intention to click on Sponsored Search Advertisements.

The determinants (consumer related factors), process (consumer attention towards SSA) and expected outcomes (consumers‟ attitudes towards SSA and consumers intention to click on SSA) have been incorporated into the preliminary conceptual model of Consumer Response Towards SSA (Figure 2.3). Each construct, presented in the model and the associated relationships, have been discussed in detail. Importantly, all the relationships between the constructs have a theoretical underpinning and support for their inclusion in the model has been derived from the extant literature. As a result of the above investigation, a series of propositions have been developed to empirically test the model (see Table 2.1 below).

Table 2.1

Research Propositions

No.

Research Propositions

P1

Prior experience with SSA is significantly positively associated with attention towards SSA.

P2

Subjective knowledge about SSA is significantly positively associated with attention towards SSA.

P3

Familiarity of brands (or websites) included in the SSA is significantly positively associated with attention towards SSA.

P4

Consumer attention towards SSA is significantly positively associated with consumer attitude towards SSA.

P5

Consumer attitude towards SSA is significantly positively associated with consumer intention to click on SSA.

Source: Developed for this research

58

CHAPTER THREE RESEARCH DESIGN AND QUALITATIVE METHODOLOGY FOR PHASE ONE: SEMI-STRUCTURED INTERVIEWS 3.1

Introduction

This chapter discusses the qualitative methodology used in the first phase of this research to confirm the appropriateness of the theoretical Model of Consumer Response Towards Sponsored Search Advertising (SSA) outlined in Chapter Two. This chapter is divided into eight sections (Figure 3.1). Firstly, the chapter outline is presented, followed by a description of the various research paradigms along with a justification for the choice of a post-positivism (realism) approach in the current research (section 3.2). In section 3.3, the stages of the research design are outlined. Following this, a discussion of interviewing methodologies is presented (section 3.4). Next, an examination of the validity and reliability of the interviewing method is presented (section 3.5), and the process involved in planning, managing and interpreting the semi-structured interviews is discussed (section 3.7). Next, the data analysis collected through qualitative interviewing, is presented (section 3.8). Finally, concluding remarks are given in section 3.9.

59

Figure 3.1

Outline of Chapter Three

3.1 Introduction

3.2 Plan of this Research

   

3.3 Paradigm Positivism Critical theory Constructivism Post positivism

3.9 Conclusion

3.8 Data Interview Analysis and Interpretation

3.7 Planning and Implementation of SemiStructured Interviewing Approach

3.4 Research Design

3.5 Choice of SemiStructured Interviewing

3.6 Validity and Reliability of SemiStructured Interviewing

Source: Developed for this research

60

3.2

Research Plan

A research plan refers to the process of conducting research to enable the accomplishment of specified goals (Aaker and Day, 1990). The selection of any research plan involves a number of related steps (Aaker, Kumar, and Day, 200l; Denzin and Lincoln, 1994), beginning with locating the research study within a proper paradigm of research inquiry, and then managing the selection of an appropriate research methodology. Finally, the chosen research methodology specifies what methods are to be applied in the collection and analysis of the data. Next, the different types of research paradigms are described and a justification for the choice of the realism paradigm is provided.

3.3

Paradigms of Research Inquiry

Prior to discussing the method that will be applied in the current research, it is important to consider the paradigm that is most suitable to the study. In particular, selecting the appropriate research paradigm is vital to the research process in all areas of the study (Mangan, Lalwani, and Gardner, 2004), especially as it helps to understand the phenomenon, particularly within the human and social sciences (Creswell, 2003). Many researchers have viewed a paradigm as “… a basic orientation to theory and research” (Neuman, 2003, p.70), a basic set of beliefs (Lincoln and Guba, 2000), and a conceptual framework (Healy and Perry, 2000) that guides action in inquiry or research (Lincoln and Guba, 2000; Rocco et al., 2003). That is, a paradigm identifies how researchers view the world, how they relate to the object under study, and what they see as the nature of reality. Paradigms are often described by the way they deal with ontological, epistemological and methodological questions (Guba and Lincoln, 1994). As shown in Table 3.1, ontological questions focus on discerning the nature of reality (Creswell, 2003; Guba and Lincoln, 1994), whereas epistemological questions explore the nature of the knowledge, and the relationship between researchers and the people or phenomena under study (Creswell, 2003). Finally, methodological questions focus on the techniques used to collect and analyse the data (Creswell, 2003). These principles are usually interconnected, as the researcher who adopts a position on one of the principles is constrained by the position that may be taken on others. Furthermore, these principles have a direct link between inquiry and the practical issues related to the conduct of the research (Creamer, 2003).

61

3.3.1 Types of Paradigms There is no consensus in the literature about classifying research paradigms. Some authors emphasise the value of positivism and post-positivism, suggesting that other paradigms have emerged for these original paradigms (for example, Taylor, 1994). Others suggest that there are three major paradigms in the social sciences; positivistic, interpretive and critical (Hunt, 2002). In contrast, Guba and Lincoln (1994) argue for the inclusion of four primary paradigms: positivism, post-positivism, constructivism and critical theory as detailed in Table 3.1. Table 3.1

Principles of the Research Paradigms

Principles

Paradigms Positivism

Ontology (The nature of reality)

Epistemology (The nature of the relationship between the researcher and the reality)

Methodology (How the researcher finds the reality)

Post Positivism

Critical Theory

Constructivism

Objective real world, driven by universal laws.

Reality can never be fully apprehended, and an objective real world does not exist beyond the human mind.

Historical realism where reality is formulated over time by social, political, cultural economic, ethnic and gender forces.

Multiple local and specific constructed realities, which often do not exist, are constructed by human beings.

Dualist, objectivist findings are true; objectivity of the researcher is essential; and participants should be a distant observer.

Knowledge of an existing world can be approached through probable statistics, modified dualist.

Subjectivist cocreated findings, transactional.

The critical inquirer cannot be distant from the subject of the investigations since the nature of the inquiry is completely value determined.

Experimental and quasi-experimental designs, manipulations, interventions and verification of hypotheses rely essentially on quantitative methods.

Experimental and quasi-experimental designs, manipulations, interventions and verification of hypotheses. Can include both qualitative and quantitative methods.

Dialectic dialogue, primarily qualitative methods include observation, grounded theory and case studies.

Gather details and utilize inductive reasoning to develop hypotheses, theories and concepts. Primarily qualitative, but can include descriptive quantitative methods.

Source: Adopted from Amaratunga, Baldry, Sarshar, and Newton, 2002; Guba and Lincoln 1994; Healy and Perry, 2000; Schulze, 2003. These major paradigms have been identified as applicable in marketing research (Healy and Perry, 2000). Accordingly, this section discusses the four major paradigms used within marketing research, and justifies the choice (within the current research) of the realism paradigmatic approach to study the factor surrounding consumer response towards SSA. 62

Positivist Paradigm. Positivism has been the dominant mode of inquiry in social sciences for more than a century (Kim, 2003). Under this paradigm, reality is identified as what can be observed through the senses, and is viewed as relying on objective facts (Neuman, 2003). Additionally, positivists argue that the social world is assumed to be independent of human consciousness (Guba and Lincoln, 1999). That is, positivists consider social phenomena as having an objective and value free reality (Bettis and Gregson, 2001) since “time and context free generalizations” are possible (Onwuegbuzie, 2002, p.519). Thus, positivists believe that truly objective research is possible, they also consider it as the optimal approach to realise the world, and then predict future outcomes (Bettis and Gregson, 2001). The fundamental nature of the positivist view of the social sciences combines deductive logic in nature with research questions expressed as hypotheses that are subject to empirical testing (Falconer and Mackay, 1999), through accurate quantitative approaches (Table 3.1).

However, as positivists consider that research should use exact and rigorous measures (Neuman, 2003), that are context, value, and bias free (Plack, 2005) and are replicable (Perry, Reige and Brown, 1999), this line of inquiry is inappropriate as this research deals with variables in terms of complex, social and real life experiences (Perry et al., 1997). An additional reason for not using this paradigm is the paucity of knowledge about consumer response towards SSA, from a consumer behaviour perspective as much of the existing literature is descriptive and practitioner-oriented (for example, Hotchkiss, 2004; Mandelli, 2005).

Constructivist Paradigm. The Constructivism paradigm began as a countermovement to the positivist paradigm (Creswell, 2003); it is based fundamentally on qualitative observations and scientific study, in relation to how people learn to construct their own understanding and their knowledge of the world through both experiences and reflections (Bettis and Gregson, 2001; Healy and Perry, 2000). Thus, reality is assumed not to exist (Lincoln and Guba, 2000). Under this paradigm, researchers and the respondents under investigation engage in a dynamic interaction that creates the meaning of the findings (Healy and Perry, 2000). Therefore, and contrary to the positivist paradigm, knowledge is always constructed by humans and so is not value free (Bettis and Gregson, 2001). However, this form of inquiry excludes an analysis of the economic and technological aspects of businesses (providing a purely subjective framework) (Healy and Perry, 2000). For this reason, the constructivist 63

paradigm is unsuitable for the current research, which requires the use of measurable and objective concepts.

Critical Theory. The Critical theory paradigm is viewed as a synthesis of alternative paradigms, such as feminism, materialism and neo-Marxism, as they share a particular set of basic beliefs (Plack, 2005). Under this paradigm, reality is formulated over time by social, political, cultural, economic, ethnic, and gender forces (Table 3.1). Further, core knowledge is established through the interactions between the researcher and the respondents, whereby the former is a transformative intellectual who changes the social world (Guba and Lincoln, 1994), while the latter exists to keep critical theorists more informed (Perry et al., 1997). However, as the current research seeks to examine and understand the determinants and outcomes of consumer response toward SSA, from a consumer behaviour perspective, rather than change the values and behaviours of consumers, critical theory is not an appropriate paradigm for this research.

Post Positivism (Realism). The Post-positivism or realism paradigm is the paradigm most often adopted for social sciences research (Bettis and Gregson, 2001). Under this paradigm, the world is perceived to be independent of researchers and open to different perceptions. Realism researchers argue that reality is assumed and can be influenced by the researchers‟ viewpoints (Schulze, 2003). That is, reality reflects the viewpoints of both the researcher and the participants. The aim of research under this paradigm is to discover cause and affect relationships and to predict and control future consumer behaviours by an examination of current behaviour (Plack, 2005). Validity is thus determined by its ability to generalise and to test the theory, based mainly on empirical and logical evidence (Schulze, 2003). Methodologically, realists employ experimental, quasi-experimental designs and/or some qualitative techniques to obtain a clearer picture of what is happening in the real world (Bettis and Gregson, 2001; Plack, 2005). That is, the realism paradigm values and recommends the use of various approaches that lead to insights that extend beyond the realm of measurable and discoverable facts (Perry et al., 1999).

Given the need to discover the real world (consumer response towards SSA), realism has been identified as the most appropriate paradigm for the current research (Denzin and Lincoln, 1994). The methodologies (used in the current research) seek to observe the real world, using qualitative and quantitative methods, for example semi-structured interviewing 64

and online surveys (Healy and Perry, 2000). Therefore, following the literature review, a series of qualitative semi-structured interviews were conducted to identify and organise observable facts. Then, a quantitative (an online survey) approach was used to collect the quantitative data for subsequent analysis.

In brief, the realism paradigm was chosen for this study as it enables the researcher to identify new variables that can be tested within the second phase of the quantitative analysis. Having discussed the choice of paradigm of inquiry, the following section addresses the case for the current research design. 3.4

Research Design

Research design is a framework used to assist in solving research problem(s) (Churchill, 1999; Kinnear et al., 1993); it involves a series of rational and strategic decisions about the selection of data collection approaches, tactical decisions regarding the type of sample to be chosen, choice of variables to be measured and scaled, and approaches to analysing the data (Aaker et al., 2001; Cavana, Delahaye, and Sekaran, 2001; Churchill, 1999). Two types of research designs (incorporating exploratory and explanatory variants) are used. A discussion of the benefits of each are presented below. Exploratory versus Explanatory Research. Exploratory research is conducted when there is limited (or an absence of) previous research in a particular area. Such research assists in providing the background to, and information about, the research objective, and is used to guide further research (Malhorta, 1999). The aim of exploratory research is to look for patterns, hypotheses or ideas that can be tested later (Burns and Bush, 2003). Typical exploratory research techniques include interviews, case studies, observation and reviews of existing related studies and secondary data.

In the current research, the exploratory approach included reviewing the literature of consumer behaviour associated with Web Advertising, and SSA in particular (Chapter Two). In addition, qualitative semi-structured interviews (Chapter Three) were used to provide insights about the research objective and to determine how consumers respond to the emerging form of Web advertising (SSA). This research commenced with an initial exploratory phase followed by an extensive explanatory research phase.

65

Conversely, explanatory research is often used to identify causes and effects of social phenomena and to predict how dependent variable (effect) will vary in response to variation in independent variable(s) (cause) (Heuser and Petersen, 2005). In the current research, the explanatory approach included using an online survey (Chapters Four and Five) that examined the extent to which specific consumer related factors impacted upon consumer response to SSA. A detailed explanation of the survey methodology is presented in Chapter Four.

Qualitative versus Quantitative Research. Broadly, research may be categorised into two distinct types: qualitative and quantitative (Bryman, 2006). Both are often used in social science studies, including marketing (Amaratunga et al., 2002; Zikmund, 2000). This section justifies the use of a combination of qualitative and quantitative research methods in the current research. Their features are compared to help identify why both techniques were used in this research. Table 3.2 outlines the main differences between qualitative and quantitative research. The purpose of the current research is to identify determinant factors influencing consumers‟ attention to, and subsequent outcomes of attitudinal and behavioural responses towards SSA. The in-depth literature review identified a range of consumer behaviour variables related to consumer response towards SSA, as presented in Figure 2.2, as well as a set of propositions. It was apparent that further qualitative research was warranted to help identify those issues (for example, not addressed in the relevant literature). The result was a series of semistructured interviews, conducted with a sample of Australian consumers who utilised the Web for information searches and the purchasing of goods and services. The interviews identified interviewee‟s responses toward SSA that were not directly observable in the imperfect reality (Healy and Perry, 2000; Patton, 1990). Those observations were incorporated into a revised conceptual Model of Consumer Response Towards SSA (section 3.4) which was later tested statistically using Partial Least Squares (PLS) (Chapter Five) (Aaker and Day, 1990; Sekaran, 2003).

66

Table 3.2

Comparison between Qualitative and Quantitative Methods Qualitative Method

Quantitative Method

Objectives

Description, exploration and discovery

Description, explanation and prediction

Type of research

Exploratory

Descriptive

Focus

Examining the breadth and depth of the phenomena

Testing specific hypotheses

Nature of reality

Multiple realities; subjective, personal, and socially constructed

One reality; objective

Reasoning

Inductive

Deductive

Research questions

What? Why? (Classification, meaning)

How many? Strength of association! (Enumeration, causation)

Hypotheses

Generation

Testing

Measurement

Researcher as instrument, „insider view‟

Psychological / physiological instruments, „outsider view‟

Approach

Flexible; natural setting (process oriented)

Highly controlled; experimental setting (outcome oriented)

Data collection

Unstructured or semi-structured techniques

Structured techniques

Sample

Purposive (evolving); small sample size

Statistical (predetermined) sample; large sample size

Data analysis

Coding, categories, themes; basic element of analysis is words/ideas

Statistical inference / statistical estimation; basic element of analysis is numbers

Outcome

Develop an initial understanding and sound base for further decision making

Used to recommend a final course of action

Source: Adopted from Amaratunga et al., 2002; Creswell, 1994, 2003; Hoepfl, 1997; Mayoux, 2000; Myers, 1997; Oka and Shaw, 2000; Patton, 1990; Zikmund, 2000. At its basic form, qualitative research is an inquiry process used to investigate social or human problems (Creswell, 1994; 2003; Myers, 1997); the process builds a holistic picture, reports detailed views of informants (Creswell, 1998) and focuses on words and observations in a natural setting (Amaratunga et al., 2002). The process elicits in-depth and new information (Hoepfl, 1997; Patton, 1990) about the phenomena that is not directly observable (such as behaviour and attitude) (Patton, 1990). 67

Further, qualitative research is used in many disciplines and fields; it includes a variety of approaches, methods and techniques (Bryman, 2006; Myers, 1997), such as observation, informal, unstructured and in-depth interviews, focus groups, and participant observation (Creswell, 1994, 2003). A study based upon qualitative research has the capability to provide detailed information about a relatively small number of objects under investigation (Patton, 1990), by exploring people‟s beliefs, experiences, attitudes, behaviours and interactions through a closer understanding of the subject‟s perspective (Falconer and Mackay, 1999; Hoepfl, 1997). For these reasons, qualitative research is considered an appropriate approach to be adopted in the first phase of the current study. The open-ended and responsive questioning techniques used are particularly suitable for encouraging participants to describe their behaviours, and to identify the major determinants and outcomes of their responses toward SSA (Fossey et al., 2002; Patton, 2002).

Additionally, qualitative research has a distinctive inductive nature, which can be used to investigate SSA, a relatively new topic in the academic marketing literature. Indeed, Denzin and Lincoln (1994) suggest that when a phenomenon is not well understood, and the relationships between the variables are not clear, the qualitative research method assists in the early phases of the research. Therefore, a qualitative approach was conducted to explore this „cutting edge‟ topic in greater depth and to generate ideas prior to measuring them.

In contrast, quantitative research involves the gathering and analysing of numerical data (Amaratunga et al., 2002) for the purpose of describing and exploring a particular phenomenon (Zikmund, 2000). Specifically, it involves drawing conclusions and testing hypotheses (Creswell, 2003). Thus, quantitative research was conducted in the second phase of the current research because of its ability to facilitate comparisons and statistical generalisations, through measuring the responses of a range of people to a limited set of questions (Neuman, 2003; Patton, 1990).

Although there are many important paradigmatic differences between qualitative and quantitative research (Johnson and Onwuegbuzie, 2004) (as shown in Table 3.2), there are also some similarities between them, which are sometimes disregarded. Also, as research is becoming increasingly interdisciplinary, complex, and dynamic (Johnson and Onwuegbuzie, 2004), researchers are adopting research approaches that complement one another (Bryman, 2006; Creswell, 2003). 68

In brief, qualitative and quantitative research methods are used in the current research in a complementary way to enable the researcher: to gain extensive in-depth information on factors surrounding consumer response towards SSA; and to clarify and confirm the appropriateness of the preliminary conceptual model (presented in Chapter Two). The model will then be tested through the quantitative research approach. Having discussed the current research design, the following section outlines the reasons for choosing the qualitative interviewing technique.

3.5

Choice of Semi-Structured Interviewing

This section identifies the nature of personal interviews and then justifies the reasons for choosing in-depth semi-structured interviews over other qualitative research methods. Then it presents an overview of the different interview methods and the relative advantages of semistructured interviewing. The final part details the approach taken by the researcher in conducting the semi-structured interviews.

Qualitative Interviewing. Interviews are often used as a key component in the social science research (Winchester, 1999). They help to improve understanding of the social phenomena (Meho, 2006), particularly in the marketing discipline (Neuman, 2003). Furthermore, they are one of the major sources of data collection commonly used in exploratory and descriptive research (Mathers, Fox, and Hunn, 2002). Qualitative interviewing refers to a basic mode of inquiry (Seidman, 1998). These interviews involve a purposeful conversation and verbal communication in which the interviewer asks pre-prepared questions and the respondents answer them; the result is a specific collection of information (Oatey, 1999). Further, qualitative interviews are a useful first step prior to applying other methodologies, such as surveys (Sewell, 1998). For example, the qualitative interview process provides a broad range of information about certain research questions resulting widening the knowledge base surrounding the problem (Mathers, et al., 2002; Oatey, 1999). Additionally, qualitative interviewing allows an in-depth analysis of an individual‟s experiences, feelings and views of the social world (Fossey et al., 2002; Sadava and McCreary, 1997; Seidman, 1998), and elicits opinions about complex and sensitive issues, which may not be directly observed through other means (Bryman, 2006; Hannabuss, 1996; Jarratt, 1996; Patton, 1990).

That is, the purpose of the interview is to explore the knowledge, opinions and beliefs of the individuals (Burns and Bush, 2003; Creswell, 2003; Sewell, 1998). Consequently, this 69

method was considered the most appropriate choice for the current research; an exploration of the core issues involved with the most important factors that influence consumer attention, attitudinal and behavioural responses towards SSA. Further, it is also an effective method for discovering other variables not detailed in the extant literature, thus adding new dimensions to the study.

Having identified the definition and purpose for collecting information using in-depth interviews, it is important to justify the main reasons for choosing in-depth interviews over other qualitative research approaches. Thus, the major features (both positive and negative) of interviews, in comparison with focus groups, case study research and observations are discussed in the next section.

Interviews vs. Focus Groups. Although focus groups are the most frequently used form of qualitative research, they were not employed in the first stage of the current research because of the type of the data to be collected (Malhotra et al., 2004). Focus group technique involves organised discussions (Malhorta, 1999) with a selected group of individuals, in order to explore their views and experiences on a particular topic (Gibbs, 1997; Threlfall, 1999). The technique assists the researcher in obtaining several perspectives about the same topic through producing data and insights that are likely to be exposed through the social gatherings and group interactions (Morgan, 1988, 1998; Rao and Perry, 2003). However, the researcher believed that exploratory information could be collected without relying on the synergistic effect of group interactions enhanced by focus groups.

Compared to in-depth interviews (which aim to understand individual behaviour, beliefs and feelings), a focus group elicits a multiplicity of views within a group context (Gibbs, 1997). The advantage of the in-depth personal interview compared to the focus group lies in the high control the researcher has over the produced data; thus, the researcher can reveal and explore each participant‟s definitive views on a topic (Gibbs, 1997; Morgan, 1988). Moreover, focus groups are not as strong as personal interviewing in terms of their ability to enhance direct probing of informant‟s knowledge (Threlfall, 1999). In addition, focus group research is limited in its ability of generalise findings into the whole population because of the difficulty of getting a representative sample, and then ensuring that the individuals participate, especially those who lack confidence or have communication problems (Gibbs, 1997). In

70

conclusion, in-depth interviewing methods offer advantages over focus groups and, consequently, are more suitable to be used within the first study phase.

Interviewing vs. Observation. Observation is the process of exploring behaviours and interactions in natural settings (Creswell, 2003; Mays and Pope, 1995). They represent a viable research approach in situations where the researcher seeks to avoid problems that may occur with other qualitative methods, such as making participants aware of their participation in the research (Judd, Smith, and Kidder, 1991). Thus, observation techniques are generally most useful when the researcher‟s presence may lead to unacceptable changes or modifications in the participant‟s behaviour (Cooper and Schindler, 2003; Creswell, 2003; Neuman, 2003). One of the main limitations of this approach is that the observer needs to be present when the observation process takes place (Cooper and Schindler, 2003). Additionally, observation techniques require the researcher to await for events or actions to happen; often, therefore, the process is slow and expensive (Cooper and Schindler, 2003; Gibbs, 1997). Taking these facts into account, the researcher decided that this method was inappropriate in the context of the current study.

Interviewing vs. Case Study. Researchers have used the case study research method for many years and across a variety of disciplines (Gerring, 2007; Tellis, 1997). The case study research technique facilitates empirical inquiries by examining contemporary phenomena within a real-life context (Myers, 1997; Seawright and Gerring, 2008; Weerd-Nederhof, 2001). This approach helps the researcher to understand complex issues and confirms what is already known through previous research. In particular, the approach is designed to elicit detailed participant opinions by using multiple sources of data (Gerring, 2007; Tellis, 1997). However, the case study method is not completely suited to the current research for two reasons. Firstly, a small number of case studies may not ensure a solid grounding for the establishment of reliable or general outcomes (Gerring, 2007; Tellis, 1997). Secondly, such research requires a medium to high level of prior theory development, while in-depth interviewing requires little or no prior theory (Riege, 2003; Seawright and Gerring, 2008; Yin, 1994).

In brief, qualitative interviewing has been identified as the most appropriate qualitative technique and the most useful collection tool for the first research phase. In addition, qualitative interview methods help to explore the research question in more depth (Mason, 71

2002), especially when the topic concerns the preliminary planning stage of a larger study, possibly a quantitative study (Amaratunga et al., 2002), as with the current research. In-depth interviewing also produces feedback, and excellent qualitative data (McClelland, 1994). Thus, the interview is highly flexible method and is ideal for the present study.

Interview Methods. Interestingly, interviewing techniques can be ranged from completely unstructured to highly structured (Bryman, 2006; Mathers et al., 2002; Oatey, 1999; Oka and Shaw, 2000). Accordingly, some interviews encourage lengthy and detailed replies, while others are designed to elicit short and specific responses. The three types of in-depth interviewing methods used in the social science literature are: unstructured, structured and semi-structured interviews (Bryman, 2006; Mathers et al., 2002; Oatey, 1999). A comparison of the main differences between these interviews types are discussed below along with a tabulated comparison (Table 3.3).

Unstructured interviews are usually a preliminary step in the research process to generate ideas; it can be defined as an informal interview, rather than being structured by a standard list of open-ended questions (Oatey, 1999). This approach encourages respondents to express, and talk freely about their thoughts (Mathers et al., 2002), allowing a greater in-depth exploration of the topic (McMurray et al., 2004). Further, the technique is beneficial in situations where fewer people are available to be interviewed and when responses are expected to differ greatly between interviewees (McMurray et al., 2004).

Structured interviews usually incorporate a list of specific questions in which the range of possible answers is known in advance (Oatey, 1999). Additionally, each respondent is offered the same set of possible responses (Mathers, et al., 2002). The advantages of this approach are the easily quantifiable answers and the ability to compare the responses. However, its main disadvantage is its lack of flexibility, due to the fixed nature of the questions (Oatey, 1999). As the research objective seeks to explore the consumer related factors associated with consumer response towards SSA, the interviewer needs some freedom to pursue or probe the interviewee responses; this opportunity would not be offered in the structured interview format. Consequently, the structured interview method was deemed inappropriate for the current research.

72

Finally, a semi-structured interview involves a series of open-ended questions in which the interviewer has the freedom to probe the interviewee, so that they can elaborate on their original responses or follow a different line of inquiry introduced by the interviewee (McMurray et al., 2004). The method is most useful when collecting attitudinal information on a large scale, or studying knowledge and assessing behaviours, as well as exploring the lifestyle and contextual issues of respondents (Gibson, 1998; McMurray et al., 2004), or when the research is exploring a new subject area and it is not possible to compile a list of possible pre-coded questions as little is known about the subject area (Mathers et al., 2002; McMurray et al., 2004; Oatey, 1999).

The current research will benefit from this approach of data collection. Its advantages will ensure sensitivity towards the participants‟ language, as well as allow a follow-up of specific ideas or issues. Such issues tend to emerge from the initial unstructured interviews during subsequent data collections (Fossey et al., 2002). As noted earlier, semi-structured interviews allow the researcher to obtain in-depth information through the process of probing and expanding on specific interviewee‟s responses (Partington, 2001).

73

Table 3.3 Comparison Interviews Comparison dimension

between

Unstructured,

Structured,

and

Semi-structured

Unstructured Interview

Semi-structured interviews

To obtain general information and an understanding of concepts and issues; can be used as a basis for further structured interviews.

To obtain a feedback and relevant information and explore an issue or a problem in more details

To obtain specific and descriptive information about a particular issue

Structure

Simple and informal. Lack of organisation

Systematic goal oriented process. Highly structured and pre-determined questions

Interviewer

Should be highly trained and expert

Control the flow of the interview as it needs. Structured and allowed open discussion Skilful and trained

Interviewer has few specific questions in mind and questions are mainly based on what the interviewee says

Open ended questions. Probe and follow up questions

Structured close-ended questions

The information may be general, vast and unrelated and it may be difficult to review, interpret and integrate

The interview guides and specifies the information to be collected. Most of the data collected is related and detailed

Most of the data collected is related, provides detailed information about the research problem

Main objective

Questions

Data collected

Structured Interviews

No need to be trained and experienced

Source: Adopted from Britten, 1995; Gibson, 1998; Hove and Anda, 2005; Patton, 1990; Leech, 2002. 3.6

Reliability and Validity of the Interviews

The achievement of reliability and validity in interviews is recognised as a major and integral part in any rigorous research methodology (Riege, 2003). While the result may differ because of individual‟s unique perceptions, and there is a difficulty in accurately calculating the measures statistically (Weaven, 2004), Reige (2003) suggests it is possible to establish valid and reliable findings for qualitative research. In particular, the type of interview conducted and the experience of the interviewer are the major determinants of the reliability and validity of the data collected via interviews (Seidman, 1998). In qualitative research, validity is concerned “…with the degree to which researchers‟ claims about knowledge correspond to the reality (or research participants‟ constructions of reality) being studied” (Cho and Trent, 2006, p.320). That is, validity is concerned with the accuracy of the measurement (Winter, 2000) in which the research results represent the phenomenon under study (Rich and Ginsburg, 1999) as well as the ability of the questions to measure what 74

they are supposed to measure (Graziano and Raulin, 1997; Winter, 2000). Thus, validity is considered one of the most important concepts that researchers can use to predict all the potential threats to the interview validity and design, and so ensure the control of the outcomes (Graziano and Raulin, 1997).

Thus, the following section focuses on how internal and external validity, as well as reliability, can enhance the quality of the interview design and process. The definitions for their validity tests are presented below. Firstly, internal validity refers to the ability of the research to establish cause and affect relationships (Zikmund, 2000); it is constructed by establishing phenomena in a credible way and highlighting the patterns of the main similarities and differences between respondents‟ opinions and beliefs in addition to specifying the type of mechanisms produced by them (Riege, 2003). In the current research, internal validity is maximised via sample selection on the basis of „information richness‟ (Patton, 1990, p.181). That is, Australian online users (who had searched online for products and services using Web search engines) were chosen in order to obtain useful and relevant information, thus adding to the quality of the responses and the internal validity of the qualitative process (Patton, 1990). Moreover, using a combination of qualitative and quantitative methods (such as the case in the current research) enhances the likelihood of producing internally valid research.

Secondly, external validity refers to the process of exploring the ability of generalising the research findings (Riege, 2003). This outcome is achieved in the current research by comparing the interview findings with evidence from the existing relevant literature. Indeed, interview findings corresponded to the findings identified by the participants in the current investigation.

Finally, reliability is related to the stability and consistency of the techniques adopted within the measuring concepts (Golafshani, 2003; Sekaran, 2003; Zikmund, 2000); it demonstrates the ability to replicate the research inquiry techniques and procedures by other researchers, thus, resulting in similar findings (Riege, 2003). The reliability of the findings from the qualitative phase of the current study was maximised through the use of several tactics. Firstly, a structured and operational process of semi-structured interviews was used including the recording, writing, and interpreting of the interview data. Thus, the techniques and procedures remained consistent with the researcher‟s theoretical positions being clarified 75

throughout the interviewing stage (Riege, 2003). As suggested by Guba and Lincoln (1994), a steering committee was used to assist in designing and administrating the interviews, namely, the researcher‟s supervisors (as steering committee) to assist in the cross-checking of the findings and providing feedback on the interview process and data integrity.

In brief, validity and reliability were achieved for the first phase in the current research through the use of consistent and verified research design and operational procedures. The following section discusses how the interviews were conducted and how their validity and reliability were maximised via careful planning throughout the interview processes.

3. 7

Planning and Implementation of Semi-Structured Interviewing Approach

This section discusses the process of implementing the semi-structured interviewing methodology followed by an exploration of the interview findings. In particular, the selection process of the interview subjects and the planning process of the interviews is discussed.

3.7.1 Identifying the Information Required The first step in the planning of interviews is to identify the nature of the information required (Dick, 1998), as well as to determine how much prior knowledge is required (Riege and Nair, 2004). Both help to provide insight into the planning process and determine the nature of the questions to be asked (Dick, 1998; Jarratt, 1996). The current research is at the cutting edge of this field of research; thus, limited scholarly guidance was available for planning the interviews. For this reason, the researcher reviewed the literature of the parent theories relating to Web advertising and consumer behaviour to develop a background knowledge of the research topic and to gain a deeper understanding of a situation (Nair and Riege, 1995; Sewell, 1998), while some areas of necessity are based on practitioner views. The practitioner literature provided a clearer picture of the research issues and guided the development of appropriate interview questions. Consequently, interview guidelines were developed that ensured a more focused and systematic interviewing process. After the first four interviews were completed, new issues and ideas raised by the interviewees enabled the expansion of the preliminary conceptual model presented in Chapter Two.

3.7.2 Identifying the Sampling Aspects The second step involved determining which participants would be interviewed to help generate the appropriate in-depth information to clarify the patterns and concepts attached to 76

the research objective (Thomson, 2004). Thus, the interview sample consisted of all Australian residents who had Internet access and who had searched online for products or services using Web search engines. As part of the preliminary conceptual model identifies the determinants of consumer attention towards SSA and the outcomes of such attention, including attitude towards SSA and intention to click on Sponsored Search Advertisements, it was essential that the sample was comprised of participants who were familiar with the Internet and who had searched online using a specific search engine.

In qualitative research, a sampling frame may be chosen by probability or non-probability approaches (Zikmund, 2000). Thus, samples tend to be relatively small and non-probability samples; whereas in quantitative research, larger random samples are more appropriate to enable the generalisability of the data (Kwortnik Jr, 2003) and, consequently maximising the external validity (Mays and Pope, 1995). In addition, the qualitative sample needs to be as heterogonous as possible and relevant to the issue being explored (Dick, 1990). Accordingly, the first phase sample of the current research was non-probability to identify heterogeneous groups who each possess common features relevant to the social issue being investigated (Dick, 1990; Mays and Pope, 1995). Thus, a purposeful rather than random sampling method is more appropriate (Patton, 1990). That is, to achieve heterogeneity, this research initially selected small and diverse sample of knowledgeable people who did not know each other; after each interview the interviewees were asked to recommend other people who may also be interviewed. Such a snowballing sampling technique was appropriate when research is concerned with small, specialised population of people who are knowledgeable about the topics (Aaker, Kumar, and Day, 2001; Neuman, 2000).

An important part of the sampling process was the selection of the first interviewee as the first snowball (Nair and Riege, 1995). That is, the first interviewee must be an expert as well as being able to direct the researcher to others who are familiar with the area of the research. Therefore, eight of the interviewees were undergraduate students at Griffith University, postgraduate students and staff at Griffith University with considerable experience with the Internet and online searching activities. These interviews represented eight different interviewees with different levels of Internet experiences and different degrees of experiences with their online searching and purchasing encounters.

77

According to Patton (1998), the appropriate sample size depends on what is to be derived from the interviews, as well as the researcher‟s available resources. Although there is no rule of thumb for calculating the sample size for qualitative interviews, most researchers recommend small samples (for example, Dick, 1990; Woodward, 1997). However, others argue that sample size is determined when data saturation is achieved (Naire and Reige, 1995). This saturation point occurs when no new data is obtained and the relationships among the categories are validated (Naire and Riege, 1995; Strauss and Corbin, 1998; Thomson, 2004). In the current research, sequential sampling was employed; participants were continuously interviewed until a point was reached where no new additional information was revealed. Eight completed exploratory interviews raised aspects not covered in the nascent literature. It was found after six interviews that no new information was being uncovered. This practice is supported by Guest, Bunce, and Johnson (2006) who found that saturation could be reached with as little as 12 interviews, although basic meta-themes were determined as early as six interviews. This supports that the interviewee sample size was sufficient in the current research.

3.7.3 Determine the Interview Process Several considerations and steps are involved in identifying or determining the setting for the interviews: the process of contacting the participants; the arrangement of the interview times and locations; the establishment of a rapport with respondents; the formulation of the opening and probing questions; and finally the approach to conclude the interview (Denzin and Lincoln, 2000; Kwortnik Jr, 2003; Dick, 1990).

Initial Contact with the Respondents. The researcher contacted the potential participants via email, inviting them to take part in the study. The initial email contact provided information about the researcher, an overview about the research, the objectives of the interview, the importance of their participation in the research, and an invitation to participate in the interview. Next followed a formal visit to the potential participants‟ workplace where a time and place for conducting the interview was determined after obtaining their acceptance of participation from each interviewee.

Time and Location of the Interview. Each interview was scheduled from approximately 30 minutes to one hour in duration; all face-to-face interviews were conducted at an assigned

78

office located at Griffith University, Gold Coast campus. Care was taken to ensure that the interview took place in an office that was quiet and guarantee interviewee privacy.

Establishing Rapport and Trust. Previous research emphasises the importance for interviewers to establish rapport and trust with the respondents (Denzin and Lincoln, 2000; Kwortnik Jr, 2003). For the current research, the initial rapport was developed by explaining several preliminary issues at the beginning of each interview (Carson et al., 2001), including informing the interviewees about the research; the purpose of the interview, to establish a person-to-person relationship (Dick, 1990); to assure interviewees of the preservation of their anonymity; and to inform them of the confidentiality of their responses. Permission was also sought from interviewees to tape record the interview.

Ask the Opening Question. Asking an appropriate opening question is vital in defining the nature of the study, as it allows more information to be given and to explore the respondent‟s attitudes towards the research topic (Dick, 1990). Thus, according to Nair and Riege (1995), the opening questions should be designed in such a way to provide a broad starting point that may provide insights for further probing questions. In the current research, the opening question was carefully designed to identify the research agenda, without limiting the extent of the potential responses. Consequently, a broad non-directive question, about the respondent‟s own experiences with online searching, became the opening question: “Could you please tell me about your experience with online searching in general?”

Probing Questions. Probing questions are designed to allow the researcher to elicit further detailed and descriptive responses, and to clarify respondents‟ particular points raised in the interview (Hove and Anda, 2005). Thus, probe questions were used in the current research, to help the respondents: express their ideas more clearly; elaborate and clarify issues; and focus the interview on the issues at hand, rather than diverging onto unrelated topics. Probing questions, such as „Can you tell me more about this issue?‟, „Can you explain more?‟ and the why? how? and what? of the issues (Yin, 1984) were used during the course of interviews.

Concluding the Interview. At the conclusion of the interview, when the researcher observed that little or no more information was being presented, the respondents were invited to give a summary of the key points they had raised (Dick, 1990). Then, they were thanked for their 79

participation in the interview process, and were asked if they would like a copy of the research findings when they become available. The method of collecting data for the first phase was followed by the analysis and interpretation of the semi-structured interviews, as discussed in the next section.

3.8

Data Interview Analysis and Interpretation

Several important factors need to be considered when analysing the interviews. The first factor is the procedure of recording and transcribing the interviews. Tape recording is most beneficial as it enables the interviewer to focus on the interview, so that the conversation flows, while taking note of the non-verbal interviewee gestures. Moreover, transcribing the interview assists the interviewer in becoming more familiar with the data and permits repeated examinations of the respondent‟s answers (Kwortnik Jr, 2003). In the current research, the analysis of the interview process followed a sequential process. Firstly, a summary of the key points of each interview was made by the researcher. Secondly, an evaluation of each taped interview then helped to clarify and modify the interview summary notes. Lastly, a verbatim transcript was produced for each interview.

The second factor involves the coding of the interview data, which allows the interviewer to make comparisons between the interviewee‟s responses. During the coding of the data, the researcher becomes familiar with the participant‟s perspectives through repetitive reading and the careful examination of the transcribed interview (Kwortnik Jr, 2003). A progressive data interpretation method was used in the current research to analyse the information given by each interviewee. The transcript of the first interview was compared with the following interview transcript (and so on), with a focus on the points of agreement and disagreement (Dick, 1990). The researcher was then able to modify the progressive report and form theoretical categories and relationships (Dick, 1990; Kwortnik Jr, 2003).

In qualitative research, the data analysis enables the understanding and exploration of categories, patterns, themes and meanings within, and through, the transcribed report of the interviews (Kwortnik Jr, 2003). In the current study, the researcher sought patterns from the compiled interviewee information allowing him to develop topic themes of particular interest. Codes from A to H were assigned to the respondents from the first phase of the study to ensure and guarantee respondents‟ confidentiality (as stated in the informed consent form).

80

Furthermore, these codes were vital in the data analysis process as they supported the theme discussed and highlighted evidence of emerging patterns.

The following section presents the findings of the qualitative interviews and discusses how these findings confirm and extend the preliminary conceptual model identified through the initial review of the extant literature (Chapter Two).

3.8.1 Findings from the Semi-Structured Interviews The process of managing and implementing semi-structured interviews was discussed previously in section 3.7. Here the key themes are identified from the semi-structured interview process. The information provided is considered to be appropriate due to its relevance to the stated research objective, and its value in revealing patterns, themes and concepts relating to the factors surrounding consumer‟s response towards SSA.

In describing the themes and patterns arising from the interviews data, a list of statements are summarised and categorised in Table 3.4, along with the research propositions. To ensure interviewee confidentiality, their names are replaced with the letters A to H.

81

Table 3.4

Summary of Results from Semi-Structured Interviews Themes / Comments

Interviewees A B

C D E F G H

1. Online searching is very important in making purchasing decision

















2. Starting searching point  Google  Yahoo  Others

√ x x

√ x x

√ x x

√ x x

√ x x

√ x x

x √ x

√ x x

3. The main reasons for using Google  Positive previous experience  Familiar with the name Google  The most famous and common search engine  Simple and well-designed  Most accurate

√ √ * √ √

√ √ * * √

√ √ √ * √

√ * * √ √

√ * * √ √

√ √ √ * √

* * * * *

√ √ √ √ √

4. Prior experience with SSA 

Prior positive experience



X √



x



x





Users tend always to check sponsored links as a result of previous positive experience with them.



X √



x



x





Users tend to ignore sponsored links as a result of previous negative experience with them. Users tend to view sponsored links as a good source of online information based on their previous experience with them.

x



x

x



x



x



X √



x



x





5. Subjective knowledge of SSA  Users are aware of the existence of SSA on the search results page.



















Users tend to pay more attention to SSA because they know about the nature and benefits of SSA.



X √



x



x





Users viewed SSA as not a beneficial tool or efficient source of information online

x



x

x



x



x

Users indicated that they tend to check SSA because they know that they are useful ads



X √



x



x











x















x









Users tend to pay more attention to SSA for familiar brands more than unfamiliar brands Users tend to avoid SSA for unfamiliar brands



















Users tend to like SSA for familiar brands



X √



x









6. Brand familiarity  Sponsored links for familiar brands are attractive. 

Notes: √ indicates interviewee agreement with the statement. X indicates interviewee disagreement with the statement. * indicates that question has not been raised. Source: Developed from the semi-structured interviews for this research

82

Table 3.4

Summary of Results from Semi-Structured Interviews Continued Themes / Comments

Interviewees A B C D E F G H

7. SSA Relevancy 

Users viewed SSA relevancy as one of the most important factors to influence their responses to SSA. Users tend to pay more attention to relevant sponsored links.

























x















x















x









SSA that is related to the user query is more favoured than irrelevant one. SSA should include relevant information to the keywords entered. SSA relevancy is judged based on the keywords entered.



















Users showed a tendency to ignore irrelevant SSA.



















Users exhibited a strong preference for SSA messages that are relevant.









x























x



x

x



x



x



X √



x



x





X √



x



x



x



x

x



x



x



X √



x



x



X √



x



* √

  

8. Credibility of SSA 

SSA credibility is mainly influenced by the nature of prior experience with such advertising. Users tend to perceive SSAs as not credible based on previous unsatisfactory experience. SSA is perceived as credible source of online information based on previous experiences. Users perceived credibility of SSA is related to their attention to SSA. Users tend not to pay attention to SSA as they perceive them as not credible and believable Users tend to view SSA positively when they perceive such advertising as credible and trustworthy.

    

9. Attitudes toward SSA 

SSA is positively viewed by users





SSA is negative viewed by users

* √ x



SSA is perceived as not intrusive ads



* √ √ * √ * √



Positive relationship between attention to SSA and attitudes toward SSA Having positive attitudes toward SSA, leads to higher intention to click on SSA.



X √



x



* √



X √



x



x



* √ * * x



Notes: √ indicates interviewee agreement with the statement. X indicates interviewee disagreement with the statement. * indicates that question has not been raised. Source: Developed from the semi-structured interviews for this research

83

3.8.2 SSA from the Consumer Behaviour Perspective - Preliminary Comments The framework developed in section 2.6 was derived from the relevant literature, most of which was based on the theoretical assumptions derived from previous Web advertising and consumer behaviour domains. The appropriateness of that framework to the Sponsored Search Advertising (SSA) realm was examined via a series of qualitative semi-structured interviews. Overall, the findings assisted in understanding the range of variables surrounding consumers‟ responses towards SSA. In particular, the interviewees provided information that was relevant and important to each of the theoretical propositions developed earlier, while also suggesting two additional areas of importance to be investigated. For example, SSA relevancy was viewed as an important factor leading consumers to pay attention to SSA. These findings have support from previous SSA research (Jansen and Spink, 2007a; Jansen and Resnick, 2006).

In addition, the respondents indicated that SSA credibility exerted an influence on their responses to such advertising. In particular, SSA credibility was influenced by the nature of prior experiences users had with SSA, and was positively related to attention towards Sponsored Search Advertisements. Moreover, SSA credibility was persistently found, during the course of interviews, to have an influence on attitudes towards SSA. These findings help to fill a gap in the available literature about SSA.

The propositions presented in section 2.6 are discussed below; a revised Model of Consumer Response Towards SSA is presented in Figure 3.2. Each research issues is addressed, having been derived from the preliminary theoretical framework within the context of the main findings emanating from the semi-structured interviews. Importantly, the information provided in this section represents the first empirical investigation of the issues included in the theoretical framework. Based on the analysis of the interview data, a number of patterns were found; they are included in the research conclusions and the implications (Chapter Six).

As mentioned in section 3.7.3, the first question asked of each interviewee was deliberately broad and general to develop and establish the initial rapport. Thus, interviewees were asked to report on their experiences with online information searching and, specifically, how long they have been using the Web and how they used it for searching and purchasing purposes. These questions were vital in obtaining a clearer picture of the importance of online searching as the facilitator for subsequent purchasing decisions. Additionally, they enabled the 84

interviewer to understand the relative importance of search engines in the consumers‟ online searching activities. All respondents agreed that: online search engines had become their first and major source of collecting and finding information about various products and services; and that online searching was essential in making purchasing decisions. For example, one interviewee (A) stated that online searching was vital in the collection of specific information to enable a comparison of products prior to purchase. Similarly, another respondent (B) indicated that “... the importance of online searching is in its flexibility. It allows people to search for information about the price and quality of the products they are interested in”, and thus, “... influences their purchasing decisions” (H).

All interviewees appear to use search engines as an initial and major source for conducting any online search activities. For example, one respondent (A) indicated that, “I generally go to Google first, but if I am looking for something that I have dealt with before, or if I purchased from some place before and I want to purchase something else, I might go directly to the site, but otherwise I would quite often go to Google”. Furthermore, the search engines were perceived as significant sources for acquiring information about different products and services before making any purchase decision. For instance, “… search engines are my best source for giving me information either to purchase product online or offline” (A). In particular, most respondents (with the exception of G) stated that they are mainly depending on Google to find online information or to browse the web. For instance, respondent D mentioned that “Google is always my first point-of-call when I need to quickly search stuff that I need to know”, and “I normally go to Google to find any online information” (B and C).

More specifically, most interviewees had beneficial experiences with Google as a primary source for finding online information. Firstly, their positive previous experiences were considered as the main reason for using Google to find online information. For instance, respondent A indicated that, “I only use Google because it‟s really successful for me … it has been for years. It provides me with the information that I was looking for and is pretty accurate whenever I do searches”. Another respondent commented that “… the main reason for using Google is because I am really happy with what Google has done for me so far…”(C). Secondly, most interviewees agreed that Google was the most accurate search engine, and that most of the time it presented relevant results. In particular, respondent D expressed that “… it‟s all about accuracy; the links are always better in Google than any 85

others (search) engine; the pages that come up first are all the ones that usually are the right ones”. Thirdly, many interviewees explained that they use Google because they are familiar with its brand, and being the most famous and common search engine. For example, interviewee F said that “… Google is well-known; everyone talks about it. It‟s famous and obviously works, just look at how they bought YouTube”. Finally, Google‟s success and attractiveness are attributed to its well designed and simple webpage (A, D, E and H). For instance, respondent D mentioned that “Google is the most well designed search engine of the moment…” Another respondent stated that “… Google is more user-friendly in the sense that the opening page is really convenient, not complicated, and simple with subcategories you can understand. No ads to start with, very empty… you have just to type your keywords and then the business begins…. Other search engines are full of text. They are all really cluttered and this reduces your time and mucks-up your results…” (E).

3.8.3 Determinants of Consumer Attention toward SSA and Subsequent Outcomes The interview results show that the areas examined in Table 3.4, provide a comprehensive list of the determinants of consumers‟ attention towards SSA and the subsequent outcomes, including attitude towards SSA and intention to click on SSA. The data were then used to test the validity of the propositions developed in section 2.6 and clarified and confirmed the appropriateness of the conceptual model, as shown in Figure 2.3. The supported propositions form the basis of the hypotheses presented in the following sections.

Prior Experience with SSA and Attention to SSA. The Interviewees were asked to describe their previous experiences with SSA and to identify the nature of those experiences. There were a variety of responses, with most interviewees (A, C, D, F and H) agreeing that they were satisfied with their experiences with SSA and, as such, they indicated that they would attend and check such advertising each and every time they conducted online search sessions. For example, interviewee A said that “… mostly, I started my search session with skimming the search results page… I always notice sponsored links [SSA] at the top and because it‟s been good to click on them in the past, I generally click on the link that closely corresponds to my search needs. But most of the time … I find what I am looking for by just clicking on the first three sponsored links [SSA] at the top of the page and then I finish searching”. Similarly, one respondent stated that “… what‟s good about sponsored links [SSA] is that you 86

go there and you find what you are looking for straight away without wasting heaps of time by going through all the links included in the page. I pay a lot of attention to sponsored links [SSA] because it‟s worked for me with so many products” (D).

Others respondents (B, E and G) stated that while they were aware of SSA, they tended to pay less attention (or ignore) to such advertisements on the search result page because of their prior negative experiences with such advertisements. For example, respondent E stated that, “I wouldn‟t pay much attention to those Google links [SSA] when I am searching online, because they always lead to expensive retailers … sometimes they also take me to a wrong page with products that are outside of what I am searching for. I just had to ignore them…” Similarly, “… I have clicked on sponsored links [SSA] many times, and always found that they just sent me to useless websites like searching for a laptop and going to an ISP sites… however, when I did click on free results, I was fairly satisfied with the outcomes… so sponsored links [SSA] need some improvements” (B).

Overall, interviewees confirmed that their prior experiences play an important role in the nature of their responses towards SSA. More specifically, a majority of interviewees attested that their prior positive experiences are positively related to their attention towards SSA. Indeed, paying attention to SSA is, in the main, subject to the nature of previous experiences with these advertisements. Thus, interviewees who possessed positive previous experiences with SSA tended to pay more attention to the advertisement, when conducting online search sessions, while those who were not satisfied with their experiences paid little or no attention to these advertisements. Given this level of qualitative support, it is hypothesised that:

H1:

Prior experience with SSA has a significant positive effect on attention to SSA.

Subjective Knowledge of SSA and Attention to SSA. As suggested in section 2.6.1.2, online users‟ subjective knowledge of SSA has an important impact on their responses towards SSA. Thus, all interviewees were asked about their awareness of SSA practises when using Web search engines. All were aware of SSA and indicated that they had noticed these advertisements when they were searching online using a particular Web search engine. For example, respondent A stated that, “I see sponsored links [SSA] every time I go to Google”. Further, the response levels and attention towards SSA messages were influenced by the amount of consumer subjective knowledge of such advertising. More specifically, 87

interviewees who exhibited higher tendencies to pay attention towards SSA perceived themselves to know more about the nature and benefits of these advertisements. The following statement is typical of most responses “… I think that getting more knowledge about what sponsored links [SSA] stand for, would make them more appealing and more reliable and I feel that most users would see them as more secure and pay more attention to these links”.

Similarly, respondent (D) stated that having knowledge about the nature of SSA (and the techniques employed by search engines to list these advertisements) assisted him in identifying the benefits and advantages of SSA over other online advertising forms; thus he perceived them as an attractive form of online advertising because he believed that they were personalised to his informational needs. “… that‟s why these links work for me when I am looking for more information about things I am looking to buy online, because companies have paid Google to be up there, so this really indicates to me that they have already been qualified as effective source for providing quick online information reliable and trustworthy and they have specific information to help my searches”(D).

In brief, the semi-structured interviews supported the proposition that there was a relationship between subjective knowledge of SSA and the amount of cognitive resources allocated in paying attention to SSA. Most interviewees emphasised that having (subjective) knowledge about the nature of SSA was necessary in achieving higher levels of attention to those advertisements. Based upon the above discussion, the following hypothesis is presented:

H2: Subjective knowledge of SSA has a significant positive effect on attention to SSA. Familiarity of Brands (or Websites) included in the SSA and Attention to SSA. The literature review revealed that one factor influencing consumers‟ attention towards advertisements was familiarity with brands. During the course of the interviews, all the respondents indicated that familiarity with a brand (or website) was vital in determining the subsequent attention they gave to SSA. Importantly, familiarity was perceived to increase the degree of attractiveness of SSA and, as such, that consumers would pay them greater attention in comparison to the advertisements for unfamiliar brands (or websites). For example, respondent F suggested that, “… I would be firstly looking for something familiar like a familiar brand, company or whatever”. Similarly, “I would say that brands that I have 88

experience with are important for me when I am searching for information online. If the sponsored link [SSA] has a name on it that I recognise then definitely it is going to grab my attention … it would look more legitimate…” (H).

Similarly, respondent D emphasised that the most important determinant of paying attention to SSA is familiarity of brands (or websites). He said, “... the first thing that attracts my attention when I am skimming through the search result page is to find brands or websites that I am aware of … for example, If I am looking for airline tickets, and the search engine displayed links for „Virginblue‟ or „Jetstar‟, I would definitely pay attention to those links that include those two companies because I know that these two airlines are economical when travelling within Australia, so, yes my eyes would firstly go there…” (D).

Furthermore, all the interviewees (with the exception of E) indicated that they would consider processing advertisements for brands or websites they had used before. For example, one interviewee A indicated that “… if I find a familiar brand or website mentioned in the message of the sponsored link [SSA], then I would read those links… but in [the] case of brand[s] or websites that I can‟t recognise or I haven‟t heard of before, I would simply ignore those links because they are including something that I am not used to…”. These findings led to support that familiarity of brands (or websites) included in the SSA has an influence on consumers‟ attention towards the advertisement. Thus, it is hypothesised that:

H3:

Familiarity of brands (or websites) included in the SSA has a significant positive effect on attention to SSA.

Sponsored Search Advertising Relevancy. During the course of the interviews, it became apparent that many online users believed that the relevancy of SSA is an important factor that influences their initial processing of such advertising. Overall, most of the interviewees indicated that the level of SSA relevancy impacted on their attentive processing of SSA. In particular, a strong majority of interviewees (A, B, C, D, F, G and H) were more likely to pay attention to SSA when these advertisements were relevant to their queries. This finding has a support in the marketing literature where it was found that online users tend to check Sponsored Search Advertisements if they are perceived as relevant (Jansen and Spink, 2007a; Jansen and Resnick, 2006).

89

Consequently, it appears that, if the content of SSA is relevant to the customer‟s situation (or is related to the subject matter they are seeking at that time) then Sponsored Search Advertisements are more favoured than other search results. That is, perceived relevant Sponsored Search Advertisements were more likely to capture consumers‟ attention and encourage them to read the associated information that may (potentially) result in clicking on the advertisement. For instance, one interviewee noted that, “... every now and then, if I see something that is of relevant to me or anything that I might be interested in or something like that I might click on it to see what it is”(F). Similarly, respondent A stated that, “I do pay attention to sponsored links [SSA] that are relevant to me… basically, because I have a specific purpose when I search online”. Indeed, according to respondent D, “… I know pretty much what I am looking for…anything that looks relevant and related…and the good thing about sponsored links [SSA] is that they give a quick snapshot because they are brief. So you can tell immediately if they are relevant or not…and for me, I look mostly for relevant sponsored links [SSA]… because I am very familiar with sponsored links [SSA] I usually find relevant information by only checking the first three sponsored links [SSA] I wouldn‟t even look at other links because I know if they are not there [SSA section], what I am looking for would not be anywhere else”.

Further, most of the respondents (with the exception of E) explained that it was important (for them) that the SSA include relevant information that is related to the used keywords (entered search terms). For instance, “…we are all bombarded with a lot of information on the Internet and you can find heaps of links within the search results page, so, I just look for links that are related to my query. I just look for links that included my keywords and those links that are specific to what I am searching for and generally I will check them” (D). Importantly, the respondents also indicated that their judgement of SSA relevancy was mostly related to the used keywords; finding the relevant links were evaluated in terms of finding the keywords included in Sponsored Search Advertisements. The respondents tended not to read the whole description within the advertisement; they would just skim over the search results for relevancy. For instance, “I usually don‟t read each link [SSA] closely and I am always skimming over what the link [SSA] is talking about… this gives me an indication if the link [SSA] is relevant or not. If it is relevant I would read it further and click through to the company‟s home page.”(A).

90

On the other hand, those Sponsored Search Advertisements, not perceived as relevant to respondents‟ specific needs, were seen as meaningless and so were more likely to be ignored. For example, one respondent (C) mentioned that each link on the search result page tended to be relevant to him because he could ignore most of the non-relevant links. Further, he explained that sometimes he also found irrelevant links on the search result page.

Additionally, it appears that the relevancy of SSA is related, to large extent, to the inclusion of familiar brands (or websites) within the SSA messages. All interviewees (except E) reported that finding familiar brands (or websites) within the content of the SSA influenced their level of relevancy. For example, one respondent H stated that, “... when I Google anything online, I used to check those sponsored links [SSA] that are relevant to me, and I used mostly not to check those sponsored links [SSA] that are irrelevant…the other day, I was looking for a computer and the top links [SSA] directed me to Dell, which is a top company I have bought from before, so it was relevant to me”. Similarly, “... when I go through the search page, I normally skim it, and if I find a sponsored link [SSA] for a website that I don‟t know, then, I would lose my interest in checking the link” (D). And, “Sometimes, if I can‟t find what I am looking for easily, I would be looking for those sponsored links [SSA] that seem relevant. Having said that, relevant to me means more specifically something that I can recognise or I have dealt with before”(C). Also, “I would probably look to the Website domain included in the sponsored link [SSA] and my immediate reaction when I see for instance „Toshiba‟, I would check this link because I know this company is good if you are looking for laptops, so, certainly I would be interested in this link” (F). In brief, the respondents‟ attentive processing could be greatly influenced by the SSA relevancy level. More specifically, paying attention to Sponsored Search Advertisements appears to depend on the relevancy within the context of the consumer search query. In turn, ignoring or skipping SSA appears to be higher for irrelevant advertisements, as perceived by consumers. Thus, relevancy appears to be an important antecedent to consumer attention towards SSA. Relevancy of SSA is also influenced by brand familiarity as finding familiar brands (or websites) within the content of SSA message may result in perceiving such advertising as relevant Therefore, an additional two hypotheses suggesting a relationship

91

between consumer attention to SSA as well as brand familiarity and SSA relevancy are presented: H4:

SSA relevancy has a significant positive effect on consumer attention towards SSA.

H5:

Familiarity of brands (or websites) included in the SSA has a significant positive effect on SSA relevancy.

Sponsored Search Advertising Credibility. During the course of the interviews, it became apparent that many respondents believed that the nature of their previous experiences with SSA influenced their perceptions of the credibility of SSA. Thus, a credibility theme emerged in relation to SSA when respondents were asked to express their opinions of SSA as an advertising tool. The respondents‟ perceptions of SSA were ranged from negative to positive based on the nature of their prior experiences with such advertising.

Three interviewees (B, E and G) indicated that SSA was neither beneficial nor credible, as the advertisements did not provide satisfactory results. For example, interviewee E suggested that SSA influences the reliability of the search engine because the search engines use money instead of phrase-relevancy as a standard for answering online users‟ queries. Another respondent B had a negative perception of SSA because the information delivered before or after clicking was not beneficial and frequently sent him to landing pages (websites) that were not related to his searches. More specifically, the same respondent B further explained that “I don‟t pay much attention to these links [SSA] and I don‟t click on them… because it wastes my time when I click on them [SSA] and go to a website where you can‟t find what you are looking for… if they want me to use them, then they need to work hard to change the way I think about these links [SSA]”.

Similarly, respondent E stated he discontinued using SSA because of problems he had experienced (with them) in the past. Specifically,” I have clicked on sponsored links [SSA] a few times, but after a while I have decided to stop paying attention to them because most of the time I was sent to sites that were misleading. You know, you ask for a specific hotel and you get sent to a third-party site waiting to charge too much. It‟s a rip-off…”. Two of those respondents who had negative experiences with SSA; have viewed such advertising as not providing adequate online information and as such they have chosen to pay less attention to (or ignore) SSA (due to their previous experiences with them). For instance, “I wouldn‟t say that I don‟t see them [SSA], I have trained myself just not to look at the sponsored links 92

section [SSA section] at all… I think now I don‟t notice them because I think they are not useful to me” (E).

On the other hand, the majority of the interviewees mentioned that previous clicking on Sponsored Search Advertisements has provided them with a positive experience and beneficial outcomes. More specifically, respondents A, C, D, F and H agreed that by clicking on Sponsored Search Advertisements, they have found more information about products, topics, and services that satisfied their needs or were consistent with what they were looking for. Also, having had pervious positive experiences with SSA, they were more likely to perceive SSA positively and to consider them a credible source of online information. For example, respondent C said that, “… most of the time, my experience with sponsored links [SSA] is positive. It‟s simple, I clicked on them and they took me to a good site with good information … that‟s a good thing… and that‟s why I would persist with using these links”. Therefore, an additional hypothesis, suggesting a positive relationship between prior SSA experience and credibility of SSA is proposed. In particular, it is hypothesised that:

H6:

Prior experience with SSA has a significant positive effect on SSA credibility.

The findings also revealed that attention towards SSA was closely related to the perceived credibility of such advertising. The majority of respondents (A, B, C, D, F, and H) tended to pay high attention to SSA because they perceived such advertising to be believable and trustworthy sources of online information. For example, one interviewee (A) reported, „ I would probably look straight away to sponsored links [SSA], simply because they seem to me like the official results in Google page that provide information quickly and immediately with no need to go through many links and pages”. Also, for respondent D, “… these links [SSA] work for me ... because they have already been qualified as effective sources for providing quick online information, reliable and trustworthy and they have specific information to help my searches”.

However, two respondents (B and E) did not often give attention to SSA, because they did not find them to have at any value. For example, respondent E observed that, “I don‟t think that I look that often to sponsored links section [SSA section], because for me, I don‟t remember that I have ever found anything by clicking on sponsored links [SSA]... for me, I don‟t trust them and I reckon they are just there to deceive me rather than guide me to what I 93

am looking for”. However, respondent E did suggest that he would be more likely to change his perceptions towards SSA, if these advertisements were more specifically directed towards the user‟s needs. “I reckon that if these paid-links [SSA] weren‟t just focused on „who could pay the money‟ it may be that I would use them more. They should focus on the user and not the advertising dollar…I trained myself not look at these sponsored links [SSA] anymore, I don‟t know ... maybe because they are so concise and simple which I don‟t like, and this does mean to me that they are not to be trusted”(E). These comments are consistent with other anecdotal evidence showing that SSA was ignored because online users viewed them as not a credible source of advertising (Marble, 2003). Such support also comes from the offline traditional advertising literature; Moore and Rodgers (2005) suggest that consumers are more likely to ignore advertisements because of less perceived credibility. Therefore, an additional hypothesis, highlighting a relationship between attention to SSA and perceived credibility of SSA is proposed. In particular, it is hypothesised that:

H7:

Attention to SSA has a significant positive effect on SSA credibility.

Indeed the credibility of SSA appears to strongly influence respondent‟s attitudes towards SSA. A majority of respondents referred to the credibility of SSA as an important factor that influenced their attitude toward SSA. Respondents A, C, D, F and H held positive attitudes towards these advertisements as they enhanced the efficiency and credibility of search engines by displaying search results that mostly corresponded to the interests of the respondent. Consequently, the respondents confirmed that their intention was to continue to use SSA because of the perceived usefulness and features of these forms of advertisements. For example, respondent A stated, “ I trust sponsored links [SSA] because it‟s been around for a long time and I have been using them for quite a while, and, I know that they are reliable and good... sometimes, I just use them without any need to go further for any other links in the Google page”.

In the same vein, respondents (C, D, and H) perceived SSA as attractive and believable and that they should be positioned on the search result page. For example, respondent D said that, “… some people would say that sponsored links [SSA]are about just making money but it‟s a business, and because we are looking for a service, they provide it to us and there is nothing wrong with that. I think that these links are great because I have nothing against people marketing their products and making people aware about what they have to offer. Google is a 94

good search engine that lots of people use and they advertise for what they do and what we are looking for; I don‟t think there is anything wrong with other companies doing the same. No, it‟s alright to me. On the contrary it‟s up there and because it‟s up there people will use it and you don‟t want to scroll down and go to page twelve; based on my experience, whatever is up there is most probably what I would be going for first… I consider them good and necessary and I will trust them as long as they are helping me get to where I want, quickly”.

Furthermore, many interviewees A, C, D, and H appeared to like SSA because such advertisements were perceived as unintrusive advertising within the context of Web advertising. For example, one respondent reported, “I know that sponsored links [SSA] are advertisements, fair enough, if I am looking to purchase a CD, than an advertisement might be a good thing because it might provide me with the information that I am looking for… [F]or me, sponsored links [SSA] are good and useful. I do not find them [SSA] intrusive at all because they are on the right hand side out of the page… people don‟t have to look at them if they don‟t want to” (A).

Overall, the interviewees had perceived SSA in different ways. While most interviewees reported that they did not hold preconceived (positive nor negative) attitudes towards SSA, two respondents (B and E) believed that such advertising was of little value; not surprisingly they held predominately negative attitudes towards SSA. This negative perceived credibility of SSA was the result of accumulated previous unsatisfactory experiences and outcomes. As noted by respondent B, “My experience: sponsored links [SSA] are not that … convincing; they just have too much information and it‟s very hard to get the right things from them...that‟s why I reckon they are not good”. From the above findings, it is hypothesised that: H8:

Credibility of SSA has a significant positive effect on attitudes towards SSA. Attitude towards Sponsored Search Advertising and Attention to SSA. The majority of interviewees agreed that the Sponsored Search Advertisements that were successful in procuring their attention, were those they most often viewed positively. Thus, the preliminary findings support that consumer‟s attention towards SSA is a determining factor of consumer‟s attitude towards SSA. For example, “I found most sponsored links [SSA] are good because they do generate some interest and often motivate me to click on them…” (D). Similarly, “I have seen a lot of sponsored links [SSA] that are advertising specific products that are so 95

related to the website that I am searching for… so, definitely I would say, sponsored links [SSA] successfully grab my attention. This means that they are good and I believe I will continue to use them as they provide good information and save me time” (A). Respondent C concurred with this view; “I have paid attention to and clicked on sponsored links [SSA] many times because they are interesting… I like them”. Given this level of qualitative support, it is hypothesised that:

H9:

Attention to SSA has a significant positive effect on attitude towards SSA.

Attitudes towards Sponsored Search Advertising and Intention to Click. The findings also confirm that a relationship exists between respondent‟s attitude towards SSA and their intention to click. Not unexpectedly, those interviewees who held positive attitudes towards SSA were more likely to click on such advertising as a result of perceiving this type of advertising to be a beneficial and useful online source of information. For example, one interviewee C commented that, “I like sponsored links [SSA], they take me right to where I want to be… I don‟t have to search through lots of links on the first result page … that‟s why I always click on them…”. And, “… sponsored links [SSA] are good I always check the links in order from the top… also I always focus on the sponsored links [SSA] because they are positioned under the search box which is obviously where I am looking”(H). Therefore, based on the preceding discussion, it is hypothesised that:

H10: Attitude toward SSA has a significant positive effect on intention to click on Sponsored Search Advertisements. Summary of the Interviews’ Findings. It appears that the eight semi-structured interviews represent the initial step in exploring consumers‟ responses toward SSA, from a consumer behaviour perspective. Overall, it was found that several factors act upon and influence consumers‟ attention to, subsequent attitudes towards, and clicking behaviour on SSA. These factors, identified as consumer related factors, incorporated prior experience with SSA, subjective knowledge of SSA, familiarity with brands (or websites) included in the SSA, and perceptions of SSA relevancy and credibility. These findings enabled the researcher to develop a revised theoretical model of Consumer Response Towards SSA (Figure 3.2) and hypotheses (Table 3.5) for the current study. Following on, Chapter Four focuses on the second phase of the quantitative data collection process which was designed to test the conceptual model of consumer response towards SSA and the developed hypotheses. 96

Table 3.5

Research Hypotheses

No.

Hypotheses

H1

Prior experience with SSA has a significant positive effect on attention to SSA.

H2

Subjective knowledge about SSA has a significant positive effect on attention to SSA.

H3

Familiarity of brands (or websites) included in the SSA has a significant positive effect on attention to SSA.

H4

SSA relevancy has a significant positive effect on attention to SSA.

H5

Familiarity of brands (or websites) included in the SSA has a significant positive effect on SSA relevancy.

H6

Prior experience with SSA has a significant positive effect on SSA credibility.

H7

Attention to SSA has a significant positive effect on SSA credibility.

H8

Credibility of SSA has a significant positive effect on attitude towards SSA.

H9

Attention to SSA has a significant positive effect on attitude towards SSA.

H10

Attitude toward SSA has a significant positive effect on intention to click on SSA.

Source: Developed for this research 3.9

Conclusion

Chapter Three presented an overview of the semi-structured interviewing method used in the first phase of the current study. The various research paradigms were described and a justification for selecting the realism paradigmatic approach as most suitable was presented. The research design was also outlined. This was followed by a discussion of the interviewing methodology and the reasons for using this method over other qualitative methodologies. The data collected from the semi-structured interviews was analysed within the context of the espoused relationships detailed in the literature review (Chapter Two). Finally, the findings of this phase of the research and a revised conceptual model of Consumer Response Towards SSA was presented.

97

Figure 3.2

Revised Conceptual Model of Consumer Response Towards Sponsored Search Advertising (SSA) Following the Semi-Structured Interviews

SSA Credibility Prior Experience with SSA

H6

H1

H8 H7

Subjective knowledge of SSA

Attention to SSA

H2

H3

Familiarity of brands (or websites)

H5

H9

Attitude towards SSA

H 10

Intention to click on SSA

H4

SSA Relevancy

Source: Developed for this research

98

CHAPTER FOUR RESEARCH DESIGN AND QUANTITATIVE METHODOLOGY FOR PHASE TWO 4.1

Introduction

In the current study, both qualitative and quantitative methodologies have been used to gain an insight into consumer response towards Sponsored Search Advertising (SSA). This chapter (Chapter Four) discusses the quantitative methodology used in the second phase of this study. It follows on from Chapter Three, which discussed the qualitative methodology which was used to inform the development of a set of the hypothesised relationships as well as the revised conceptual Model of Consumer Response Towards SSA (Figure 3.3).

This chapter is divided into six sections (Figure 4.1). Following the Chapter introduction, section 4.2 presents a justification for employing a survey methodology in the second phase of this research. The data collection method including the development of measures, questionnaire design and administration are detailed in section 4.3, while section 4.4 details the selection of the sample. Sections 4.5 and 4.6 respectively, describe the rationale and procedures involved in analysing the collected data, and the ethical considerations related to the data collection process. Finally, the chapter conclusion is presented in section 4.7.

99

Figure 4.1

Outline of Chapter Four 4.1 Introduction

4.2 Justification of Survey and Methodology for the Second Phase

4.3 Survey Design and Administration

4.4 Sampling Strategy

4.5 Anticipated Data Analysis

4.6 Ethical Considerations

4.7 Conclusion

Source: Developed for this research

4.2

Justification of Survey Methodology for the Second Phase of this Research

Surveys are among the most widely used data collection methods in social science research (Neuman, 2003) and especially empirical studies related to Web advertising and consumer 100

behaviour (Bell and Tang, 1998; Burns and Lutz, 2006; Gordon and Lima-Turner, 1997; Park and Kim, 2003; Patsioura, Vlachopoulou, and Manthou, 2009; Shamdasani et al., 2001; Shim et al., 2001). For the current study, the use of a survey was deemed appropriate because of its ability to collect quantitative data for further statistical analysis (Zikmund and Babin, 2007). Although experiments or laboratory studies have been used in some Web advertising studies, for example, to examine brand recall, memory, and clicking behaviour (Chang and Thorson, 2004; Danaher and Mullarkey, 2003; McCoy et al., 2007; Moore et al., 2005; Sundar and Kalyanaraman, 2004), the generalisation of these results may be questionable as they were conducted in controlled, and not „real life‟, settings. Further, surveys have additional advantages including the provision of quick, inexpensive, efficient and accurate quantitative data (Burns and Bush, 2003; Kumar, Aaker, and Day, 2002; Neuman, 2003; Zikmund and Babin, 2007). For these reasons, the survey method was chosen as the most appropriate approach to test the research hypotheses developed for the current study. Having provided a justification for using a survey method in this research, the following section details the process used to design and develop the survey to achieve precise and complete information about the research problem (Hair et al., 2008; Malhotra, 1999; Malhotra et al., 2004).

4.3

Survey Design and Development

The survey‟s design and development is a systematic process that includes a sequence of logical activities (Hair et al., 2008) that directly affects the quality of the collected data (Burns and Bush, 2003). Therefore, for the current study, several issues were considered throughout the survey process. Firstly, an appropriate set of concepts was developed and translated into a set of specific questions to elicit the information relating to the research objective and associated research questions (Lyberg et al., 1997; Malhotra et al., 2004). Secondly, the survey items were formulated to be easily understood by respondents and, therefore, ambiguous terms and complex statements were avoided (Lyberg et al., 1997). To improve the quality of the collected data and minimise response error, attention was also given to the questions format and wording, the instructions, and the logical order of the questions (and scales) (Hair et al., 2008; Malhotra et al., 2004). These issues are discussed in more detail throughout the chapter.

The survey instrument used in the current research was designed, developed and administrated according to guidelines presented in the existing marketing research literature (Churchill, 1999; Creswell, 2003; Hair et al., 2008; Malhotra et al., 2004). 101

Figure 4.2

Questionnaire Design and Development Process

Step 1

Specifying the data needed and operational definition

Step 2

Determining the survey method

Step 3

Determining the question content, wording and structure

Step 4

Determining the scale and response format

Step 5

Pretesting and revising the questionnaire

Step 6

Step 7

Assessing the reliability and validity of the survey

Administrating and placing the questionnaire online

Source: Adapted from Churchill, 1999; Creswell, 2003; Hair et al., 2008; Malhotra, 1999; Malhotra et al., 2004. As shown in Figure 4.2, Step 1 involved identifying the information needed to achieve the research objective as well as the operational definitions of the constructs being considered in the current study. In Step 2, the researcher chose the most appropriate survey technique by investigating the advantages and disadvantages of each available survey method. The issues relating to question content, wording and structure were then considered (Step 3), followed by Step 4 in which the response format and sequence of the questions were addressed. In Step 5, the questionnaire was pre-tested with a small sample of the target population to identify

102

and eliminate any problems. Then, in Step 6, the validity and reliability of the questionnaire was addressed. Finally, the questionnaire was administrated online to the target sample (Step 7).

4.3.1 Identify the Information Needed and Operationalisation of the Constructs The first step in the survey design process is concerned with determining the data needed to address the research objective and answer the research questions (Hair et al., 2008; Malhotra et al., 2004). Therefore, an important element of a good research design involves designing a questionnaire that addresses the needs of the research (Creswell, 2003), and serves as a tool to obtain the relevant information (Taylor-Powell, 1998). A review of the relevant literature in Chapter Two, and the findings of the exploratory phase of the data collection (Chapter Three), enabled the researcher to identify a number of relevant variables to answer the research questions (as shown in Table 4.1). However, prior to collecting the data for these variables, the conceptualisation and operationalisation of the research constructs was undertaken so that they could be measured (Sekaran, 2003).

Table 4.1

The Information Needed for this Research including Research Questions and Related Constructs

Research Questions

Investigated Constructs

RQ1: What are the consumer related factors that account for Prior experience with SSA, consumer attention towards SSA and to what extent do they influence subjective knowledge of attention towards SSA? SSA, SSA relevancy, and familiarity of brands (or websites) and attention to SSA RQ2: To what extent does attention to SSA influence attitude Attention to SSA and towards SSA? attitude towards SSA RQ3:

To what extent does SSA credibility influence attitude towards SSA credibility and attitude SSA? towards SSA

RQ4:

To what extent does attitude towards SSA influence intention Attitude towards SSA and to click on SSA? intention to click on SSA

Source: Developed for this research Operational Definitions of the Constructs The conceptualisation and operationalisation processes represent two basic elements that measure the focal constructs in the proposed theoretical model. Conceptualisation refers to the process of applying conceptual or theoretical definitions to constructs (Neuman, 2003; 103

Sekaran, 2000); that is, identifying what is meant by the construct. In the current research, consideration was given to provide clear and unambiguous definitions that link the theoretical framework to the research context (Neuman, 2003). These conceptual definitions are included in Table 4.2. Operationalisation is “… the process of transforming abstract constructs into a set of concrete indicators that can be observed and measured” (Smith, 1988, p.39). These specific operational definitions describe a concept in terms of its observable characteristics (Hair et al., 2008; Neuman, 2003). The current construct items were selected because of their alignment with the conceptual definitions of their constructs (Neuman, 2003).

The eight constructs are: prior experience with SSA, subjective knowledge of SSA, familiarity of brands (or websites) included in the SSA, SSA relevancy and SSA credibility, along with attention towards SSA, attitude towards SSA, and intention to click on Sponsored Search Advertisements. These variables, identified from an examination of offline and online advertising and consumer behaviour literature, were considered to be relevant to the current research objective, namely, „to explore the impact of consumer related factors on consumer response towards Sponsored Search Advertising‟. Additionally, the measurement scales for all of the focal constructs were sourced and/or adapted from existing scales as summarised in Table 4.2 and discussed in the following section.

Table 4.2 Construct

Prior experience with SSA

Subjective knowledge of SSA

Adaption of Existing Measures for the Focal Constructs Adapted Definitions

Reference works

Information learned and conclusion drawn from prior experience with SSA that functions both as an important feedback as well as determining how likely the consumer will engage with SSA in the future (Cho and Cheon, 2004).

Cho, C., and Cheon, H. J. (2004). Why do people avoid advertising on the Internet? Journal of Advertising, 33(4), 89-97.

Subjective knowledge of SSA is defined as the consumer‟s perception of the amount of knowledge of SSA they have stored in their memory (Flynn and Gold smith, 1999).

Flynn, L. R., and Goldsmith, R. E. (1999). A short, reliable measure of subjective knowledge. Journal of Business Research, 46, 57-66.

Items in original scale

11

5

104

Table 4.2

(Continued)

Construct

Familiarity of brands (or websites) in the SSA

SSA relevancy

SSA credibility

Attention towards SSA

Attitude towards SSA

Intention to click on SSA

Adapted Definitions

Reference works

Familiarity of brands (websites) B in the SSA refers to the degree of importance attached by consumers to familiar brands or websites included in the SSA advertising message (Simonin and Ruth, 1998).

Simonin, B. L., and Ruth, J. A. (1998). Is a company known by the company it keeps? Assessing the spillover effects of brand alliances on consumer brand attitudes. Journal of Marketing Research, 35(1), 30-42.

This construct is concerned with the degree to which the SSA and its keywords are perceived as pertinent, applicable and related to the consumer‟s search needs (Lastovicka, 1983).

Lastovicka, J. L. (1983). Convergent and discriminant validity of television commercial rating scales. Journal of Advertising, 12(2), 14-23.

SSA credibility represents online users‟ perceptions of the truthfulness and believability of Sponsored Search Advertising in general (Mackenzie and Lutz, 1989).

MacKenzie, S. B., & Lutz, R. J. (1989). An empirical examination of the structural antecedents of attitude toward the ad in an advertising pretesting context. Journal of Marketing, 53(2), 48-85,

The extent of cognitive resources a person devotes to the Sponsored Search advertisement (Muehling, Stoltman, and Grossbart, 1990; Laczniak and Muehling, 1993).

Muehling, D. D., Stoltman, J. J. and Grossbart, S. (1990). The impact of comparative advertising on levels of message involvement. Journal of Advertising, 19, 4, 41-50.

A learned predisposition to respond in a consistently favourable or unfavourable manner towards sponsored search advertisement during a particular exposure occasion (Pollay and Mittal, 1993). Intention to click on Sponsored Search Advertisements is defined as the likelihood that a consumer will engage in a clicking behaviour on Sponsored Search Advertisements (Shamdasani et al., 2001).

Laczniak, R. N., and Muehling, D. D. (1993). The relationship between experimental manipulations and tests of theory in an advertising involvement context. Journal of Advertising, 22 (3), 59-74. Pollay, R. W., and Mittal, B. (1993). Here's the Beef: Factors, determinants, and segments in consumer criticism of advertising. Journal of Marketing, 57 (3), 99-114.

Shamdasani, P. N., Stanaland, A. J. S., and Tan, J. (2001). Location, location, location: Insights for advertising placement on the Web. Journal of Advertising Research, 41(4), 7–21.

Items in original scale 3

6

3

5

4

1

Source: Developed for this research

105

Measuring Prior Experience with Sponsored Search Advertising. The items to measure prior experience with SSA were adapted from Cho and Cheon (2004) who designed a scale that assessed online user‟s prior experience with Web advertising. According to Cho and Cheon (2004), prior experience with Web advertising refers to information learned and conclusion drawn from prior experience with Web advertising which functions as important feedback, and determines how likely it is that the consumer will engage with the advertising in the future. Cho and Cheon‟s (2004) scale was adapted from earlier measures of prior negative experiences associated with advertisements by Oliver (1980) and Davis, Bagozzi, and Warshaw (1989). The prior experience with the advertisement scale comprises 11 items and is administrated in a Likert format, with seven responses, ranging from strongly disagree to strongly agree. Cho and Cheon‟s (2004) reported Cronbach‟s alpha as ranging from 0.75 to 0.99. As the original scale was applied to Web advertising generally, the items were adapted to relate specifically to SSA. Two examples of the original and adapted prior experience scale items are shown below in Table 4.3. Table 4.3

Examples of Prior Experience with Sponsored Search Advertising Scale Items from original scale

Modified items used in this study

I am dissatisfied with my decision to click Internet ads.

I am dissatisfied with my decision to click Sponsored Search Advertising.

I am not happy with my earlier decision to click Internet ads.

I am not happy with my earlier decision to click Sponsored Search Advertising.

Measuring Subjective Knowledge of Sponsored Search Advertising. The items to measure subjective knowledge of SSA were adapted from Flynn and Goldsmith‟s (1999) who developed a scale that assessed consumer subjective knowledge of fashion clothing. Subjective knowledge (of SSA) is defined as the consumer‟s perception of the amount of knowledge (of SSA) they have stored in their memory (Flynn and Goldsmith, 1999). Flynn and Goldsmith‟s (1999) five-item scale utilises a seven-point Likert scale, ranging from strongly disagree to strongly agree. The scale demonstrated a high reliability with alphas ranging from 0.87 to 0.93 (Flynn and Goldsmith, 1999). Therefore, these items were considered suitable for use in the current study, with minor adaptations being made to accommodate their use in the SSA context. Two examples of the original and adapted subjective knowledge items are shown below in Table 4.4.

106

Table 4.4

Examples of Subjective Knowledge of Sponsored Search Advertising Scale Items from original scale

Modified items used in this study

I know pretty much about fashion clothing among my circle of friends.

I now pretty much about Sponsored Search Advertising.

I am one of the experts on fashion clothing.

I am one of the experts on Sponsored Search Advertising.

Measuring Familiarity of Brands (or Websites) included in the SSA. The items to measure familiarity of brands (or websites) included in the SSA were adapted from Simonin and Ruth‟s (1998) three-item measure of brand familiarity. Familiarity of brands (or websites) in SSA represents the degree of importance consumers attach to familiar brands (or websites) in the message (Simonin and Ruth, 1998). The Simonin and Ruth‟s (1998) brand familiarity scale is a seven-point semantic differential scale. Simonin and Ruth (1998) reported alphas of .80 and .94 for the scales representing familiarity with the car and microprocessor brands, respectively. Therefore, these items were considered appropriate for use in the current research with adaptations to fit the SSA context. Two examples of the original and adapted familiarity of brands scale items are shown below in Table 4.5. Table 4.5

Examples of Familiarity of Brands (or Websites) Included in the SSA Scale Items from original scale

Modified items used in this study

The brand … is familiar

It is important that I am familiar with brands included in the SSA.

I recognise the brand

It is important that I recognise brands included in the SSA.

Measuring Sponsored Search Advertising Credibility. The items to measure SSA credibility were sourced from MacKenzie and Lutz‟s (1989) three-item measure of advertising credibility. SSA credibility represents online users‟ (consumers) perceptions of the general truthfulness and believability of SSA in general (Mackenzie and Lutz, 1989) (see Table 4.2). The advertising credibility scale, administrated on a seven-point semantic differential scale, has been used in the Web context by Choi and Rifon (2002), reporting Cronbach‟s alpha of 0.80. These items were deemed appropriate for use in the current

107

research, with the wording adapted to suit their use in the SSA context. Two original and current items of advertising credibility are shown below (Table 4.6). Table 4.6

Examples of Sponsored Search Advertising Credibility Scale Items from original scale

Modified items used in this study

Advertising is believable

SSA is believable

Advertising is convincing

SSA is convincing

Measuring Sponsored Search Advertising Relevancy. The items to measure SSA relevancy were sourced from the Lastovicka‟s (1983) 16-item scale to measure advertising relevance, confusion, and entertainment. However, only the relevance dimension was used for the current research. SSA relevancy is defined as the degree to which the advertising (SSA) and its keywords are perceived to be pertinent, applicable and related to the consumer‟s search needs (Lastovicka, 1983). The Lastovicka‟s (1983) advertising relevancy scale comprises six-items and is operationalised as a six-point Likert scale. Lastovicka (1983) reported Cronbach‟s alpha of 0.85, thus showing acceptable levels of reliability. In the current study only five items were used, as one item was deemed unsuitable in the Web advertising context. The item was pertinent only to television advertising; “I felt as though I was right there in the commercial experiencing the same thing”; this item could not be adapted to fit the Web advertising context. Minor adaptations were made to the items so that they related more specifically to the SSA context. Two examples of the original and adapted SSA relevancy items are shown below in Table 4.7.

Table 4.7

Examples of Sponsored Search Advertising Relevancy Scale Items from original scale

Modified items used in this study

The ad did not have anything to do with me or my needs.

The SSA did not have anything to do with me or my needs.

The commercial gave me a good idea.

The SSA gave me a good idea.

Measuring Attention towards Sponsored Search Advertising. The items to measure attention to SSA were adapted from existing measures by Laczniak et al. (1989) and Muehling et al. (1990) related to attention to an advertisement. Attention towards SSA is defined as the extent of cognitive resources a person devotes to the SSA (Muehling, Stoltman, and Grossbart, 1990; Laczniak and Muehling, 1993). The original scale 108

demonstrated acceptable reliability estimates (Laczniak et al., 1989; Laczniak and Muehling, 1993; Muehling et al., 1990). However, more recently the scale has been applied in the Web context by Bruner and Kumar (2000), and Stevenson, Bruner, and Kumar (2000), who reported alpha scores of 0.91 and 0.76 respectively. Minor adaptations were made to the original scale for use in the SSA context. Two examples of the original and modified items are presented in Table 4.8.

Table 4.8

Examples of Attention towards Sponsored Search Advertising Scale Items from original scale

Modified items used in this study

How much did you notice … the advertising

I noticed SSA.

How much thought did you put into evaluating …

I put significant thought into evaluating SSA.

Measuring Attitude towards Sponsored Search Advertising. Attitude towards SSA is defined (see Table 4.2) as a learned predisposition to respond in a consistently favourable or unfavourable manner towards SSA during a particular search session (Holbrook and Batra, 1987; Pollay and Mittal, 1993). The items to measure attitude towards SSA were sourced from Holbrook and Batra‟s (1987) measure of attitude towards specific advertisements which has four items and is administrated with a seven-point semantic differential scale. More recently, Shamdasani et al. (2001) used this scale in the Web context. Reliability estimates of 0.99 and 0.81 were reported by Holbrook and Batra (1987) and Shamdasani et al. (2001), respectively. Slight changes to the wording of the questions were made to reflect the SSA context. Examples of the original and adapted items are shown in Table 4.9.

Table 4.9

Examples of Attitude toward Sponsored Search Advertising Scale Items from original scale

Modified items used in this study

I like the advertisement.

I like SSA.

I react favourably to the advertisement.

I react favourably to SSA.

Measuring Intention to Click on Sponsored Search Advertisements. Intention to click on Sponsored Search Advertisements is defined as the likelihood that a consumer will engage in a clicking behaviour on Sponsored Search Advertisements (Shamdasani et al., 2001). The single item to measure intention to click on Sponsored Search Advertisement was sourced 109

from the intention to click on banner advertisement scale originally developed by Shamdasani et al. (2000). The intention to click scale has been used in both online advertising effectiveness studies and banner advertising research (Daugherty and Eastin, 2001; Rodgers, 2002; Yaveroglu and Donthu, 2008). Further, the scale is measured by asking respondents to provide a response, on a seven-point Likert scale to the following statement: I will be most likely to click on SSA to acquire further information. Following the conceptualisation and operationalisation of the focal constructs in the proposed model of the current study, attention was then given to determine the most appropriate survey method with which to collect the data (information that addresses the research objective).

4.3.2 Specifying the Survey Method Historically, survey data have been collected by a number of methods (self administered or personal face-to-face, telephone, observation or mail) (Burns and Bush, 2003; Churchill and Iacobucci, 2005; Hair et al., 2008; Malhotra et al., 2002; Sekaran, 2003; Zikmund and Babin, 2007). However, more recently, the use and application of Internet based surveys has been growing in the social science research area (Hanna et al., 2005; Malhotra et al., 2004). Each data collection method has advantages and disadvantages (Churchill and Iacobucci, 2005) and, therefore, the choice of the appropriate survey collection method is contingent upon the research objective, as well as the nature of the survey interaction and the mode of administration (Malhotra, 1999; Malhotra et al., 2004). A comparison of these survey methods, including the relative strengths and limitations of each method, is presented in Table 4.10.

110

Table 4.10

A comparison of Survey Methods

Dimensions

Online

Mail

Telephone

Personal

Cost

Low

Moderate

High

Speed of data collection

Very fast

Slow

Low to moderate Very fast

Geographic flexibility

High

High

High

Limited

Respondent cooperation

Low

Good

Excellent

Response rate

Varies depending on website Fair

Fair

Low

High

Anonymity of respondent

High

High

Moderate

Low

Follow up

Difficult

Easy

Easy

Difficult

Degree of interviewer influence Item non-response rate

None

None

Moderate

High

High

Low

Medium

Low

Moderate to fast

Source: Adapted for this research from Aaker, Kumar, and Day, 2005; Burns and Bush, 2003; Cavana et al., 2001; Churchill and Iacobucci, 2005; Hair et al., 2003; Malhotra, 1999; Malhotra et al., 2004; McDaniel and Gates, 2005; Neuman, 2002; Zikmund and Babin, 2007. A summary of the advantages and disadvantages of each survey method is presented in Table 4.10. The dimensions, such as cost, time, information control, sampling control, and the characteristics of the researcher and respondents need to be considered in choosing the most appropriate research method. Personal or self-administered questionnaires have advantages because of their ability to establish rapport with the respondents, enable immediate feedback and potentially provide high response rates. However, they also have disadvantages associated with the cost and the injection of researcher bias (Cavana et al., 2001; Churchill and Iacobucci, 2005; Kumar et al., 2002; Malhotra et al., 2004; Zikmund and Babin, 2007).

Telephone-based surveys have the advantages of being relatively quick to administer and of reducing the interviewer‟s bias; however, they are limited by the high costs and difficulties related to their inability to observe non-verbal responses (Cavana et al., 2001; Churchill and Iacobucci, 2005; Kumar et al., 2002; Malhotra et al., 2002; Neuman, 2003; Zikmund and Babin, 2007). In comparison, mail administrated surveys promise low costs, respondent anonymity, and wide geographic reach; however, their limitations include poor response rates and a reduction in the researcher‟s control during the data collection process (Churchill and Iacobucci, 2005; Cavana et al., 2001; Malhotra et al., 2002; Neuman, 2003).

111

Due to the growth in Internet usage (Aaker et al., 2005; Burns and Bush, 2003), online administered surveys are becoming a popular method in marketing research (Lazar and Preece, 1999; Malhorta et al., 2004). Online surveys offer many advantages over traditional survey methods, including greater response accuracy, rapid deployment, and fast turnaround (Burns and Bush, 2003; Hanna et al., 2005; Wright, 2005; Zikmund and Babin, 2007), with perhaps the most important advantage being their low cost (Sheehan and Hoy, 1999; Weible and Wallace, 1998; Wright, 2005). Indeed, the cost per response decreases significantly as sample size increases (Watt, 1999). However, the disadvantage of this approach is that online respondents may not to be representative of the general population (Cavan et al., 2002; McDaniel and Gates, 2005; Wright, 2005). For this reason, care was taken in selecting the most representative sample frame for the current research to counter this identified limitation.

After considering the advantages and disadvantages of the different survey methods, as well as the research objective and associated research questions, an online survey method was considered the most appropriate for the current study. Five reasons for this decision are outlined below:

An online survey method was considered appropriate as the population of interest consisted of consumers who have Internet access and who have searched for, and/or purchased products and services online using Web search engines;

Online surveys provide the facility to transfer survey responses directly into a database, thus, eliminating transcription errors and preventing survey alteration by the survey respondents (Burns and Bush, 2003; Hair et al., 2008);

In the current research, it was important to reach the geographically-dispersed population in Australia (Zikmund and Babin, 2007);

Online surveys provide a fast turnaround time, with relatively low cost, and ease of administration (Burns and Bush, 2003; McDaniel and Gates, 2005);

Most importantly, online surveys can utilise software programs to enhance the implementation of „branching‟ and „go to‟ questions electronically (Aaker et al., 2005; Zikmund and Babin, 2007). 112

In summary, an online survey was deemed the most appropriate method to reach the target market, because of rapid deployment, flexibility in questionnaire design, and the provision of computer data file results (Kumar et al., 2002; Malhotra et al., 2004). From specifying the appropriate data collection method, the next step is to determine the appropriate question design, for example, question wording, content and structure.

4.3.3 Determine the Question Wording, Content and Structure The guiding principles of good question design were adopted to collect accurate and relevant information, to minimise response error and, consequently to enhance the validity of the collected data (Burns and Bush, 2003; Cavana, et al., 2001; Churchill, 1999; Malhotra et al., 2004). Overall, issues such as the question content, wording, structure and sequence were considered (Cavana et al., 2002; Churchill and Iacobucci, 2005; Malhotra et al., 2004). In terms of the question content, the questions were designed to be brief, necessary, relevant, and understandable (Aaker et al., 2005); therefore, every effort being made to ensure they were complete and the meanings clear. Thus, the questions: only dealt with one topic; were not double-barrelled or ambiguous. Further, sensitive questions were kept to minimum and were placed towards the end of the questionnaire to reduce non-completion. The keywords and phrases were highlighted in bold and abbreviations were avoided to improve understanding. A range of structured questions were employed (scaled, multiple choice and dichotomous questions), with close-ended questions used with a specific ordered choice (Likert scale). The respondents were asked to choose the response that most closely corresponded to their opinion.

For the sequencing of the questions, simple and interesting opening questions were used to maximise respondent involvement; therefore, the questions were ordered from the most interesting (at the beginning) to the most sensitive (at the end). Importantly, the questions were easily categorised into groups according to the constructs and all sections were introduced by instructional statements. Those questions requiring personal information were placed at the end of the questionnaire, following accepted practice (Burns and Bush, 1995; Churchill and Iacobucci, 2005). The final survey was administrated online and, therefore, the guidelines suggested by Dillman (2000) were followed. As such, the presentation of the online survey was similar (and therefore familiar) to those used in traditional mail surveys in an attempt to maximise response rates (Ranchhod and Zhou, 2001). The online survey was designed with an open 113

format and vertical flow and instructions were given in order to ease the process of answering and completing the survey. Further discussion about how the online survey was designed for the current research is presented in section 4.3.8.

Through following the principles outlined above, the questionnaire was designed to maximise respondent comprehension, to increase response rates, to minimise measurement errors (in questionnaire design) and to enable the researcher to collect accurate and relevant data. As important as question design, is the process of choosing scale and response format (Davis and Cosenza, 1993) which is described in the following section.

4.3.4

Determine the Scale and Response Format

The process of scaling is an essential tool in almost every marketing research situation (Malhotra et al., 2004) and is most commonly used for measuring how people feel or think about objects or constructs (Neuman, 2003). Indeed, scaling is concerned with creating a continuum on which the measured objects are located (Kumar et al., 2002; Malhotra et al., 2004), and has two primary purposes. Firstly, scales facilitate the process of conceptualisation and operationalisation of the constructs and, secondly, they represent variables in quantitative measures to assist in testing the research hypotheses (Neuman, 2003; Zikmund and Babin, 2007). Thus, there are several scale characteristics that require consideration when identifying the most appropriate scale to be used in a study including: the form and structure of the scale in terms of labelling options, type of poles, balance of scales, and response wording (Kumar et al., 2002).

A number of scaling techniques are commonly employed in marketing research: the Likert scale, the Semantic Differential scale, the Thurstone scale, the Bogardus Social Distance scale, and the Guttman scale (Churchill and Iacobucci, 2005; Kumar et al., 2002; Malhotra et al., 2004; Neuman, 2003). However, the most commonly used marketing research scale is the Likert scale (Zikmund and Babin, 2007).

Likert Scale. Likert scales consist of a series of complete statements pertaining to the given object, which provide evaluative response categories ranging from strongly disagree to strongly agree (Kumar et al., 2002), with which the respondents are asked to indicate their level of agreement or disagreement (Kumar et al., 2002; Malhotra et al., 2004; Neuman, 2003; Zikmund and Babin, 2007). Likert scales are also called summated-rating or additive 114

scales because the scores of the individual statements are summed to produce a single score (Kumar et al., 2002; Neuman, 2003). Having taken into consideration the criteria for selecting a scaling technique, for example, information needed by the study and the characteristics of the respondents, as well as the mode of administration (Burns and Bush, 2003; Churchill and Iacobucci, 2005; Hair et al., 2008), the Likert scale was chosen for the current study. The scale was chosen for its simplicity, ease of construction, and administration via self-completion questionnaires, such as online surveys (Hair et al., 2008; Malhotra et al., 2002).

The next step in scale development (as advocated by Neuman, 2003) is to determine the appropriate number of response categories in each scale. The recommended number of Likert scale points is from three to nine; further, they should be evenly balanced with a neutral point (e.g., undecided), where respondents can express a neutral response direction (Aaker et al., 2001; Neuman, 2003). Hence, all the focal constructs, except „familiarity of brands (or websites)‟ were measured on a seven point Likert scale ranging from „Strongly disagree‟ to „Strongly agree‟ with „Undecided‟ as a neutral point response, to effectively allow respondents to express their opinions in this research. Familiarity of brands (or websites) included in the SSA was measured on a seven point Likert scale ranging from „Very unimportant‟ to „Very important‟. In addition, to achieve questionnaire uniformity and to ensure ease of completion, all scale and response formatting were standardised to a seven point Likert scale (O‟Cass and Pecotich, 2005).

Once the scale and response format were identified, the selection of the product or service to be used as a stimulus was addressed. “Travel tickets” (were adopted as the stimulus for the survey instrument); the rationale behind this decision is discussed in the following section.

Choice of Travel Tickets as Stimuli Travel tickets were chosen as stimuli for the current study because they are a popular online product and service category, as reported by DoubleClick in 2006. Further support for the choice of travel tickets as stimuli was also based on a pre-test conducted online (as discussed following in section 4.3.5) with a convenience sample of 30 postgraduate students at Griffith University. The pre-test survey asked respondents to list their purchases in the last 12 months, within the following nine categories: books, apparel, CDs/DVDs/ video games, electronic equipment, travel tickets, hotel/rental car reservations, personal care and household care 115

products, house ware and furniture, and movie tickets. These categories represented the most frequently searched and purchased products and services online (DoubleClick, 2006). The findings of the pre-test indicated that travel tickets, and hotel and car rental reservations were the most highly purchased product or service categories used by respondents. Additional support for „travel tickets‟ as stimuli was provided by interviews conducted with individuals (representative of the population of interest) and the results indicated that travel tickets were the most searched and purchased product or service category online in the cohort. Google AdWords (2008) also revealed that „Travel Tickets‟ were among the highest ranking keywords entered for online searching purposes in Australia. Specifically, in one year, the search volume for the keyword „Travel Tickets‟ reached one million (Google AdWords, 2008). Thus, „Travel Tickets‟ was selected as the stimuli for the current study.

Having determined the items to be included in the draft questionnaire, and chosen the scale response format, the question sequence and design, and the survey layout, the next step was the pre-testing of the survey instrument to ensure its validity, and to identify any changes needed prior to administering the final version (Burns and Bush, 2003).

4.3.5 Pre-testing and Revising the Questionnaire As mentioned above, an important step in the survey development process involves pretesting the instrument with a small sample of the target population to identify and eliminate any possible problems prior to administering the survey (Burns and Bush, 2003; Cavana et al., 2002; Malhotra et al., 2004). In the current study, the questionnaire was pre-tested in three stages. First, a hard copy of the questionnaire was evaluated by a panel of five academics in the Marketing Department at Griffith University and based on their feedback, some minor changes were made. Then, the questionnaire was assessed and reviewed by the researcher‟s supervisors to reach an agreement about the final version of the draft questionnaire.

Second, and guided by other survey researchers (for example, Churchill and Iacobucci, 2005; Neuman, 2003), the survey was pre-tested via focus group interviews with a pool of potential respondents. Three focus groups were conducted where the respondents were asked to complete an electronic copy of the survey instrument. Each focus group consisted of four online users who had searched for, or purchased, a product (or service) using Web search 116

engines. In this way, the researcher ensured that the participants were similar to the target population to be included in the actual survey (Malhotra et al., 2004). Initially, the respondents were briefed about the research, the purpose of the focus group and instructed on how to complete the survey. Once they had completed the online survey, they were asked to click on the submit button located at the end of the questionnaire. Participants were then requested to provide feedback on the following issues: the questionnaire design, the clarity of the instructions and questions, the question wording and phrasing, and any other comments relating to the survey instrument. Based on their feedback, several changes were made to the questionnaire. For example, the participants indicated that the questionnaire took too long to complete and that minor revisions and refinements were needed for clarity and comprehension of instructions as well as the wording of the items. Importantly, no items (the majority of which were sourced and adapted from existing measures) were deleted at this stage of the research.

Based on this pre-testing stage of focus groups, a number of areas in the survey instrument were identified for improvement, particularly in relation to clarifying and refining instructions and question wording, revising the length of the questionnaire, evaluating the questions flow and assessing the overall layout of the questionnaire. This process resulted in 38 items being retained in relation to the focal constructs; 15 classification questions were added to cover demographic, Web usage and search engine usage information. This stage of the pre-testing facilitated the development of the draft questionnaire in readiness for the pilot study.

Third, implementing the pilot study required the draft questionnaire to be administrated online to a small sample of the target population (Malhotra et al., 2004). The purpose of this stage of pilot testing is to measure the validity and reliability of the survey instrument. Therefore, the draft questionnaire was administrated to 39 respondents, selected on a convenience sampling basis. The collected data were then analysed using exploratory factor analysis and reliability analysis to assess its validity and reliability. Overall, the results demonstrated acceptable scores of reliability, ranging from .70 to .98. However, the factor analysis revealed that some items within the construct had low loading scores; consequently, some items were deleted (see Table 4.11).

117

Based on the previous three stages of pre-testing and pilot testing, several changes were made to the questionnaire relating to refining the wording of some items, deleting items and the survey layout. The final survey instrument comprised 36 items to measure the major constructs. The outcome of the pilot testing process is provided in Table 4.11.

Table 4.11

Changes Made to Draft Survey Items for Final Survey

Construct

Prior experience with SSA

Changes made

Items deleted2

Examples “No incentive is offered for the continued clicking on sponsored search advertisement”

Subjective knowledge

Wording changed-1

“ I know pretty much-to- I know a lot”

SSA credibility

Wording added -3 item added-1 Wording changed- 2

SSA is credible SSA is believable (to me) SSA is trustworthy “…did not have anything to do with me or my needsto- did not meet my needs ” “…was meaningful for me- to- SSA was relevant to me and the entered keywords”

SSA relevancy

Item deleted-1 Familiarity of brands (or websites)

Attention towards SSA

Attitude towards SSA Intention to click

Wording changed and added Item added-1

# of items in final survey 9

5

“As I was scanning…., I thought of reasons why would I buy or not buy the product”. (How important is it to you that SSA includes) familiar brands?

4

4

4

How important is it to you that SSA includes familiar Websites

Wording modified

“I noticed…-to- I notice..”

Item deleted-1 No changes

I was involved with SSA.

Item added-1

“I will be most likely to click on SSA located on the left side)”

4

4

2

4.3.6 Reliability and Validity Considerations The researcher gave consideration to the issues of validity and reliability throughout the measurement stage of the current study, including an assessment of the use of accurate and consistent measures. The following section outlines the validity and reliability strategies employed.

118

Reliability. Reliability is concerned with the consistency of responses (McMurray et al., 2004; Rich and Ginsburg, 1999); the research method demonstrates a consistent measure of performance and data (Palmer, 2002; Schraw, 1998) and is relatively free of measurement errors (Malhotra et al., 2004; Neuman, 2003). Thus, reliability is concerned with the degree to which similar findings are achieved each time the same study is replicated (McMurray et al., 2004; Riege, 2003). In the current study, reliability was maximised by using clear conceptualisation of the constructs and ensuring accurate measurements (Neuman, 2003; Riege, 2003). Further, all constructs were operationalised with multiple indicators. Finally, the questionnaire was pretested and modified to ensure that it was easily understood.

Validity. Validity is concerned with the accuracy of measurement (Winter, 2000), namely, which is the ability of the survey instrument to assess what it is supposed to measure (Graziano and Raulin, 1997; Winter, 2000), so that the research results represent the phenomenon studied (Rich and Ginsburg, 1999). The validity may be assessed in many ways; for example, in the current research, validity was assessed in terms of content, face, criterion, convergent and construct validity.

Content validity is concerned with the degree to which the content of the construct contains all the measures that should be included (McMurray et al., 2004) and it can be examined from a number of perspectives (face, content, convergent and construct validity) (Parasuraman, Grewal and Krishnan, 2004). Face validity is the degree to which the measurement reflects what is to be measured (Burns and Bush, 2003; Neuman, 2003). Interestingly, content validity is a particular type of face validity (Neuman, 2003) that represents the extent to which a measure represents all facets of a given construct (Burns and Bush, 2003; Neuman, 2003). Both face and content validity were achieved in the current study, as it followed several procedures recommended by Cooper and Shindler (1998), and Davis and Cosenza (1993). These procedures consisted of the appropriate examination of the existing literature, thus defining the conceptual dimensions, along with consulting panel of experts to assess the items within each construct. At that point, the survey instrument was developed and pilot tested on a small sample. Criterion validity “… examines whether the measurement scale performs as expected in relation to other variables selected as meaningful criteria” (Malhotra, 1999, p. 306). To strengthen the criterion validity, the current study substantiated the measures by deriving 119

information from online users during the qualitative stage of the research (detailed and discussed previously in Chapter Three). Specifically, the measures were linked to relevant Web advertising and consumer behaviour theories within the context of observations made in the qualitative stage of the current study (Cooper and Emory, 1995; Neuman, 2003; Zikmund and Babin, 2007).

Convergent validity occurs when multiple measures of the same construct operate in a similar way to form a single measure (McMurray et al., 2004; Otto, Najdawi, and Caron, 2000; Neuman, 2003). Within the current research, the convergent validity was maximised using multiple items to measure each construct; they are also further examined statistically in section 5.5.8. Construct validity is established when the empirical evidence of a measure is consistent with the theoretical implications underlying the concept (Malhotra et al., 2004; Zikmund and Babin, 2007), being achieved during the statistical data analysis stage (Chapter Five).

As noted above, a more detailed examination of the validity and reliability of the current research was conducted using statistical data analysis. While Chapter Five provides a more comprehensive assessment of the validity and reliability of the collected data, the initial examination provided some insights into the reliability and validity of the measures used. The next step in the survey design and development was the online administration of the questionnaire.

4.3.7 Administering the Questionnaire Online The online questionnaire was created with a survey generator application, Survey Monkey. As previously outlined in section 4.3.6, the current study followed guidelines for established and emerging online survey design (Andrews, Nonnecke, and Preece, 2003; Couper, Traugott, and Lamias, 2001; Dillman, 2000; Wright, 2005). Therefore, attention was given to designing an online survey that was user-friendly, and with features that promoted the logical flow of the questionnaire. Additionally, the inclusion of a motivational introductory screen, easy to follow instructions, the format, and the placement of questions and response boxes were continuously addressed (Dillman, 2000; Dillman and Bowker, 2001). The introductory screen which was the introduction to the survey included an informative title, information related to the research and researcher, and instructions on how to proceed to the next page and how to respond to the presented questions. As recommended by Couper et al. (2001), a multi-item 120

format (per screen) was used instead of only one question being shown per screen. The questionnaire was constructed to appear page by page, with each page containing a bank of questions. This approach assisted in reducing the completion time and lessened the possibility of missing data. It also facilitated the skip patterns for branching questions as the computer was programmed to automatically insert relevant questions on subsequent pages. The skip logic used two screening questions. For example, a filtering question was used where participants responded to a question about whether they had searched for and/or purchased a travel ticket online. Depending on their answer, respondents were either able to access all parts of the survey, or they were automatically sent to an end-page informing them that they did not meet the research requirements to continue with the survey.

Depending on the specific questions, text boxes, check boxes and radio buttons were used. The check boxes were applied to multiple choice questions, with instructions to click beside the responses. The radio buttons were used for questions scaled on a seven-point Likert scale which could be answered by placing a tick in the chosen button. Importantly, a progress indicator was included at the top of survey pages to reduce the possibility of respondents abandoning, and thus not completing all sections, in the questionnaire (Dillman and Bowker, 2001). In addition, each page ended with instructions to click a „Next Page‟ icon.

Once the questionnaire was completed, the respondents were provided with the contact details of the researcher, should they wish to receive a summary of the final report. In addition, respondents were asked to click on the submit button where a note of appreciation for their participation was presented at the close of the survey. After clicking on the submit button, the data were transferred automatically into the host company server, where it was saved. The database was then downloaded and exported into SPSS for further data analysis.

In Australia, the researcher used a specialised online data collection agency to assist with the data collection process. As a preliminary step, the researcher contacted the agency via email explaining the purpose of the research, the target sample, and the time limit for the completion of the data collection process. They outlined that the best way to collect the data as being via an invitation email to the target sample contacts on their existing databases. Further details of the process of administrating the online survey is discussed in the next section.

121

After deciding upon the administration process, the researcher must identify which subjects to survey in order to obtain the needed information to address the research objectives (Malhotra et al., 2004). Therefore, the sampling strategy employed in the current research is discussed next.

4.4

Sampling Strategy

When planning the sampling strategy, specific factors need to be considered by the researcher are: identifying the population to be surveyed, selecting the sampling frame, and the sampling technique and size (Churchill and Iacobucci, 2005; Kumar et al., 2002; Zikmund and Babin, 2007). Within the current study, the population of interest consisted of all Australian citizens who had access to the Internet in 2008. This population was then convergent with the research objective and, therefore, assisted in minimising the population specification error (Tull and Hawkins, 1993). According to estimates from Internet World Stats (2008), the Australian online (Internet users) population, as of December 2007 was 15,300,000. Indeed, according to Malhotra et al. (2004), information about the target population may be sought by specifying a sample of the population about which inferences may be made. In the current study, a representative sample was selected from which to draw conclusions that would be generalisable across the entire population being examined (Sekaran, 2003; Zikmund and Babin, 2007).

The second factor, the selection of the sampling frame; refers to the list of elements from which the sample may be drawn (Zikmund and Babin, 2007). In the current research, the sample frame was drawn from an Australian database provided by a commercial market research company (The Final Prospect). Of concern for any researcher, is the sample frame error which can occur when specific sample elements are not precisely represented in the sampling frame (Zikmund and Babin, 2007). However, every effort was made here to ensure that the frame was accurate and representative and, that the commercial market research company was aware of the researcher‟s requirements and had sufficient expertise in this area to ensure that the researcher directions were met. In addition, this commercial sampling frame was deemed most suitable for the current research because (1) it is the largest database marketing company in Australia (2) its ability to provide a detailed snapshot of Australian online consumer behaviour and (3) data is updated every three months to ensure data accuracy (The Final Prospect, 2008).

In addition, a total of two months was spent on

122

searching for a better sampling frame from universities and other databases marketing companies but this was the best one available in Australia, albeit an expensive one.

In the context of sampling procedures for online surveys, Forrest (2003) identifies three sampling techniques: unrestricted, screened and recruited samples. Unrestricted samples are concerned with obtaining a sample of everyone on the Internet who comes into contact with the survey. This sampling technique was not appropriate for the current research due to the potential limitation of a poor representation fit to the target sample (Forrest, 2003). Screened samples restrict the sample selection to only those participants who meet specific criteria, whereas recruited sampling uses email, or particular websites, to select and approach potential respondents seeking their participation in the research (Forrest, 2003). Therefore, taking into consideration the time constraints and the research objective, a combination of recruited and screened sample techniques were used in the current research. Specifically, the online sample for the current research was recruited from The Final Prospect Company database of 3000 Australian consumers who had access to the Internet and those who had previously conducted online purchasing online. More specifically, this particular database included Australian consumers who have signed up to an online panel to participate in online surveys conducted by The Final Prospect Company. That is, potential respondents were approached by posting an email invitation to 3000 Australian online users, requesting their participation in the study. An additional screened sampling approach was used for including only those participants who had searched for and/or purchased a travel ticket online, thus making them eligible to complete the questionnaire.

To direct participants to the Web survey link, The Final Prospect Company sent an email, on behalf of the researcher, to Australian online users inviting them to participate in the study. The email informed them of the purpose of the research and the length of the survey; it also assured them that the collected information would be treated confidentially and would be used only for the stated purpose of the research. In addition, a Web survey link was included in the email, to allowing respondents easy access to the questionnaire. In an endeavour to obtain a high response rate, The Final Prospect Company offered an incentive to respondents to complete the survey (a random draw for a $30,000 cash prize). The data collection for the study ceased on the 23rd November 2008 and any data collected after that date was not used in the subsequent data analysis. In total, 415 responses were received; 325 responses were usable, being a 13.9 percent response rate, and thus fitting within the acceptable range of 6 to 123

22 percent, a typical rate for online surveys in marketing field (Comley, 1997; Tse et al., 1995).

4.5

Anticipated Analysis

The anticipated data analysis to be used in the current research is outlined below. The summary statistics used in the data analysis are described briefly, followed by a preliminary analysis of the data, and a preview of the statistical procedure to be used to examine the theoretical model.

Firstly, the data analysis will begin with a summary of the descriptive statistics, such as frequencies, percentages and means related to the sample profile, in order to obtain a feel for the data (Sekaran, 2003). Secondly, it is anticipated that data analysis will include the measurement of central tendency and dispersion in addition to visual inspection of frequencies for all construct items. Thirdly, the preliminary analysis including the factor analysis will examine the factor structures of the constructs and reduce redundant items (Newman, 2003). By computing the construct reliability, the internal consistency of the measures will be examined. Then, an appropriate subsequent analysis will be performed to test the hypotheses developed in Chapter Three.

Given the nature of the relationships proposed in Chapter Three, the small sample size, and the research objective, Partial Least Squares (PLS) analysis technique was chosen to evaluate the revised theoretical model presented in Figure 3.3. PLS is a second generation of multivariate analysis technique (Barclay, Higgins and Thompsons, 1995), which was developed by Wold (1980) to analyse statistical models that involve a set of constructs and multiple indicators (Chin, 1998; Fornell and Bookstein, 1982). Further, PLS is considered the most appropriate approach for a study such as the current one. It has several advantages over other Structural Equation Modelling (SEM) techniques. PLS allows the assessment of the psychometric properties of the measures (Chin, 1998; Hulland, 1999), and also facilitates simultaneous tests of the measurement model and the structural model (Barclay et al., 1995). PLS is also compatible with interval-style data, and can assess a model with a relatively small sample size (Chin, 1998; Gefen, Straub, and Boudreau, 2000; Thompson, Barclay, and Higgins, 1995). In addition, PLS is an appropriate technique when the major concern is the prediction of the dependent variables (Chin, 1998; Fornell and Bookstein, 1982).

124

4.6

Ethical Considerations

The final section of this Chapter briefly describes how ethical conduct was applied to the process of data collection, and thus maintain the ethical standards expected (Freed-Taylor, 1994) to achieve moral research (Neuman, 2003) and correct decision-making (McMurray et al., 2004). Hence, the current research followed the ethical guidelines of The Research and Higher Degree Committee of Griffith University. Importantly, the study was granted ethical approval by the committee prior to the two stage data collection process being undertaken.

An ethical approach to the surveying was achieved by providing the respondents with an informed consent form, in the first phase of this research, which the respondents were asked to sign. In addition, an informed consent form was also provided for the online survey respondents. The detailed information on the form presented the participants‟ with the benefits, rights, risks and consequences of engaging in the study as well as outlining outlined the nature and the purpose of the research. Moreover, the respondents were informed of the voluntary nature of the survey and, therefore, were encouraged to respond by being offered an incentive to participate, and to ensure their continued cooperation. No identifiable information (for example, names and addresses) was requested so that the participants‟ personal privacy and anonymity were ensured.

4.7

Conclusion

This chapter has described the quantitative research methodology and survey design used in the current research. The justification for the choice of survey over other quantitative methods was presented. This was followed by a discussion of the steps adopted in the survey design and development, including the data collection method, measurement process and online questionnaire design and structure. The conceptual and operational definitions for all constructs identified in the proposed model were then described. In addition, an online survey was chosen as the most appropriate data collection method to obtain the required information and to test the set of hypotheses developed in Chapter Three. The next discussion detailed the sample selection procedures and the anticipated data analysis approaches. Finally, the ethical considerations relating to the data collection process were described. The next chapter (Chapter Five) will present the data analysis and results.

125

CHAPTER FIVE ANALYSIS AND RESULTS OF SURVEY DATA 5.1

Introduction

Chapter One provided an overview of the study, the topic under investigation and research issues. Chapter One also introduced the research objective and associated research questions, presented a justification for the study, and summarised the methodological approaches as well as outlined the content of the thesis. Next, Chapter Two provided a foundation or a theoretical framework for the study by reviewing the existing literature concerning Web advertising and consumer behaviour domains whereby a range of relevant constructs to the study were identified and discussed. This review resulted in the conceptualisation of a preliminary conceptual Model of Consumer Response Towards Sponsored Search Advertising (SSA) that served as a framework for the development of the propositions for this study. This was followed by Chapter Three which discussed the qualitative methods and presented the findings from the exploratory stage of the study. The findings of this first phase of the study served in the development of the hypotheses for the study as well as the development of the revised conceptual Model of Consumer Response Towards SSA. Following on, Chapter Four presented a justification of the quantitative research methodology of the second phase of the study used to collect data to test the hypotheses. Chapter Four also described the data collection method, sample selection procedure, and questionnaire development and design.

This chapter (Chapter Five) presents the results obtained from the online survey conducted (in the second phase of data collection). It has six sections. Following the Chapter introduction, section 5.2 discusses how the data were edited, coded, and transcribed in preparation for further analysis. Section 5.3 describes the sample profile. Next, section 5.4 discusses the data preparation conducted for this study which involves data screening for normality and outliers. Section 5.5 presents the results of the preliminary analysis including the evaluation of the data using factor analysis and reliability estimates. In section 5.6, the results of the hypotheses obtained through Partial Least Square (PLS) regression analysis are presented. Finally, section 5.7 provides a summary of the chapter. 126

Figure 5.1

Outline of Chapter Five

5.1 Introduction

5.2 Data Preparation

5.3 Sample Profiles

5.4 Preliminary Data Analysis

5.5 Preliminary Results

5.6 Data analysis Using Partial Least Squares (PLS)

5.7 Conclusion

Source: Developed for this research 127

5.2

Data Preparation

Data preparation is the process of transforming the collected data into a form suitable for data analysis (Malhotra et al., 2004; Zikmund and Babin, 2007). According to Kumar et al. (2002), the preliminary preparation of the data is essential to achieving good quality of data and a meaningful analysis. In the current research, editing, coding and transcribing of the data were used as a means for preparing the data for a subsequent analysis.

5.2.1

Editing the Data

Editing is the process of checking the completeness, consistency and legibility of data in readiness for coding and transference to storage (Malhotra et al., 2004; Zikmund et al., 2007). In the current research, the initial editing (for pilot study) was done before the commencement of the data collection process by assigning a code to each response; a pilot study was then conducted to check the accuracy and consistency of the coding process. Consequently, inconsistent coding was eliminated.

5.2.2

Data Coding

Coding the data refers to assigning a code to each possible response to each question in the survey (Malhotra et al., 2004). The current research used an online data collection method, and so the coding was done automatically, after the data collection process. Then the data were stored in a database downloaded directly into SPSS. One advantage of such online data collection (as discussed in section 4.3.2) is the automatic transfer and storage of the responses into the database. Thus, data entry errors are reduced (Zikmund et al., 2007).

5.2.3

Transcribing the Data

Transcribing the data involves transferring the coded data from the questionnaire into the computer software analysis (Malhotra et al., 2004). In the current research, the online questionnaire was linked to a database that recorded participants‟ responses; in turn these data were transferred into SPSS spreadsheet. It is noteworthy that there were no occurrences of missing data for the interval data, a consequence of undertaking checks for all sections to remind respondents to provide a response to individual items on the questionnaire. However, a small amount of missing data occurred for two demographic variables (education level and income) 128

where respondents were given the freedom to choose whether to provide responses to those questions. After preparing the data for analysis, the descriptive statistics relating to the respondents‟ profiles are discussed next.

5.3

Sample Profiles

Data were gathered from a convenience sample of 325 respondents via an online survey. The data relating to respondents‟ profiles were tabulated to obtain a better feel of the data, as recommended by Sekaran (2003). Therefore, the respondents‟ demographic profiles were tabulated for gender, age, education level, and income (Table 5.1). Then, as the sample comprised online users, the respondents‟ profiles regarding Web usage and search engine usage were tabulated, with the results presented in Table 5.2 and 5.3. Table 5.1

Demographic Profile of the Sample

Variable

Category

Response information N= 325

Gender

Male Female

Age

Range Mean

Education (optional)

Postgraduate Graduate Diploma Bachelor Advanced Diploma Yr 12 or equivalent Less than Yr 12 Missing

12% 17.9% 21.6% 15.4% 23.7% 9.3% 0.3%

Less than $25,000 $25,001 - $50,000 $50,001 - $75,000 $75,001 - $100,000 More than $100,000 Missing

12.6% 29.5% 32.9% 14.2% 10.2% 0.6%

Income (optional)

38% 61% 18 yrs to 78 yrs 39yrs

Demographic Profile of the Sample. As shown in Table 5.1, female respondents accounted for the majority of the sample (61%) and the age of respondents ranged from 18 to 78 years with an 129

average reported age of 39 years. In relation to the educational level, 33% of the respondents reported completing secondary education, while 67% reported achieving a university degree. For income, over 62% of the respondents reported an income of between $25, 001 and $75,000.

Web Usage Profile. To gain an insight into respondents‟ experiences with the Web, they were required to indicate: (1) how many hours they spend on the Web each week; and (2) how experienced they considered themselves with the Web. These responses are presented in Table 5.2 and are discussed next.

In terms of weekly usage of the Web, more than one third of the respondents (35%) used the Web for over 20 hours a week, while only 11% of the respondents used the Web for less than five hours a week (Table 5.2). These responses suggest that there is a high level of Web usage among the target sample. In relation to the respondents‟ experience with the Web, approximately 90% of the respondents expressed their opinion that they were either somewhat or very experienced with the Web (Table 5.2). Therefore, these results suggest that overall, the respondents could be considered to be experienced with the Web.

Table 5.2

Web Usage Profile

Variable

Category

Response information N= 325

using the

Less than 5hours/week 5-10 hours/week 11-15 hours/week 16-20 hours/week 21-25 hours/week 26-30 hours /week 31+ hours/week

11% 16% 21% 17% 18% 5% 12%

Do you consider yourself… with the Web

Very experienced Somewhat experienced Not at all experienced

35% 55% 10%

Frequency of Web

Search Engine Usage. The search engine usage profiles of the respondents showed the frequency of usage, the search engines most often used and searching techniques.

130

As shown in Table 5.3, the majority of the respondents (75%) used search engines daily; of these, 58% used them several times a day. The most often used search engine appears to be Google. The majority of respondents (83%) indicated that they use Google for online search activities. This result appears to be consistent with the number of Australian online users (85%) who depend on Google as the major source for online information (ITWire, 2006). Table 5.3

Search Engine Usage Profile

Variable

Category

Frequency of search engine usage

Once a day Several times/day Once a week Several times/week

Search engine most often used

Google Yahoo! MSN Others

Number of links being examined before clicking

Only one Only a few The first page The first two pages The first three pages More than three page

At what point do you Revise your search?

When you perform a search and don’t find what you are looking for, what are you typically more likely to do?

After reviewing the first few entries After reviewing the first page After reviewing the first 2 pages After reviewing the first 3 pages After reviewing more than 3 pages

Enter a few more words to better target the search Switch to another search engine and enter the same keywords Switch to another search engine and enter different keywords Give up

Response information N= 325 17% 58% 9% 16% 83% 7% 9% 1% 3% 23% 37% 14% 14% 9% 23% 33% 18% 18% 8%

85% 10% 1%

4%

In relation to respondent searching techniques, almost 63% of the respondents indicated that they only examined search results displayed within the first page before deciding to click on a 131

particular link. Further, 33% of the respondents indicated that they would revise their search after reviewing the first result page. However, only 8% of the respondents said that they would review their search after reviewing more than three pages. This result highlights the importance of being listed within the first result page; indeed existing search engine research shows that 75% of online users never look further than page one on search engine pages (Georgia Institute of Technology, 2006). In the current research, the majority of respondents (85%) also indicated that they were more likely to keep searching within the same search engine, by modifying the entered „used‟ keywords if they did not find what they were looking for within the search results pages.

5.4

Preliminary Data Analysis

A rationale and discussion of the preliminary data analysis used in the current research is presented below. The preliminary data analysis focused on evaluating the data via measures of central tendency and dispersion; correlation analysis, exploratory factor analysis and reliability estimates.

As mentioned previously, only a small amount of missing data were reported for two demographic variables (education level and income); there were no missing data in the interval data as the forced response approach was employed. According to Tabachinck and Fidell (2001), an important step in any preliminary multivariate analysis is the screening of the data for normality. Here normality refers to an assumption that each variable is normally distributed (Hair et al., 1998; Tabachinck and Fidell, 2001). Normality of variables is inspected by assessing skewness and kurtosis (Tabachinck and Fidell, 2001). In the current research, the analysis indicated that the data were multivariate normal as all the variables‟ values fell within the acceptable range of +2 and -2.

The data were also assessed for the presence of outliers, which can have an impact on the nature of the results (Hair et al., 1998). An outlier refers to “… a case with such an extreme value on one variable or such a strange combination of scores on two or more variables that they distort the statistics” (Tabachinck and Fidell, 2001, p. 66). According to Hair et al. (1998) any data that falls outside the range of three to four standard deviations should be identified as an outlier. In

132

the current research, an inspection of the data revealed that no outliers were found to fall outside the range of three to four standard deviations (see Appendix B).

The second step in the preliminary analysis involves examining the factor structures of the constructs and reliability estimates (which discussed in the following section). Factor analysis is one of the most widely used multivariate analysis technique in marketing and consumer behaviour research (Peterson, 2000) and is concerned mainly with reducing the set of data by “… expressing a large number of (observed) variables or indicators by means of a smaller set of linear composites, known variously as components, variates.… or most commonly, factors” (Peterson, 2000, p.262). In other words, factor analysis assists in bringing inter-correlated variables together under more general, underlying variables, as well as reducing the number of variables studied to a more limited number of underlying factors. Consequently, factor analysis provides a clearer understanding of the data and ensures that the data are appropriate for further analysis (Field, 2000).

Two types of factor analysis techniques have been identified and used in the marketing related literature, namely, Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) (Tabachinck and Fidell, 2001). EFA is used to identify and summarise data by simplifying interrelated measures, while CFA is used to verify the factor structure for a set of variables (Tabachinck and Fidell, 2001; Suhr, 2000). Given that EFA is a widely utilised and applied statistical technique in the social sciences (Costello and Osborne, 2005), EFA was considered as a suitable technique to be used in the current research. More specifically, EFA was used because of its advantages in enhancing the understanding of the measurement quality, and determining and simplifying the factor structure for the set of observed variables, based on a correlation matrix (Cheng and Shiu, 2008; Netemeyer, Bearden, and Sharma, 2003; Shi and Wright, 2000).

In evaluating the adequacy of the collected data for factor analysis, several points were taken into consideration. Firstly, an examination of the correlation matrix was conducted to inspect any items with low correlations (Tabachinck and Fidell, 2001). This approach was important as Tabachinck and Fidell (2000) suggest that the use of factor analysis needs to be reconsidered if the correlations are less than .30 within the correlation matrix. Another approach to identify the 133

appropriateness of the collected data for factor analysis is an examination of the entire correlation matrix by testing the presence of correlations among the variables (Hair et al., 1998). Inter-correlations among variable may be tested via statistical measures, such as the Bartlett test of sphericity and the Kaiser‟s measure of sampling adequacy (KMO). The former examines whether the correlation matrix is an identity matrix, whereas the latter tests whether the partial correlations among the variables are small (Hair et al., 1998). A significant value of Bartlett test (