The Role of the Audience in Product Placement: Development of an Audience Engrossment Scale

A thesis submitted in partial fulfilment of the requirements for the degree of Doctor of Philosophy

from

The School of Marketing Australian School of Business University of New South Wales

by

Jane Scott

© August 2008

 O              P     !  " # $% &  '    (( ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

______

Jane Margaret Scott 25th August 2008

 O              P     !  " # $% &  '  (( COPYRIGHT STATEMENT

‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

Signed ……………………………………………......

Date ……………………………………………......

AUTHENTICITY STATEMENT

‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’

Signed ……………………………………………......

Date ……………………………………………......  O              P     !  " # $% &  '    (( ABSTRACT

Product placement is now a US$7.76 billion industry, flourishing as advertisers attempt to combat audience sophistication, zipping, zapping, muting of commercials, TiVo, media multi-tasking, the Internet and digital television, all of which may signal the death knell of the interruptive commercial model. Yet whilst research on product placement is growing, it has not kept pace with the practice, and many findings do not converge across studies. This is likely the case because parameters remain undefined and there is no operational framework to describe how product placements are processed, and no agreement as to what effects are possible or how they should be examined. Most effects-based research has focussed on executional factors and what the product placement does to the audience member. This assumes that the recipient is a passive participant. However this thesis argues that the audience member is actually an active processor who should be the focus of research.

This research distinguishes product placement from related activities and develops a new conceptual model of product placement processing. It puts a strong focus on the role of the audience member, stating that their level of familiarity of the placed brands, and their level of engrossment with the entertainment story will impact their recognition of product placements in that story. Applying Rasch Measurement Theory, an Audience Engrossment scale is developed and refined over four stages of data collection, with 1360 respondents across seven films, to capture the quality of people’s interaction with a film. The result is a scale comprising 19 feeling items, 10 arousal items, 6 appraisal items and 7 cognitive effort items. The scale was then tested as part of the conceptual model, with 191 participants watching The Island and completing questionnaires after the film relating to their recognition of brands within the film and their level of engrossment. Brand familiarity information was collected four weeks earlier. Onset prominence, high plot connection, dual modality and use by star were found to have the strongest direct effects on recognition, with brand familiarity and the four audience engrossment dimensions generally found to interact with the product placement characteristics as hypothesised.

 O              P     !  " # $% &  '  (( DEDICATION



                            



 O              P     !  " # $% &  '  (( ACKNOWLEDGEMENTS

I have a lot of people to thank for getting me this far. The fact that I have such a long list is testament to the quality of people I am fortunate enough to have in my life. To all these people, thank you for getting me through the tough times, but thanks also for sticking around to share the good times too!  Margaret Craig-Lees Thank you for teaching me the fundamentals in my Honours degree, and for continuing this guidance throughout my PhD. Thank you for being there through all my losses and for understanding my pain and being a ready ear. Thank you for inviting me and Dad into your family, and for letting me take time off when I needed it and being so supportive of my time at London Business School. Thank you for making my brain hurt more times than I wish you had. Thank you for your brilliant mind and your much valued friendship.  Jennifer Harris Thank you for coming on board when Margaret moved to Auckland. Thank you for being the stats guru, the calming influence, the patient teacher, and so very quick and efficient. Thank you for your hands-on approach to helping me with my data collection and for having the Midas touch for getting cinemas and schools on board. I feel truly blessed to have had the two loveliest supervisors in the history of supervisors, and for the special friendship and mutual respect we all share.  Thomas Salzberger Thank you for answering an email from a random Australian PhD student. Thank you even more for answering a thousand more emails since. You are the most generous person with their knowledge and their time that I have met on this journey, and you have shared this with me so willingly and graciously that I will forever be indebted to you. I hope your mission to spread the Rasch message continues with gusto, and that my thesis may help you do this. Come back to Australia whenever you want – you will always have a friend in Sydney.  O              P     !  " # $% &  '  (( I would also like to thank some other UNSW colleagues whose friendship and encouragement has been invaluable, namely Marion Burford, Ian Wilkinson, Mark Uncles, Roger March, Mathew Chylinski and Tini Mathies. Thank you also to the wonderful Margot DeCelis, Paula Aldwell and Nadia Withers for their support.

Closer to home, I am also blessed by the love of two wonderful men.

Dad My best friend. Thank you for giving me perspective and being unfailingly proud of me even when I doubted myself or wanted to give up. Thank you for being the greatest research assistant in the world and helping me with my data collection, giving up countless days to convince strangers at the cinema to help your daughter with her PhD. The silver lining in the last four years has been the consolidation of a father-daughter relationship that is the envy of everyone I know. I love you.

Christian Thank you for your love, which came at the most unexpected time, but which has made me the happiest girl in the world, and been the perfect counter-balance and distraction from all this study. I mean it when I say that I could not have got through all of this without you.

Finally, thank you to all of my other friends (especially those who assisted with my data collection) who have been an enormous support and tower of strength to me, as well as a bundle of fun. Your love, faith and pride in me have given me the strength to give of my best, especially when it all seemed too hard.

 O              P     !  " # $% &  '    (( PUBLICATIONS DERIVED FROM THIS RESEARCH

Scott, J. and M. Craig-Lees (2005a), "Measuring Media Audiences: The Need for an Audience Engrossment Scale", Proceedings of ANZMAC 2005, Perth, Australia.

Introduction of audience engrossment concept; inadequacy of existing scales; components of audience engrossment (i.e. Sections 3.6.2, 3.7, 3.8)

Scott, J. and M. Craig-Lees (2005b), “Product Placement: Developing Concepts, Constructs and Measures”, Advances in Consumer Research, vol. 33. Cornelia Pechmann and Linda Price (eds.) (also presented at ACR Conference, Texas, 2005).

Brandcasting continuum; relationship between product placement and advertainment; distinguishing product placement from sponsorship and endorsement (i.e. Sections 2.7.1, 2 .7.2)

Scott, J. and M. Craig-Lees (2006), "Conceptualisation, Consumer and Cognition: The 3 Cs that will advance product placement research”; Advances in Consumer Research Asia Pacific, vol. 7. Margaret Craig-Lees, Gary Gregory and Teresa Davis (eds.) (also presented at ACR Conference, Sydney, 2006).

Introduction of brandcast processing model, need for stronger distinction between explicit and implicit memory tests; extension of re-conceptualisation issues (i.e. Section 2.7, Chapter 3)

Scott, J., J. Harris, M. Craig-Lees (2007) – “Refining Audience Engrossment: A Comparison of the Usefulness of Rasch Modelling and Factor Analysis in Scale Development”; Proceedings of ANZMAC 2007, Dunedin, New Zealand.

Justification of Rasch measurement as opposed to Classical Test Theory in scale development; presentation of results (i.e. Sections 4.5, 4.6, 4.9, Chapter 5)

Scott, J. and T. Salzberger (2008) – “Investigating the Threshold Ordering of the Audience Engrossment Scale Using the Polytomous Rasch Model”; 3rd International Rasch Conference, Perth, Australia

Importance of ordered thresholds; demonstration of modified response scale categories (post-hoc versus new data collected); presentation of results (i.e. Section 4.9, Chapter 5, Appendix 4.2)

 O              P     !  " # $% &  '    (( TABLE OF CONTENTS

ABSTRACT...... IV

DEDICATION...... V

ACKNOWLEDGEMENTS...... VI

PUBLICATIONS DERIVED FROM THIS RESEARCH ...... VIII

CHAPTER 1: INTRODUCTION ...... 1

1.1 – INTRODUCTION ...... 1 1.2 – THE GROWTH OF PRODUCT PLACEMENT ...... 4 1.3 – THE RESEARCH PROBLEM ...... 7 1.4 – RESEARCH OBJECTIVE AND QUESTIONS ...... 10 1.5 – ISSUES INFORMING THE METHODOLOGY ...... 11 1.6 - METHOD...... 12 1.7 – CONTRIBUTION AND KEY FINDINGS OF THIS RESEARCH...... 13 1.7.1 – Theoretical and Methodological Contribution and Key Findings...... 13 1.7.2 – Social Contribution and Key Findings ...... 14 1.7.3 – Managerial Contribution and Key Findings...... 15 1.8 – THESIS OUTLINE...... 16

CHAPTER 2: PRODUCT PLACEMENT LITERATURE REVIEW ...... 19

2.1 – INTRODUCTION ...... 19 2.2 – THE ACCOUNTABILITY ISSUE ...... 19 2.3 – OVERVIEW OF PAST RESEARCH INTO PRODUCT PLACEMENT ...... 21 2.4 – INVESTIGATING PRACTITIONER ATTITUDES AND PRACTICES REGARDING PRODUCT PLACEMENT...... 22 2.5 - UNDERSTANDING AUDIENCE ATTITUDES AND PERCEPTIONS ABOUT PRODUCT PLACEMENT ...... 23 2.6 – PAST RESEARCH MEASURING PRODUCT PLACEMENT EFFECTIVENESS ...... 26 2.7 – EVALUATING PAST RESEARCH AND MOVING FORWARD ...... 30 2.7.1 – Conceptualising Product Placement...... 31 2.7.2 - Distinguishing product placement from other activities...... 34 2.7.3 - Measuring effectiveness...... 40 2.8 – RESEARCH GAPS...... 43 2.9 – CONCLUSION ...... 45

 O              P     !  " # $% &  '  ()(

CHAPTER 3: ADVANCING PRODUCT PLACEMENT RESEARCH ...... 46

3.1 - INTRODUCTION...... 46 3.2 – THE NEED FOR A BETTER CONCEPTUAL MODEL ...... 46 3.3 – DEVELOPING A MODEL OF BRANDCAST PROCESSING...... 49 3.4 – MEASURING EFFECTS ...... 51 3.5 – BRANDCAST QUALITY ...... 52 3.5.1 - Modality ...... 53 3.5.2 - Prominence...... 53 3.5.3 – Star presence and star use...... 54 3.5.4 – Method of Depiction ...... 55 3.5.5 – Temporal Quality...... 56 3.5.6 – Plot Connection ...... 56 3.6 – AUDIENCE CHARACTERISTICS ...... 57 3.6.1 - Level of Product / Brand Familiarity ...... 58 3.6.2 – Level of Audience Engrossment...... 59 3.7 - INADEQUACY OF EXISTING CONCEPTS AND MEASURES WHICH COULD APPLY TO AUDIENCE ENGROSSMENT...... 61 3.8 – COMPONENTS OF AUDIENCE ENGROSSMENT ...... 67 3.9 – CONCLUSION ...... 71

CHAPTER 4: DEVELOPMENT OF THE AUDIENCE ENGROSSMENT SCALE...... 72

4.1 - INTRODUCTION...... 72 4.2 – THE RESEARCH SETTING ...... 72 4.3 – UNDERLYING RESEARCH TRADITION AND STRATEGY ...... 73 4.4 – THEORY OF MEASUREMENT ...... 76 4.5 – JUSTIFICATION FOR RASCH MODELLING FOR SCALE DEVELOPMENT ...... 77 4.5.1 - What is the Rasch model?...... 78 4.5.2 - Advantages of the Rasch Measurement approach over the Classical Test Theory approach ...... 80 4.6 – ITEM GENERATION ...... 84 4.6.1 – Specifying the domain of engrossment...... 84 4.6.2 – Generating items...... 84 4.6.3 – Applying Rasch Measurement Theory to item development ...... 86 4.7 – INITIAL AUDIENCE ENGROSSMENT SCALE...... 92 4.8 – INITIAL CONTENT VALIDITY CHECKS ...... 97

 O              P     !  " # $% &  '  ()( 4.9 – REFINEMENT #1 – PILOT STUDY – NOVEMBER 2006 ...... 97 4.9.1 – Sample and Implementation...... 97 4.9.2 – Use of RUMM2020 software ...... 99 4.9.3 – Snapshot of AUDENG scale tested at Stage 1 ...... 100 4.9.4 – Results for Feelings Dimension ...... 100 4.9.5 – Results for Arousal Dimension ...... 102 4.9.6 – Results for Appraisal Dimension ...... 102 4.9.7 – Results for Cognitive Effort Dimension ...... 104 4.9.8 – Modifications to the scale before Stage 2...... 104 4.10 – REVISED AUDIENCE ENGROSSMENT SCALE...... 107 4.11 – CONCLUSION ...... 110

CHAPTER 5: REFINEMENT OF THE AUDIENCE ENGROSSMENT SCALE...... 111

5.1 - INTRODUCTION...... 111 5.2 – REFINEMENT #2 – MARCH 2007 ...... 111 5.2.1 – Sample and Implementation...... 111 5.2.2 – Snapshot of AUDENG scale tested at Stage 2 ...... 114 5.2.3 – Results for Feelings Dimension ...... 114 5.2.4 – Results for Arousal Dimension ...... 116 5.2.5 – Results for Appraisal Dimension ...... 117 5.2.6 – Results for Cognitive Effort Dimension ...... 119 5.2.7 – Modifications to the scale before Stage 3...... 120 5.3 – MAJOR SCALE TESTING – APRIL/MAY 2007...... 122 5.3.1 – Sample and Implementation...... 122 5.3.2 – Snapshot of AUDENG scale tested at Stage 3 ...... 125 5.3.3 – Results for Feelings Dimension ...... 125 5.3.4 – Results for Arousal Dimension ...... 131 5.3.5 – Results for Appraisal Dimension ...... 135 5.3.6 – Results for Cognitive Effort Dimension ...... 140 5.3.7 – Modifications to the scale before Stage 4...... 143 5.4 – CONFIRMATORY TEST IN HIGH SCHOOLS – JUNE 2007...... 145 5.4.1 – Sample and Implementation...... 145 5.4.2 – Snapshot of AUDENG scale tested at Stage 4 ...... 146 5.4.3 – Results for Feelings Dimension ...... 146 5.4.4 – Results for Arousal Dimension ...... 150 5.4.5 – Results for Appraisal Dimension ...... 154

 O              P     !  " # $% &  '  ()( 5.4.6 – Results for Cognitive Effort Dimension ...... 160 5.5 – FINAL AUDIENCE ENGROSSMENT SCALE ...... 162 5.6 – ASSESSING GENERALISABILITY OF THE AUDIENCE ENGROSSMENT SCALE ...... 165 5.7 – VALIDITY OF THE AUDIENCE ENGROSSMENT SCALE ...... 166 5.8 – CONCLUSION ...... 169

CHAPTER 6: TESTING THE BRANDCAST PROCESSING MODEL ...... 170

6.1 - INTRODUCTION...... 170 6.2 – RESEARCH QUESTIONS AND HYPOTHESES...... 170 6.3 – RESEARCH STRATEGY ...... 173 6.4 - THE SAMPLE ...... 174 6.4.1 – Procurement of sample ...... 174 6.4.2 – Appropriateness of a teenage sample ...... 174 6.5 - THE FILM STIMULUS ...... 176 6.5.1 – Understanding the film preferences of teenagers ...... 176 6.5.2 – Movie Selection and Analysis ...... 177 6.6 - CONTENT ANALYSIS ...... 180 6.7 – OPERATIONALISATION OF CONSTRUCTS AND QUESTIONNAIRE DEVELOPMENT...... 181 6.7.1 – Brand familiarity ...... 181 6.7.2 – Audience Engrossment...... 182 6.7.3 – Recognition...... 182 6.8 - RESEARCH MANAGEMENT ...... 183 6.8.1 – Pre-film testing ...... 183 6.8.2 – Film Screening...... 184 6.8.3 – Post-Film Testing...... 184 6.9 – DATA PREPARATION AND METHOD OF ANALYSIS ...... 185 6.9.1 – Descriptive Analysis ...... 185 6.9.2 – Model Testing ...... 188 6.10 – RESULTS: DESCRIPTIVE STATISTICS...... 189 6.11 – RESULTS: DIRECT EFFECTS ...... 191 6.12 – RESULTS: INDIRECT EFFECTS ...... 194 6.12.1 – Brand familiarity as a moderator ...... 195 6.12.2 – Feelings as a moderator ...... 196 6.12.3 – Arousal as a moderator ...... 197 6.12.4 – Appraisal as a moderator ...... 198 6.12.5 – Cognitive effort as a moderator...... 199

 O              P     !  " # $% &  '    ()( 6.13 – SUMMARY OF HYPOTHESIS TESTING ...... 201 6.14 – REVISED MODEL OF BRANDCAST PROCESSING ...... 202 6.15 - CONCLUSION ...... 204

CHAPTER 7: CONCLUSIONS AND IMPLICATIONS ...... 205

7.1 – INTRODUCTION ...... 205 7.2 – CONCLUSIONS RELATING TO THE RESEARCH QUESTIONS ...... 206 7.2.1 - Conceptual clarification of product placement and related activities (RQ1, RQ2) ...... 206 7.2.2 - Development of a new theoretical framework of brandcast processing (RQ3, RQ4, RQ5) ...... 206 7.2.3 - Development of the Audience Engrossment Scale (RQ6, RQ7, RQ8) ...... 207 7.2.4 – Direct effects of product placement quality on product placement recognition (RQ9) ...... 209 7.2.5 – Moderating effect of brand familiarity on product placement recognition (RQ10)...... 212 7.2.6 – Moderating effect of audience engrossment on product placement recognition (RQ11)...... 214 7.3 – CONTRIBUTIONS OF THIS RESEARCH...... 215 7.3.1 – Theoretical Contribution ...... 215 7.3.2 – Methodological Contribution ...... 217 7.3.3 – Managerial Contribution...... 218 7.3.4 – Societal Contribution...... 220 7.4 – LIMITATIONS ...... 221 7.4.1 – Limitations of the Audience Engrossment scale development ...... 221 7.4.2 – Limitations of the Brandcast Processing Model testing ...... 222 7.5 – FUTURE RESEARCH DIRECTIONS ...... 223 7.5.1 – Research issues that still need to be resolved...... 223 7.5.2 – Extending Audience Engrossment Research...... 225 7.5.3 – Extending Product Placement Research...... 226 7.6 - CONCLUSION ...... 229

REFERENCES...... 230

 O              P     !  " # $% &  '  ()( LIST OF TABLES

TABLE 2.7.1 – EXISTING DEFINITIONS OF PRODUCT PLACEMENT ...... 32 TABLE 2.7.2 – KEY DIFFERENCES BETWEEN PRODUCT PLACEMENT AND ADVERTAINMENT...... 39 TABLE 5.2.1.1 – RESPONSE RATES AT CINEMA #1...... 112 TABLE 5.3.1.1 – RESPONSE RATES AT CINEMA #2...... 123 TABLE 5.3.1.2 – RESPONSES PER DAY AT CINEMA #3 ...... 124 TABLE 5.3.3.1– INITIAL SUMMATION OF FEELINGS DIMENSION – 21 ITEMS ...... 127 TABLE 5.3.3.2– FEELINGS DIMENSION AFTER REMOVING ‘APPALLED’ AND ‘CROSS’ – 19 ITEMS...... 128 TABLE 5.3.3.3 – GENERAL MODEL FOR FEELINGS – 21 ITEMS ...... 129 TABLE 5.3.3.4 – OPTIMAL FEELINGS SCALE PER MOVIE ...... 130 TABLE 5.3.4.1– AROUSAL SCALE – 11 ITEMS...... 134 TABLE 5.3.4.2– AROUSAL SCALE – 10 ITEMS...... 135 TABLE 5.3.5.1– RESULTS AFTER COLLAPSING SCALE CATEGORIES AND REMOVING NOTW2 – 11 ITEMS...... 137 TABLE 5.3.5.2– LIKING / ENGAGED DIMENSION – 8 ITEMS ...... 138 TABLE 5.3.5.3– NOT LIKING / DISENGAGED DIMENSION – 4 ITEMS ...... 138 TABLE 5.3.5.4– NOT LIKING / DISENGAGED DIMENSION WITHOUT NOTW2 – 3 ITEMS...... 139 TABLE 5.3.6.1– COGNITIVE EFFORT DIMENSION WITH THRESHOLDS CORRECTED – 9 ITEMS ...... 141 TABLE 5.3.6.2– COGNITIVE EFFORT DIMENSION WITH THRESHOLDS CORRECTED AND EASY2/3 REMOVED...... 142 TABLE 5.3.6.3– PROPOSED SCALE FOR COGNITIVE EFFORT WITH DIFF1/2, ATTN1/2/3 – 5 ITEMS ...... 143 TABLE 5.4.4.1– RESULTS FOR LOW VS HIGH AROUSAL (AS PER PILOT STUDY)...... 150 TABLE 5.4.4.2– RESULTS FOR POSITIVE AROUSAL VS NEGATIVE AROUSAL VS BORED (AS PER STAGE 2) ...... 151 TABLE 5.4.5.1– POTENTIAL GENERAL APPRAISAL SCALES ...... 156 TABLE 5.4.5.2 – PREVIOUS EXPOSURE TO THE MOVIE ...... 157 TABLE 5.7.1 – PERSON SEPARATION INDEXES PER MOVIE, PER DIMENSION ...... 167 TABLE 6.5.1 - KEY FINDINGS FROM EXPLORATORY STUDY...... 177 TABLE 6.9.1 – ACCURACY OF RESPONSES ...... 188 TABLE 6.10.1 – RECOGNITION AND FAMILIARITY OF BRANDS IN THE MOVIE ...... 189 TABLE 6.11.1 – DIRECT EFFECTS OF EACH PRODUCT PLACEMENT CHARACTERISTIC ON RECOGNITION ...... 192 TABLE 6.11.2 – DIRECT EFFECTS OF ALL PRODUCT PLACEMENT CHARACTERISTICS ON RECOGNITION ...... 193 TABLE 6.12.1 – MODERATING EFFECTS OF BRAND FAMILIARITY ON RECOGNITION...... 195 TABLE 6.12.2 – MODERATING EFFECTS OF FEELINGS ON RECOGNITION ...... 197 TABLE 6.12.3 – MODERATING EFFECTS OF AROUSAL ON RECOGNITION...... 198 TABLE 6.12.4 – MODERATING EFFECTS OF APPRAISAL ON RECOGNITION...... 199 TABLE 6.12.5 – MODERATING EFFECTS OF COGNITIVE EFFORT ON RECOGNITION ...... 200

 O              P     !  " # $% &  '  ()( LIST OF FIGURES

FIGURE 2.7.2 – THE RELATIONSHIP BETWEEN BRANDCASTING, ADVERTAINMENT AND PRODUCT PLACEMENT40 FIGURE 3.2.1 – BALASUBRAMANIAN, KARRH AND PATWARDHAN’S (2006) PROPOSED MODEL FRAMEWORK..47 FIGURE 3.3.1 – BASIC THEORETICAL MODEL OF BRANDCAST PROCESSING ...... 50 FIGURE 4.5.1 – GENERAL POLYTOMOUS RASCH MODEL ...... 78 FIGURE 5.4.3.1 - ITEM MAP FOR FEELINGS DIMENSION – THE ISLAND ...... 149 FIGURE 5.4.4.1 - ITEM MAP FOR AROUSAL DIMENSION – THE ISLAND ...... 153 FIGURE 5.4.5.1 - ITEM MAP FOR APPRAISAL DIMENSION – THE ISLAND ...... 159 FIGURE 5.4.6.1 - ITEM MAP FOR COGNITIVE EFFORT DIMENSION– THE ISLAND ...... 161 FIGURE 6.14 – BASIC THEORETICAL MODEL OF BRANDCAST PROCESSING (REVISED)...... 203

 O              P     !  " # $% &  '  ()( LIST OF APPENDICES

 O              P     !  " # $% &  '  ()(

 O              P     !  " # $% &  '  ()( CHAPTER 1: INTRODUCTION

“The entertainment and advertising industries must converge to survive”

Scott Donaton, Publisher of Advertising Age (2004)

1.1 – Introduction The need for alternative vehicles of communication is becoming more urgent as consumers become increasingly difficult to reach, despite increases in advertising targeted towards them. Underlying this are developments such as TiVo, commercial-free satellite radio, do- not-call lists and improved email spam filtering which are making it much easier for consumers to avoid traditional advertising communications. TiVo in particular is making advertisers justifiably nervous that the traditional interruptive model is becoming less effective and relevant (Samuel 2004). Furthermore, the targeting of specific market segments has become more difficult for marketers because of the boom in new media channels resulting from deregulation and the rise of digital communications technology. Media vehicles have proliferated, reflecting the fragmentation of consumer audiences into many interest groups and sub-cultures, thus making it difficult for a single medium to capture large audiences within a defined target. And even if audiences are reached, media sophistication has increased viewers’ beliefs in their own ability to ‘see through’ the tactics of advertisers (Bloxham 1998; Hackley and Tiwsakul 2006).

In order to address this issue, marketers have developed an array of new, more comprehensive communication strategies to penetrate awareness (Solomon and Englis 1994). One tool is product placement (PPL), sometimes known as brand placement, which we define as the inclusion of products – branded and/or unbranded – to support entertainment story content (see Chapter 2). In this quest to combat commercial clutter, declining television ratings, muting, zipping, zapping, and increasing audience

 O              P     !  " # $% &  '  (*( fragmentation, product placement is an intrusive, yet subtle way of displaying (branded) products in movies, television programs, music videos, radio programs, songs, video games, plays and novels (Gupta and Lord 1998; Avery and Ferraro 2000; Macklem 2002). Each product placement is comprised of a bundle of characteristics which determine its overall quality. These relate to its modality, prominence, connection to the plot and method of depiction (i.e. whether the actual brand is shown or spoken, or a logo, packaging, signage or advertising is shown). All product placements also have a temporal component, which depending on the media they feature in, may relate to total length of exposure, frequency of exposure, or both. Similarly, depending on the type of media, a celebrity may be positioned near the brand or even use it.

With consumers now more able to avoid traditional advertising techniques, the benefit of product placement lies in the fact that the product placement is consumed in conjunction with the entertainment program. The audience is relatively ‘captive’ during the program, and their ‘this is advertising’ antennae are turned off. It is this unobtrusive entry of the commercial message that makes product placement different from most other forms of marketing communication (McCarty 2004). When consumers recognise a communication as a persuasion attempt, they process the message differently than they would if no such recognition occurred (Balasubramanian 1994). They may get distracted from the message, disengage from the communication or form assessments of the persuasion effort and the company related to the communication (Friestad and Wright 1994). But since the audiences perceive the media vehicle as a form of entertainment and not persuasion, they may not generate counterarguments, thus enhancing the persuasive impact of the product placement (Solomon and Englis 1994). This effect is heightened since entertainment programs (e.g. films, TV shows) are said to have a strong persuasive influence over audience’s social judgements, at least in the short-term (Karrh 1998). The assumption is that receptive audiences are willing to become highly involved in the stories and characters portrayed on- screen, and have powerful responses to a story. These affective responses may then transfer to any brands included in the program (Karrh 1998).

 O              P     !  " # $% &  '  (+( Product placement strategies usually involve the pairing of products with characters who may explicitly or implicitly endorse a product. Therefore, through the mechanics of social learning theory and role modelling, people may consciously or unconsciously remember and later choose to use products that they see fictional characters and actors use because they identify with, and want to be like those people (Bandura 1977; Moschis and Churchill 1978; King and Moulton 1996). Product placement operates under the premise that whilst watching an entertainment program, audiences are in an active processing state, scanning the program for cues about characters, their identity characteristics, and their likely actions (Karrh 1998). Whilst audiences may not be able to mirror all the desired attributes of a successful actor, they may be able to use some easily available consumer goods as a link to desired attributes (Gabriel and Lang 1995; Karrh 1998; Gould, Gupta and Grabner-Krauter 2000).

Therefore, because of the fracturing of media, and the potential effects of product placement on brand awareness and perception, the implementation of product placement continues to grow, but it is still not a well understood practice. In 2006, global product placement expenditure was valued at US$3.36 billion, with growth of 30.3% to US$4.38 billion anticipated in 2007 (PQ Media 2007). When considering the barter/exposure value of non-paid placements, the value of the global product placement industry is estimated to be US$7.76 billion, with an anticipated growth of 20.3% to US$9.33 billion in 2007 (PQ Media 2007). With such large sums of money being spent on the practice, the cost effective management of product placements is now imperative. Yet product placement practitioners’ decision making is often intuitive and based on rather superficial information (Karrh 1995; Pardun and McKee 1996; Pardun and McKee 1999; Karrh, McKee and Pardun 2003; Russell and Belch 2005; Craig-Lees, Scott and Wong 2006). We partly attribute this to the quality of effects-based research in product placement, which to date has shown mixed and inconclusive results. This in turn is due to the lack of a viable product placement processing framework and the lack of agreement as to what effects (or outcomes) are possible and how they should be examined (see Chapter 2).

 O              P     !  " # $% &  '  (,( This thesis therefore presents a theoretical framework to describe product placement processing1 and highlights what effects are possible. This framework is based on the premise that when brands are featured in an entertainment story, they provide a stimulus that an audience member then processes, and that for a product placement stimulus to be deemed effective, some form of memory trace should ensue. We also acknowledge that individual audience members are likely to process the same entertainment program / story and the same product placements differently, depending on their familiarity of the brands that are placed, how connected they are to the program (Russell, Norman and Heckler 2004) and how involved they become with the story. We term this final concept audience engrossment (Scott and Craig-Lees 2005a, p2) and define it as “the degree to which individuals are engaged - affectively, cognitively and behaviourally - with the entertainment content they are consuming, at the time of consumption” (see Chapter 3).

While there are efficient means to assess connectedness (Russell et al. 2004) and brand familiarity, no such tool to assess the quality of an audience’s cognitive and emotional engagement with a program or story (i.e. audience engrossment) exists. Therefore, to remedy this, the central task of this thesis is the development and testing of a scale to capture this concept (see Chapter 4 and 5).

1.2 – The Growth of Product Placement The use of, and reference to, products and brands in entertainment programs to support creative content has a long history. The origins of product placement can be traced back to 1896 when the Lever Brothers secured the placement of their soaps in some of the first films ever produced (Hudson and Hudson 2006; Newell, Salmon and Chang 2006). Although there is evidence that brand/product managers in the early 20th century recognised the valuable exposure that product placements could provide (for example, tobacco companies joined with Hollywood studios to arrange for actors to smoke cigarettes in

1 The model that is developed is actually a model of brandcast processing, which encompasses both product placement and advertainment. However, since the term brandcasting will not be introduced until Section 1.3, we will refer to the model simply as a product placement processing model at this stage.  O              P     !  " # $% &  '  (-( movies in the 1920s), it remained a casual process or a by-product of public relations initiatives until the 1980s whereby the product sponsor did not pay the moviemaker, but rather, loaned the product for use as props in movie production, or provided a service for the production crew in return for exposure in the movie (Spillman 1989; Balasubramanian 1994; Brett 1995; Brennan, Dubas and Babin 1999; Brennan and Babin 2004; Wenner 2004). The 1982 film ET is frequently cited as the pivotal contributor to the recognition of the strategic value of placements, as following its exposure in the film, sales of Reese’s Pieces rose 66% in three months (Reed 1989).

Today, firms have become cognisant of the commercial opportunities afforded by the cameo appearances of branded products and their cost-effective benefits2 and willingly pay entertainment / content producers substantial amounts for product placements which are perceived as advantageous. To efficiently locate these product placement opportunities, the product sponsor often hires a specialist product placement agency to act as a liaison with production studios and secure story scripts before production commences. These scripts are carefully reviewed to locate desirable story contexts for placing a specific product. When these contexts are acceptable to both the producer and the product sponsor, a product placement results (Balasubramanian 1994; Gupta and Lord 1998). Recently however, major advertising and media agencies have become more involved in the process, establishing dedicated branded entertainment divisions (Hudson and Hudson 2006). Products may also appear in entertainment vehicles as a result of mega corporation ‘sweetheart deals’ (e.g. McDonalds and Disney), which are more long-term and reflect a large financial investment, or firms may deal directly with production companies (Cowlett 2000). The industry is also becoming institutionalised with trade associations such as the Entertainment Resources and

2 On a cost-per-contact basis, product placement is very attractive. In fact, the price of many paid placements could not even buy a one-time 30-second advertisement on prime–time television – communication which is likely to be viewed by a smaller number of people and is completely perishable since once it is screened, it will not be screened again unless paid for. Product placement in film (for example) extends both message reach and message life because movies have a worldwide audience and are able to outlast the initial phase of theatre exhibition through video and DVD rentals and sales, and television broadcast opportunities (Balasubramanian 1994). Indeed, multinational marketers select product placement over advertising because “the number of tickets sold for a moderately successful movie worldwide well exceeds the reach of an average advertisement” (Bhatnagar, Aksoy and Malkoc 2004, p103).  O              P     !  " # $% &  '  (.( Marketing Association (ERMA) and the Branded Content Marketing Association (BCMA) providing a forum for advertisers, entertainment creators and distributors, agencies and regulators to discuss and address the changing needs of the advertising and branded content market. They also ensure high quality ethics and standards of operation to the industry and promote the practice of product placement (BCMA 2006; ERMA.org 2006). The beliefs and practices of these various practitioners are discussed in Section 2.4.

The US is the world’s largest paid product placement market, accounting for two thirds of spending, with projected spending of US$2.9 billion in 2007, and double digit (but slowing) growth over the next four years (PQ Media 2007). The next largest markets are Brazil, Mexico, Australia and Japan. Other markets also emerging as significant players include India (whose massive growth is driven by the booming Bollywood film industry and the establishment of specialist product placement agencies and product placement on television), and China, who will be the world’s fastest-growing product placement market in 2007 (up 34.5%) (Nelson and Devanathan 2006; PQ Media 2006; PQ Media 2007). Product placement in the European Union and United Kingdom is also likely to flourish due to the relaxation of rules governing paid product placements on television (Sweney 2006).

Product placement will continue to grow as the digitisation of media makes it easier for product messages to be placed in both new and pre-existing entertainment programs. With more entertainment programs being downloaded from the internet, especially as more homes worldwide acquire broadband access, it is likely that the product placements in entertainment downloaded by individuals will be tailored to those individuals (Galician 2004). In addition, digital television could allow for greater interactivity via the insertion of Internet hyperlinks in television programs, the placing of direct orders by consumers during these programs, virtual advertising and virtual product placement (Reding 2001). A common vision is that viewers, enabled by set-top boxes, will be able to select items onscreen to find out more about them, and possibly buy them directly (Taffel 2004). Such advances in technology may mean that product placement will become less of a simple  O              P     !  " # $% &  '  (/( promotional tool and more of a selling tool (Karrh 1998). This means that in the future, we may see a shrinking of the brand exposure-to-purchase opportunity time lag, with product placement more closely linked to an immediate buying response from the audience (Karrh 1998).

However, the veiled nature of product placements, coupled with this phenomenal growth, has resulted in societal discord. The ethical aspect of product placement has sparked much discussion, with the possible impact on vulnerable audiences the key concern. Younger audiences are one of these vulnerable audiences as they are heavy consumers of entertainment media (Nelson and McLeod 2005) and have less sophisticated information processing skills and ability to counter-argue (Roedder 1981; Strasburger 1995). Moreover, they are keenly trying to establish an identity, often using celebrity role models to assist them with this (Galician 2004). This makes them potentially more susceptible to the associations implied via product placements, especially where a valued celebrity is associated with the product/brand use. Underlying this concern is the notion that via social learning, audiences are more likely to integrate the products and brands portrayed into their everyday lives, and that whilst this learning can be conscious, it is more likely to be unconscious (Bandura 1977).

1.3 – The Research Problem Whilst research on product placement has begun to appear in recent years, it has not kept pace with the growth of the practice. To date, there have been seven major dimensions of product placement research (see Chapter 2), but of most concern for this study is that relating to “effects measurement”. Balasubramanian, Karrh and Patwardhan (2006, p116) point out that “inferences about placements often do not converge across empirical studies (which report qualified impact on memory for relevant brands)”. They attribute this to two factors - the lack of a fully developed conceptual framework and the different research designs used (i.e. lab-based experiments versus field-based research). This coupled with the relative small number of studies conducted, and the disparate focuses of these studies has meant that any validation or generalisation is limited, although some findings have been  O              P     !  " # $% &  '  (0( confirmed (for example, prominent placements have consistently been found to be more effective than subtle placements) (see Chapter 2).

As Balasubramanian, Karrh and Patwardhan (2006) point out, a viable conceptual framework to understand product placement effects is needed. However for such a framework to be operational there needs to be some consensus as to what is being examined (i.e. the nature of the stimulus). To date, over twelve definitions of product placement have been formulated, however this thesis argues that none of these really hit the mark. They are generally either too specific or too vague in describing the activity (see Section 2.7.1). In addition to this, an emerging concern is that, particularly in the context of radio and television, product placement is being associated with a number of diverse practices that should be conceptually distinct, such as sponsorship, endorsement, advertainment and plugs. Therefore, product placement needs to be properly classified to ensure that when we compare research studies, these studies are measuring the same phenomenon. Of particular importance is the resolution of the relationship between product placement and advertainment. As both describe the presence of products and brands in entertainment programs, we develop the umbrella term of brandcasting (the inclusion of products - branded and/or unbranded - in entertainment story content) to envelop both concepts. Following on from this, advertainment relates to the development of entertainment story content to support a brand, whereas product placement relates to brands supporting entertainment story content, as per our definition presented in Section 1.1 (see also Chapter 2). This clarification of concepts informed the research design and data collection approach, giving confidence that the subsequent research was done with a clear and certain idea in mind of what was to be measured and what effects were likely.

Since nearly all past research has focussed on the impacts of the various product placement characteristics which are all under the control of the advertiser and program creator (and usually examined in isolation of each other), the audience member has been relegated to being a passive processor, with no role in the “success” of a product placement. Such an assumption would imply that any given product placement would have the same effect on  O              P     !  " # $% &  '  (1( all people, and that all people watch the same program in the same way. But this approach to research is patently myopic.

The need to understand how individual audience members process product placements underlies this study, with the argument that it is not what product placements do to audiences, but what audiences do with them that governs their success or failure. Yet past studies have generally ignored how audience characteristics (except for basic demographics) may make people more or less likely to counter-argue or be persuaded by a product placement or impact consumer reactions and memory differently. That limited attention has been given to this area also relates to the lack of a viable framework and suitable tools to test this framework. Whilst Balasubramanian, Karrh and Patwardhan (2006) have recognised the need to investigate the role of the audience member by including various individual factors in their theoretical model of product placement, their model is neither operational nor manageable, and in essence, is not a processing model (see Section 3.2). For these reasons, their model is not suitable for this research.

Therefore, there remains a gap for developing an operational model of brandcast processing that involves the audience member as a key participant and identifies the possible effects which may result from this processing3. A new measure of audience engrossment is also needed to properly understand the role of the individual audience member in this process. Given that what is being paid for is exposure during a discrete entertainment program, then as with advertising research, practitioners need to know how the quality of the exposure (i.e. the product placement) and audience characteristics interact, what effects are feasible, and have some way to measure these effects4. To develop the model, academic literature covering areas such as advertising effects, information processing, memory, media usage,

3 Whilst the brandcast processing model should apply to both product placement and advertainment, in this research, the model was tested in a product placement context. 4 As academics we are interested in the interaction of these factors and the concurrent processing of advertising and entertainment. These results are also of interest to regulators and other opponents of product placement in negotiating restrictions over the process.  O              P     !  " # $% &  '  (2( emotions and social learning theory have been consulted, as well as the existing product placement studies.

1.4 – Research Objective and Questions Understanding the impact of context and audience state on brandcast effectiveness is important and raises two issues that this research seeks to address. One is the theoretical issue – the role of the viewer in the success of a brandcast. The second is a methodological problem – how can we investigate different reactions to the same content and how might these cause different effects? This is particularly significant given that a viewer’s engrossment with a program’s content may influence the effectiveness of its embedded placements (Bhatnagar et al. 2004). With this research focussing on understanding how individuals process product placements, the key objective was to develop an Audience Engrossment scale as this would provide theoretical and practical insights into how audiences interact with and process entertainment and the brands contained within these stories.

Therefore, the central research question underlying this research is:

What role does the individual audience member play in the “success” of a product placement?

Germane to this are the following substantive issues or sub questions:

Conceptual clarification: 1. Is product placement well defined? 2. Can product placement be clearly delineated from related activities such as advertainment, sponsorship, endorsement and plugs?

 O              P     !  " # $% &  '    (*3( Development of a new theoretical framework of brandcast processing: 3. Is there an existing operational framework that describes product placement (brandcast) processing? 4. What factors may influence how individual audience members process brandcasts? 5. What effects (outcomes) are realistic to expect from exposure to a brandcast?

Development of the Audience Engrossment Scale: 6. Is there an existing measure which captures the concept of audience engrossment? 7. What dimensions might comprise audience engrossment? 8. What items might comprise these dimensions of audience engrossment?

Preliminary testing of the Brandcast Processing Model: 9. Are higher quality product placements5 more likely to be remembered than lower quality product placements? 10. Does an individual’s level of brand familiarity affect their memory for a product placement of that brand? 11. Do viewers experiencing different levels of engrossment have the same ability to process product placements? Will their memory for the placed brands be the same?

1.5 – Issues informing the methodology Assessing interaction between an advertisement and an entertainment program to measure audience engrossment ideally needs to be done at the time of processing, especially since attention, interest levels, feelings and reactions are likely to change over the duration of the story. To do this would require an experiment-based research design and sophisticated physiological equipment and testing, measuring physical responses such as pupil dilation, eye movement, heartbeat and body heat.

5 By high quality, we refer to placements that may be more prominent, have high plot connection, appear both aurally and visually, or be used by the star. Prior research and theory suggests these should all have significant impacts (see Section 3.5).  O              P     !  " # $% &  '    (**( However, our strong desire to maximise external validity in collecting this data, minimise intrusion on the respondent, and the hope that this audience engrossment measure could be used by a wide variety of parties in naturalistic settings, suggested that such testing was unsuitable. Therefore, a self-report measure (a pen and paper scale) was developed, whereby respondents described the frequency with which they felt certain feelings and had certain physical reactions to an entertainment program, described the amount of cognitive effort they needed to expend in order to follow it and provided an overall appraisal of how much they enjoyed various aspects of the program. Using this method, this information could be collected after watching the entertainment program (not during) as this was considered less intrusive. Whilst such an approach is not without its limitations, least of which is its retrospective nature and requirement to label an emotion, it is an empirical approach to the measurement of emotion that is common and well accepted within the marketing discipline (Stewart, Morris and Grover 2007) (see Section 4.3).

1.6 - Method After assessing the strengths of various measurement approaches, Item Response Theory, and in particular, the one parameter Rasch model6, was selected as the most appropriate to meet the strict assumptions that any measure developed should meet (see Chapter 4). Thus, the Audience Engrossment scale was developed, in a film context, using Rasch Measurement Theory (Rasch 1960; Andrich 1988b). Elements of Churchill’s (1979) classical test theory approach and Rossiter’s (2002) C-OAR-SE model were also incorporated in the item generation phase and specification of the construct as there were certain aspects of both these theories that were compatible. Doing this led to a better developed construct and list of items to refine. To maximise external validity, scale refinement was conducted in the field over four phases and across seven movies, with 1360 actual cinemagoers completing the scale immediately after seeing a movie of their choice. Rasch analysis (using RUMM2020) was used to analyse each stage of data collection and to refine the scale.

6 This case is also referred to Rasch Measurement Theory or Rasch modelling throughout this thesis  O              P     !  " # $% &  '    (*+( After thorough development and testing of the Audience Engrossment scale, preliminary testing of the brandcast processing model was conducted using a previously identified ‘vulnerable’ group: a teenage sample (aged 15-17). This testing, to understand factors driving product placement recognition, was via a multi-stage study involving a quasi- experiment with 191 participants. In this quasi-experiment, participants watched The Island, a movie selected by the researcher with known product placements. They completed questions after the film relating to their recognition of brands within the film and their level of engrossment and program liking (Murry, Lastovicka and Singh 1992)7. Information regarding their level of brand familiarity had been collected four weeks prior to this. SPSS 15 and LIMDEP 9.0 were used to test the theoretical model presented in Chapter 3 and hypotheses presented in Chapter 6.

1.7 – Contribution and Key Findings of this Research 1.7.1 – Theoretical and Methodological Contribution and Key Findings With product placement yet to be well defined and operationalised, the identification and examination of its possible effects on audiences has been problematic and led to inconsistent findings and less-than-optimal practical implementation. Therefore, a significant contribution of this research is the conceptual clarification of product placement, clearly distinguishing it from advertainment, sponsorship, endorsement and plugs. We suggest that product placement and advertainment both fall under our newly proposed umbrella of ‘brandcasting’ (Scott and Craig-Lees 2006). Product placement is positioned as an activity whereby the product or brand is used to support the story, whilst advertainment is said to be when a story is built around the product or brand (Scott and Craig-Lees 2006).

Another theoretical contribution is the development and testing of a theoretical framework of brandcast processing. Extending and refining the ideas of Balasubramanian, Karrh and Patwardhan (2006), this model is arguably the first operational model focussing on

7 Program liking was measured to assess convergent validity of the Audience Engrossment scale. It was not part of the proposed brandcast processing model.  O              P     !  " # $% &  '    (*,( brandcast processing, and is certainly the first to put the audience in a central and active role in the process as a moderator of the effects of product placement or advertainment. Audiences are conceptualised as active processors who bring to each placement processing task a range of characteristics (namely brand familiarity, audience engrossment and connectedness) that affect the quality of the intended memory trace.

To enable this strong audience focus, and have the research conducted from the standpoint of the audience member, an Audience Engrossment scale is developed using Rasch measurement to measure an individual’s level of engrossment with, in this study, a film (see Chapters 4 and 5). Indeed, the scale itself, and the application of Rasch Measurement Theory in a marketing context are important contributions of this research. Importantly, this Audience Engrossment scale allowed us to demonstrate that audience characteristics do moderate the proposed stimulus-response relationship. Specifically it was found that differing levels of audience engrossment (via its four underlying dimensions of feelings, arousal, appraisal, and cognitive effort) do affect how audiences process the stimulus (in this case, a movie) and the product placements contained within it. Brand familiarity was also found to have very strong moderating effects on recognition of those brands placed in the movie. Connectedness was not examined because it is not relevant in a film context (see Section 3.7).

Taking a big picture perspective, this research contributes to the accumulating body of work examining non-traditional marketing communication initiatives and information processing within entertainment programming. It also promotes a change of focus to the role of the audience member in media and advertising research.

1.7.2 – Social Contribution and Key Findings Controversy over product placement stems from many avenues, but most centre around concern for vulnerable audiences and anger towards greedy advertisers and entertainment providers. Opponents fear that audiences generally do not perceive these brand appearances to be advertising and that exposure cannot be avoided since it is part of the entertainment  O              P     !  " # $% &  '    (*-( program. Others dislike the fact that the practice expands the advertising revenue opportunities beyond the thirteen minutes per hour now allowed by the Australian Broadcasting Authority (Canning 2003) and that producers are not required to disclose the names of sponsors of product placement contained within a program.

Whilst in the United Kingdom and European Union regulations surrounding product placement have disallowed the practice (although these are now being lifted), in Australia (as well as most other countries), there has been no regulatory involvement. However, the argument for regulation would be strengthened if effects were demonstrated. By investigating how well the teenage audiences in this study consciously process product placements, this research also offers a strong societal contribution, providing preliminary support for this ethical debate. In this research, teenagers were found to have very high recognition rates of the brands included in the movie they watched, thus giving credence to the complaints of opponents to product placement who worry about the effects of repeated exposure to brands and celebrities endorsing these brands by starring alongside them (see Section 6.4.2).

1.7.3 – Managerial Contribution and Key Findings This research highlights specific actions the product placement planner can take to maximise the effectiveness of their placement in regards to program selection, suitability of different brands, and specific executional strategies.

The Audience Engrossment scale will also help practitioners determine whether the same product placement placed in two different programs having the same viewership and the same audience profiles will generate a different response depending on the individual audience member’s level of engrossment with the program. For example, it was found that films which evoke a lot of intense feelings and physical reactions, or which require considerable cognitive effort would not be as attractive as a film with likeable characters and which audiences really enjoy and can easily follow. Furthermore, once the most suitable program is selected, the results from this study will help guide planners as to which  O              P     !  " # $% &  '    (*.( executional characteristics will most likely enhance product placement recognition (for example, high prominence and plot connection or having the star use the brand).

In regards to which brand to place in which program, it was found that matching the placed brands to the intended audience leads to greater recognition of the product placement. This suggests that product placement may not be a suitable vehicle for launching a new product or attempting to get the product in front of the eyes of a new market. Indeed, with such high costs involved, marketing managers who demand maximum returns on their investments must better understand the relationship between product placement execution, the audience and their combined outcomes.

1.8 – Thesis Outline This chapter has presented background information surrounding product placement, explaining the reasons for its growth and what theoretical mechanisms underlie its premise. It has also discussed the research problem and research questions being addressed and summarised some key findings and contributions of the research.

Chapter 2 provides a critical review of prior research, highlighting areas where research is lacking or has been poorly conducted. It surmises that conceptualisation of product placement and related practices has been poor, and that there has been a misunderstanding of what effects are feasible and how these should be measured. It addresses these issues by re-conceptualising product placement, distinguishing it from advertainment, endorsement, plugs and sponsorship, and discussing what effects are possible (with a strong focus on conscious and unconscious memory). Other research gaps are identified, all attributable to the absence of an operational framework of product placement processing.

Chapter 3 offers a new model of brandcast processing which addresses the limitations of previous studies outlined in Chapter 2. It also argues for the importance of studying product placement (and indeed, any brandcast) from the perspective of the audience member, suggesting that the audience member’s level of brand familiarity and their level of  O              P     !  " # $% &  '    (*/( connectedness and engrossment with the entertainment program will moderate the relationship between the brandcast and recognition. The need for a new Audience Engrossment scale is justified after a thorough literature review revealed that no scale existed that could capture this idea.

Chapter 4 develops this Audience Engrossment scale. Rasch Measurement Theory is introduced and reasons why it is a superior measurement and scale development tool to Classical Test Theory are justified before outlining the item generation process with this comparison in mind. The initial Audience Engrossment scale is introduced, the results of a pilot study involving 69 participants are discussed, and refinements to the scale made.

Chapter 5 presents the next three stages of scale refinement, with the scale tested in cinemas with 1291 actual cinema-goers seeing a movie of their choice. Each of these samples is described, as are the results from each stage and the changes made. The final version of the Audience Engrossment scale is then presented, followed by a discussion of its generalisability and validity.

Chapter 6 uses the final version of the Audience Engrossment scale, as well as acquired brand familiarity data to conduct some preliminary testing of the theoretical model developed in Chapter 3. Specific hypotheses are proposed and tested. A film containing relevant and plentiful product placements is identified as the research stimulus, and a teenage sample sourced. The steps involved in content analysing the film are outlined, as are the steps in the development of the other survey tools. The implementation of the quasi- experiment is then explained, as are the techniques used to analyse the data, before the results presented. A revised model of brandcast processing is then presented.

Chapter 7 offers a discussion of the implication of these results and compares them to existing research and relates them back to the original research questions. Next, the theoretical, methodological, managerial and social contributions of the study are outlined, namely the re-conceptualisation of product placement and related activities, the  O              P     !  " # $% &  '    (*0( development of a model of brandcast processing, the development of the Audience Engrossment scale via the application of Rasch Measurement Theory to a marketing context, and the testing of the brandcast processing model with a teenage (vulnerable) audience. Limitations of the research are discussed and the thesis concludes by suggesting possible directions for future research.

 O              P     !  " # $% &  '    (*1( CHAPTER 2: PRODUCT PLACEMENT LITERATURE REVIEW

“The absence of standards—in measurement, pricing, even definitions—will be the highest hurdle to marketers’ acceptance of branded entertainment as a legitimate marketing tool”

(Sauer 2004, p4)

2.1 – Introduction In this chapter, issues surrounding the accountability of product placement are introduced, highlighting the need for better effects measurement and the setting of viable objectives. Past research regarding product placement is discussed, and the significant gaps in our understanding highlighted. This leads to the identification of four key problems in product placement research, with the proposition that without suitable definitions and parameters in place, effects measurement is limited. These issues are addressed, with the chapter concluding with a discussion of the need for a model of product placement (brandcast) processing and a stronger research focus on the individual audience member.

2.2 – The Accountability Issue With increasing expenditure on the practice, practitioners need to ensure that they are getting value for the money they are outlaying on product placements. To determine this, they need to be able to set measurable objectives and have a clear understanding of what effects are possible. DeLorme and Reid (1999) discussed product placement’s role in the integrated marketing communications mix and argued that because attitudes towards brands develop over time, product placements can strengthen brand name recognition and positioning, thus contributing towards long-term influence. They also suggest that product placement can create associations that are important in building a brand’s image. However, in conjunction with the Hierarchy of Effects model (Lavidge and Steiner 1961), we maintain that building and sustaining awareness and preference should be the fundamental

 O              P     !  " # $% &  '    (*2( goals of product placement, and that the other benefits flow on from these basic objectives being met and as a result of other marketing activities. That said, testing this range of outcomes and seeing whether they actually occur should remain a focus of research.

Previously the only two commercial measurements in practice were a memory-based measure developed by CinemaScore, which calculates a score for the placement based on audience perceptions of the film and recall of the product, and a rather crude measure by Creative Entertainment Services which uses ticket sales as a proxy for the number of impressions arising from a product placement in a movie. However, attempts are being made to develop tools which measure either the monetary value of the product placement or other memory-based outcomes (Russell and Belch 2005), with several companies such as Intermedia Advertising Group (IAG), Brand Advisors, Media Bridge, TAM Media Research and iTVX now developing valuation models and quantitative measures to determine the true worth and effectiveness of placements. Arguably the dominant player (in television measurement at least) is Nielsen Media Research who have developed a service called Place*Views which considers characteristics of the product placement and records audience size at the time of the placement, and offers “the most comprehensive database of brand appearances and mentions on television since the Fall 2003 season” (Nielsen Media Research 2007)8.

However, these measurement tools provide only limited information about whether, when and how product placements are effective, the nature of their impact, and how best to measure this impact. Balasubramanian, Karrh and Patwardhan (2006) suggest that our knowledge is weakened because of the lack of a fully developed conceptual framework and the use of different research designs (i.e. lab-based experiments versus field-based research) and state that our understanding of consumers’ responses to product placement is therefore “not fully evolved” (p116). That said, this does not mean that the research that has

8 In actual fact however, this is solely a content analysis tool, recording what brands are using product placement (advertiser, brand, category), what type of placement they are using (audio, visual, foreground, background, character interaction with the product), and when and where these placements are occurring (network, program, episode, date, time).  O              P     !  " # $% &  '    (+3( been conducted in this area is not substantial or has not been useful. It has, and it continues to grow, but more is still needed. This existing body of research shall be discussed next.

2.3 – Overview of Past Research into Product Placement A thorough review of the literature dealing with product placement research has led to the conclusion that to date, there have been seven major dimensions of product placement research – 1. Content analysis of films, television programs and music videos; 2. Investigation into practitioner attitudes and practices regarding product placement; 3. Research dealing with audience attitudes towards product placement (often with a focus on ethical issues); 4. Analysis of the impact of product placement on brand awareness via recall and recognition tests (i.e. memory-based research using explicit measures); 5. Examination of behavioural impacts such as purchase intention and changes in brand image and brand attitudes (using both explicit and implicit measures); 6. Cross-cultural studies comparing how different cultures respond to one or more of these aforementioned dimensions; 7. Broader exploratory qualitative research

Alternatively, this same research could be categorised into three broader but related strands: practitioner attitudes and practices regarding product placement (1, 2, 7); consumer attitudes to product placement (3, 6, 7); and effects measurement (4, 5, 6, 7). An extended review of the research may be found in Appendix 2.1, but an overview of the major studies and trends in these three groups of research will be presented in the following sections.

Practitioner focussed research is important because it aids our understanding of practitioner concerns and skills and identifies information shortfalls. Understanding consumer attitudes to product placement is also useful because this can give insight as to whether product placement is viewed in a positive or negative light, which in turn can affect processing. However, research measuring product placement effects is the most relevant to this study as  O              P     !  " # $% &  '    (+*( we seek to understand what objectives are feasible, what effects are possible and the best ways to assess them.

2.4 – Investigating practitioner attitudes and practices regarding product placement9 In measuring and understanding product placement effects, it is important to understand what practitioners believe to be the goals of product placement and what effects they hope to achieve. In the early days especially, decisions regarding what made a good product placement were made on superficial grounds (e.g. portraying the product favourably, price of the placement, theme of the movie) (Karrh 1995; Pardun and McKee 1996; Pardun and McKee 1999). Over time though, Karrh, McKee and Pardun (2003) and Craig-Lees, Scott and Wong (2006) found that practitioners began to think more broadly and considered more factors to be pivotal to a product placement’s success. These included using the brand on- screen, having no competitive brands appear in the same program, and ensuring that the brand has a natural relatedness to the program content.

However, there still remained a high degree of reliance upon subjective criteria for decision making across these studies, with the results of the Karrh, McKee and Pardun (2003) study indicating only a mild move towards more quantitative measures of product placement effectiveness. Craig-Lees, Scott and Wong (2006) found sales and purchase intention to be the benchmark of success amongst Australian practitioners. These responses raise the question as to what objectives are being set and why effectiveness is being measured in so many different (and potentially unsuitable) ways. Qualitative work by Russell and Belch (2005) offers good insights here.

Russell and Belch conducted interviews with 56 American product placement practitioners and found that in many organisations, product placement was not part of the integrated

9 Due to a lack of direct relevance to this research, prior content analysis research will not be discussed here, but can be found in Appendix 2.1.  O              P     !  " # $% &  '    (++( marketing plan, but rather an additional activity, tacked on the side, often with no real objectives of its own, except for gaining exposure or to match competitor behaviour. Furthermore, they found that there were no agreed-upon ways to assess product placement’s effectiveness and value (even regarding whether such testing should be done at all), with some focussing on the monetary value of the product placement, some focussing on outcomes such as recall and association, and others simply satisfied with “impressions and/or ego involvement objectives” (p38). Most interesting however was the suggestion that product placement agencies appear not to advocate any form of measurement to avoid being held accountable for their performances and for fear that product placement does not actually work. Other practitioners believed that the limited financial investment did not necessarily warrant much attention being paid to the returns. This may be an acceptable excuse on an individual company level, but not on a billion-dollar industry level which does need to be accountable. Indeed, as Russell and Belch (2005, p83) state “without knowledge of how and when placements work, the value of product placement in marketing planning is greatly reduced”. Perhaps it is best left to academics to take on this task, if there is too much vested interest from industry to do this (i.e. potentially showing that all this money spent does not have any impact). Therefore, there is a major need to assist practitioners in understanding what objectives and effects are reasonable and explaining how product placement works on an individual processing level.

2.5 - Understanding audience attitudes and perceptions about product placement How audiences view product placement as a commercial activity can impact on brand attitudes and/or preferences. Although it may not have an effect on the quality of the memory trace during the consumption of a placement, it could affect the global set of brand associations that the product placement produces (Edell and Burke 1987; Holbrook and Batra 1987).

 O              P     !  " # $% &  '    (+,( Research has consistently found positive attitudes to product placement, including high levels of awareness and acceptance. As early as 1993, Nebenzahl and Secunda found that 70.1% of their sample had positive attitudes towards product placement, and that these attitudes were more positive than those towards commercials. Furthermore, 77.9% agreed that they would allow product placement with varying degrees of encouragement or restrictions. Morton and Friedman’s (2002) findings echoed those of Nebenzahl and Secunda nearly ten years earlier, confirming that consumers do not want product placement prohibited, and that they are not willing to pay more for a movie that contains no product placements. The only objections to product placement have been made on ethical grounds, with guns, cigarettes and alcohol considered less acceptable than other products, especially to women (Gupta and Gould 1997; Gould et al. 2000). It has also been consistently found that people who watch more movies find product placement more acceptable (Gupta and Gould 1997; Argan, Velioglu and Argan 2007).

International research has found that there is general support for product placement across all cultures, especially American (Gould et al. 2000; Karrh, Frith and Callison 2001; McKechnie and Zhou 2003), Turkish (Argan et al. 2007) and Indian (Panda 2004) consumers. Gould, Gupta Grabner-Krauter (2000) found that there was no difference between American, French and Austrian consumers in accepting non-ethically-charged products. Generally however, the Americans found product placement the most acceptable, with the Austrians finding it the least acceptable. Karrh, Frith and Callison (2001) found that there was no difference between the amount of attention paid to product placement by Americans and Singaporeans. However, American audiences were savvier, being more likely to believe that brand appearances in programs were the result of paid advertising efforts. Singaporean audiences had more ethical and regulatory concerns about the practice. Similarly, McKechnie and Zhou (2003) found that Chinese consumers were less accepting of product placement than American consumers. Yet unlike American consumers, male and female Chinese consumers differed little in terms of their attitudes towards the placement of ethically-charged products. Argan, Velioglu and Argan (2007) found that there was a favourable attitude towards product placement amongst Turkish moviegoers, but extensive  O              P     !  " # $% &  '    (+-( commercial activity in movies was perceived as less ethically acceptable. Finally, Indian consumers were also positive about product placement, describing it as acceptable, frank, amusing and pleasant (Panda 2004).

Mediaedge:cia found that in all countries, 16-34 year olds were the most likely to notice product placements and to consider trying the products they see in films. They also (more than any other age group) felt it makes sense to see brands in films (Hall 2004). Furthermore, Nelson and McLeod (2005) found that those adolescents who were more attuned to brands and were more brand conscious were more aware of product placements and were more favourable towards the practice. Adolescents also considered other people to be more influenced by product placements than themselves, especially those peers that they were not as close to (i.e. other classmates as opposed their own friends). Baby boomers have also been found to have a positive attitude, regardless of level of movie- going consumption. Any concerns this group had related to ethically charged products or product placement activity directed towards children (Schmoll, Hafer, Hilt and Reilly 2006).

Attitudes to how products are portrayed in programs have been found to be affected by a number of factors. For example, d’Astous and Seguin (1999) found that consumer evaluations of product placements differed depending on the type of program in which they occurred, with evaluations most negative when they featured in a mini-series / drama. In contrast, high sponsor-program congruity led to better evaluations and better ethical judgements about product placement. They also found that implicit placements (i.e. subtle, background placements) were perceived as less ethical than more explicit placements. Russell (2002) and Panda (2004) found that a high level of congruence between the brand and the storyline was received more favourably by audiences and that when the brand looked out of context or was forced upon the plot, it created irritation among the audience. Panda (2004) also found that the consumer’s evaluation of an explicit integrated product placement was more positive than that of an implicit product placement and that implicit

 O              P     !  " # $% &  '    (+.( placements were judged to be significantly less ethical than explicit placements, confirming the earlier research by d’Astous and Seguin (1999).

Therefore, we can see that product placement is a widely accepted activity across all genders and ages and people from all countries, meaning that people should be quite receptive to processing product placements and forming positive brand associations. Teenagers are particularly receptive to product placement, however given their potential vulnerability (see Sections 1.2 and 1.7.2), effects of product placement on this group should be better understood. The only consistent concerns relate to the placement of ethically charged products and a dislike of placements that are out of context or too surreptitiously placed.

2.6 – Past research measuring product placement effectiveness10 Considering its use in practice, there has been relatively little research regarding product placement effectiveness. Past research has primarily focussed on examining the impact of various executional elements of product placement and can be divided into two groups. The first group, started by the earlier researchers, and which is by far the biggest group, employs explicit memory-based research, generally aided or unaided same-day or day-after recall or recognition (e.g. Steortz 1987; Sabherwal, Pokrywczynsji and Griffin 1994; Babin and Carder 1996; Gupta and Lord 1998; Brennan et al. 1999; d'Astous and Chartier 2000; Law and Braun 2000; Russell 2002; Scott and Craig-Lees 2003; Brennan and Babin 2004; Russell et al. 2004; Scott and Craig-Lees 2004). The second group of researchers have measured effects implicitly (e.g. Law and Braun 2000; Auty and Lewis 2004a; Law and Braun-LaTour 2004; van Reijmersdal, Neijens and Smit 2007).

Early studies (e.g. Steortz 1987; Ong and Meri 1994; Babin and Carder 1996) found relatively mild and mixed effects on memory from product placement, suggesting that not

10 N.B. This review only discusses research using movies or television programs as the stimulus, as this is the most relevant area for this study.  O              P     !  " # $% &  '    (+/( all product placements are equally effective. Indeed, the mixed and weak results of these early studies on effectiveness are in part because they generally failed to recognise the multi-dimensional nature of product placement. They aimed to determine how many brands could be remembered and made no attempt to explain why this may be. In doing so, they tended to define product placements as similar, regardless of their modality, prominence, or level of plot connection, and considered all audience members to process placements in the same way. Recently however, there have been some studies that have attempted to consider these complexities.

Gupta and Lord (1998) were the first researchers to extend the product placement literature, by considering the effects of prominence and mode of product placement on recall. In doing so, they demonstrated that some placements were better recalled than others. For example, prominent placements accrued higher recall than subtle placements. Brennan, Dubas and Babin (1999) also concluded that prominent placements were better recognised than subtle placements and determined that prominence accounted for a greater percentage of variation in viewer recognition than that which was explained by placement exposure time. d’Astous and Chartier (2000) also found that prominence enhanced recognition (but had a negative impact on recall).

Research addressing the effect of aural versus visual versus audio-visual delivery on product placement recall and recognition has produced conflicting results. Sabherwal, Pokrywczynski & Griffin (1994) found that audio-visual placements accrue the highest recall, as did Law and Braun (2000) who found that audio-visual placements have the highest recall, followed by visual and then aural placements (although audio placements are better recognised than visual placements). However, Russell (2002) and Gupta and Lord (1998) found that audio placements led to higher recall than visual placements.

The role of plot connection and integration has been researched by Russell (1998) and d’Astous and Chartier (2000), and is an important area to understand given that the success of product placement is supposedly grounded in the notion of seamless integration between  O              P     !  " # $% &  '    (+0( product and plot. A brand with a higher plot connection contributes much to the story, providing a major thematic element due to its high integration to the story (Russell 1998). d’Astous and Chartier (2000) found that integration of the placement with the plot had a positive impact on consumer liking and acceptance of the placement, but a negative impact on recall. This finding was reinforced by Russell (2002) who found that incongruent placements (i.e. those that were visual and had high plot connection, or aural and had low plot connection) were better recalled than congruent placements. This was particularly the case for visual placements – plot connection did not have significant effects on recall of audio placements. However, like d’Astous and Chartier (2000), Russell (2002) also found that congruent placements led to greater attitudinal / persuasive changes than incongruent placements.

The impact of audience characteristics on product placement processing has had limited attention, but the effect of connectedness on recall is one aspect that has been studied. Connectedness refers to the intensity of the relationships viewers develop with television programs and the characters in those programs (Russell et al. 2004). The authors suggest that the processing and storage of program-specific information will differ between high and low levels of the connectedness construct, with highly connected viewers finding the information in the program to be more essential to their lives than less connected viewers. Highly connected viewers consider the content more important and relevant to their world and may even mould characteristics of their lives to match those depicted in the program, forming relationships with the characters who then become a source of influence, especially in relation to their product consumption. Therefore Russell, Norman and Heckler (2004) hypothesised that as connectedness increased, memory for product placements would improve, and found that highly connected viewers recalled significantly more brands than low-connected viewers. Extending this research, Russell and Stern (2006) looked to answer the question of how characters’ relations to placed products and consumers’ relations to those characters may affect consumers’ attitudes to the product. They found that consumers align their attitudes towards products with the characters’ attitudes to the product, and that this alignment process is driven by the consumers’ extra-program  O              P     !  " # $% &  '    (+1( attachment to the characters. This demonstrates the character’s effect on viewers’ attitudes towards product placements, especially when the character’s attitude towards the product is positive.

In the one study, Scott (2002) aimed to identify which audience characteristics and which product placement characteristics led to improvements in recognition, and found that pleasure, cognitive effort and star liking had positive relationships with recognition, as did product placements which were visual in nature, had a high amount of screen time, and depicted products that were familiar to the audience. Prior to this research, no single study had investigated so many variables that could impact conscious memory, or considered both audience and executional perspectives. Of particular interest and significance was the concept of movie involvement (comprised of both emotional and cognitive aspects)11, which was considered in product placement research for the first time.

Despite using explicit measures herself, Russell (2002) recognised the limitations of pure explicit memory-based research, stating that the reliance on brand recall and recognition measures presumes that the effects for memory are similar to the effects for attitude, and pointed to the absence of correlations between memory and attitude measures often found in the persuasion literature (e.g. Petty, Cacioppo and Schumann 1983) which suggest that the memory-attitude relationship is not necessarily linear. Since recall may be a poor predictor of persuasion (Mackie and Asuncion 1990), she suggested that research on the effectiveness of product placements should investigate both memory and attitude effects. Her study supported this contention, showing that conditions that maximised memory did not necessarily maximise persuasion. Whilst incongruency between modality and plot connection improved memory, congruency enhanced persuasion (Russell 2002).

However, Law and Braun (2000) were the first researchers to consider using implicit measures to explore product placement effects. In doing so, they found that although placed

11 This movie involvement concept is akin to the audience engrossment concept that this research conceptualises and develops (see Chapters 4 and 5).  O              P     !  " # $% &  '    (+2( products were chosen more frequently than products that were not placed, this choice was not found to reliably correlate with recognition or recall. Auty and Lewis (2004a) also found that there was no difference in the product children chose following exposure to a film, regardless of whether they could recall the product placement or not. van Reijmersdal, Neijens and Smit (2007) used implicit tests to look at the effects of product placement on brand image – something that had previously remained unstudied, which is surprising given that brand image change is often mentioned as one of the benefits of using product placement (Karrh 1998; DeLorme and Reid 1999). They found that brand image changed in the direction of the program and that there was no effect of conscious brand memory on brand image. Therefore, without consciously remembering having seen the product placement, exposure to the product placement still affected brand image. These memory results are in line with Law and Braun (2000) and Auty and Lewis (2004b) who showed that product placement effects on brand choice were unrelated to explicit memory. They also indicate that brand image is influenced implicitly, which means that image is influenced without conscious memory of the exposure. In sum, these results support the idea that brand image and brand memory are processed differently, and are in agreement with the evolving view that different measures are needed to estimate the different effects of brandcasts, depending on the goals of that brandcast.

2.7 – Evaluating past research and moving forward The varied and inconsistent findings of past product placement research (e.g. Gupta and Lord 1998 versus Sabherwal, Pokrywczynski and Griffen 1994 in regards to effects of modality; d’Astous and Chartier 2000 versus Law and Braun 2000 in regards to the effect of prominence on recall; the use of implicit versus explicit memory measures) have been attributed by Balasubramanian, Karrh and Patwardhan (2006) to the lack of a fully developed conceptual framework to understand effects. The first step to correcting this and being able to measure any effects is to first precisely specify the domain of the construct under investigation - product placement. As suggested by Churchill (1979), the researcher must be exact in delineating what is and is not included within the definition. Indeed, to  O              P     !  " # $% &  '    (,3( measure the presence, absence or amount of any property in the real world, we must provide an operational definition of the property that is free from ambiguity and arbitrariness (Clancey 1994). But with no such parameters yet established for product placement, it remains somewhat ill-defined and difficult to operationalise. The parameters that are needed relate to distinguishing product placement from the other different forms of entertainment-based communications messages and their different purposes. So first, a suitable definition must be developed, and then a typology which distinguishes product placement from related activities.

2.7.1 – Conceptualising Product Placement As can be seen in Table 2.7.1, there are many definitions of product placement currently in circulation. Whilst many do iterate some key themes, namely, the inclusion of products or brands in an entertainment program; the fact that this inclusion has a reward factor (monetary or other); and that this inclusion enhances the value of the product or brand, many have differing levels of inclusivity. Some definitions are very specific, such as “product placement is the inclusion of consumer products and services in motion pictures distributed to theatres by major Hollywood studios in return for cash fees or reciprocal promotional exposure for the films in marketing advertising programs" (Clark 1991, in Nebenzahl and Secunda 1993 p1, emphasis added). Does this mean that products featuring in an Australian or Bollywood movie would not be classed as product placements? Is payment only in the form of cash or reciprocal promotional exposure? Many other definitions also attempt to limit product placement to just film or television (e.g. Gupta and Gould 1997). However, as outlined in Chapter 1, product placements can appear in a multitude of entertainment media not just limited to these two. Most also limit product placement to the placement of brands only and exclude the fact that the inclusion of a generic product can also have an impact on the overall product category (e.g. Balasubramanian 1994; Gupta and Gould 1997; Karrh 1998; Panda 2004). Finally, although not explicitly stated, definitions tend to infer that product placement is a visual

 O              P     !  " # $% &  '    (,*( Table 2.7.1 – Existing Definitions of Product Placement • “the inclusion of a brand name, product, package, signage or other trademark merchandise within a motion picture, television show, or music video (Steortz 1987, p22) • "the inclusion of consumer products and services in motion pictures distributed to theatres by major Hollywood studios in return for cash fees or reciprocal promotional exposure for the films in marketing advertising programs" (Clark 1991, in Nebenzahl and Secunda 1993, p1). • “a paid product message aimed at influencing movie (or television) audiences via the planned and unobtrusive entry of a branded product into a movie (or television program)” (Balasubramanian 1994, p31) • “incorporating brands in movies in return for money or for some promotional or other consideration” (Gupta and Gould 1997, p37) • “the paid inclusion of branded products or brand identifiers, through audio and/or visual means, within mass media programming” (Karrh 1998, p33) • “the inclusion of a product, a brand name, or the name of a firm in a movie or in a television program for promotional purposes” (d'Astous and Chartier 2000, p31) • “commercial insertions within a particular media program intended to heighten the visibility of a brand, type of product, or service” (La Pastina 2001, p541) • “the presentation of branded goods on screen either visually (if the product is shown) or verbally (if it is mentioned or described)” (Rössler and Bacher 2002, p99) • “the practice of including a brand name, product, package, signage or other trademark merchandise within a motion picture, television or other media vehicles for increasing the memorability of the brand and for instant recognition at the point of purchase” (Panda 2004, p7) • “the manipulation of features of television and movie material for commercial purposes” (Kretchmer 2004, p39) • “the inclusion of a brand name, product or logo within a scripted medium” (Schneider and Cornwell 2005, p323). • “the intentional incorporation of a brand into editorial content” (van Reijmersdal et al. 2007, p403)

 O              P     !  " # $% &  '    (,+( phenomenon. Yet defining product placement in such a way as Stoertz (1987, p22, emphasis added), and describing “package, signage or other trademark merchandise” ignores the fact that product placements may be aural or audio-visual in nature.

A good definition will clearly, tightly and succinctly describe what product placement is. We feel that product placement is better described by defining it in terms of the actual act, namely, the inclusion of products - branded and/or unbranded – to support entertainment story content. This is a slight modification to our original definition (Scott and Craig-Lees 2006)12 to ensure that the definition remains as broad as possible in regards to what entertainment media a product placement can feature in. Our previous definition included the word ‘program’, but we feel that this has too many connotations to television and film, and may be interpreted too narrowly.

Note the term ‘paid’ is not included in our definition. Karrh’s (1998) definition infers that product placement only refers to cash payments, and it is unclear whether he is also considering other activities that might constitute product placement such as the lending of products and use of sets. We know this unpaid aspect of product placement is still very prevalent, as per the spend figures outlined in Sections 1.1 and 1.2 which estimated the value of unpaid product placements to be US$4.4 billion (PQ Media 2007). Under Karrh’s (1998) definition, this activity would not be classified as product placement.

Further justifying the decision to remove the concepts of ‘paid’ and ‘promotional purpose’ from the definition is that they infer that all placements are there to influence audiences, and ignores the fact that not all placements are directed by brand/product managers. To insinuate that everything is put in a movie to influence audiences would be to say that directors and writers have lost complete control of their creative processes and that it is advertisers who are making entertainment, not them. Every minute detail (including each prop) contributes to the bigger picture, with each subtlety revealing something vital about

12 This original definition was: Product placement refers to the inclusion of products – branded and/or unbranded – to support an entertainment program (Scott and Craig-Lees 2006).  O              P     !  " # $% &  '    (,,( the characters or the action that is about to take place or adding realism to the scenes. Whilst formal product placement deals do take place (and sometimes may include completely irrelevant brands featured gratuitously in a scene), many products still appear without any such deal and explicit payment.

As both consumers and researchers, we cannot always know which products appear as a result of a deal and which ones appear purely for creative purposes. Sometimes we can, when sponsors are transparent about such deals, and partake in publicity to maximise the product placements effects (e.g. Audi in I, Robot or BMW in Tomorrow Never Dies). But generally, we have no way of knowing what products in the entertainment program were paid for and which were not. Therefore, we process all the different props, costumes and settings in entertainment stories in the same way, regardless of whether it was the result of a formal placement or not. So relative to processing, the paid nature of product placement would seem irrelevant13. What is important is that the product appears, how it appears, and how it is incorporated into a scene.

2.7.2 - Distinguishing product placement from other activities Having explored conceptualisation issues, another crucial task is to establish the parameters of product placement. This means clearly differentiating it from related communication forms such as sponsorship, endorsement, plugs and advertainment. Such distinctions are presently unclear, with many of these concepts and practices morphing in to each other. For proper effects measurement, these need to be disentangled.

Common use and dictionary meanings suggest that to sponsor means vouching for, favouring, being responsible and/or to be surety for, or to support (someone) in a fund-

13 However, in terms of effects, the issue of payment does become important because advertisers want to know that they are gaining value from their advertising spend. When the product placement is paid for, issues of accountability become relevant. The advertiser needs to know what they are buying, and they need some way of measuring whether they are getting value for money. So in this way, the research need surrounding product placement comes from the issue of payment and accountability, and therefore has a strong managerial purpose and motivation.  O              P     !  " # $% &  '    (,-( raising activity by pledging a certain sum for each unit completed. On the other hand, to endorse is to confirm, sanction, countenance, or vouch for (statements, opinions, acts, people), through an endorsement. Thus, there is considerable overlap in these terms, especially in common every day use.

It is also understandable why these terms have been (mis)used in the context of product placement as some elements of product placement may be related to these acts. For example, the use or reference to a brand, in a positive context, by a star or character in an entertainment program can be seen as an implied endorsement14 (i.e. vouching for or supporting the brand). This endorsement within the product placement context could also be linked to an aspect of product placement quality - the use of that brand by the star(s) (see Section 3.5.3). Sponsorship in a marketing context can be used to describe the act of paying for or contributing towards the expenses of a media program, performance or other event or work, in return for advertising space or rights, thus linking a brand name, product or logo with the medium in a supposedly philanthropic but overt way (Schneider and Cornwell 2005). However, this act needs to be properly identified as sponsorship, and not product placement even though it pairs a brand and an entertainment program. Indeed, to avoid future confusion, it is important that these concepts be understood and properly distinguished from both each other, and also from product placement. Any further use of these terms in a product placement context should be avoided.

A further activity that also needs to be distinguished is plugs, which Roehm, Roehm and Boone (2004) define as an on-camera discussion of a brand, either off-the-cuff or as part of a formal endorsement a person might have with a brand. Whilst these plugs occur during an entertainment program (generally news programs, talk shows and interviews), they cannot be considered product placement because they are linked to the individual and not to the program content itself. For this reason, they are most closely related to endorsement.

14 However, this would remain the least common method of endorsement, with the more explicit type seen outside of program content in traditional advertising the more common approach.  O              P     !  " # $% &  '    (,.( The relationship between product placement and advertainment is somewhat more complex. Advertainment describes programming that is designed to sell as it entertains, and thus sees the merging of entertainment programming and advertising (Deery 2004). The key element of advertainment is that the messages do not appear to be advertisements, but entertainment, thus capturing attention whilst the brand takes either a central or incidental role. The term advertainment (also known as branded entertainment, brandtainment, brandsploitation and brand integration) emerged in the late 1990s and was originally associated with the use of a short film format (anything from two to ten minutes in length), but other genres such as computer games and music video clips have also adopted this format. Examples of advertainment include The Scout, a six minute film about a baseball scout who is past his prime, and which supports the lawn equipment brand Craftsman, DKNY in the eight minute film Friday Night Fever, the Visa-sponsored 17 minute Ecology of Love featuring the bands N.E.R.D and The Neptunes, and the 22 minute film screened at the 2004 Tribeca Film Festival in which Toyota’s Scion was the star of a docudrama about two musicians who get their first driver’s licence.

Kretchmer (2004) argues that two forms of advertainment can be identified - one where the advertisement is the entertainment; the other where the entertainment is the advertisement. The former is exemplified by Anheuser-Busch’s Bud Bowls which originated in 1989 and saw the Bud and Bud-Light battle for gridiron glory. This advertisement migrated to the Internet and transformed in the real domain with people betting on the outcome. For the latter she gives the example of Heather Locklear endorsing Preference during ads in Spin City, Katie Holmes endorsing Lumia in ads during Dawson’s Creek and Jessica Alba supporting Feria in ads during Dark Angel, arguing that when we watch the show, we are primed to think of the product when we see the stars from the ad sporting their beautifully coloured hair. Since she defines advertainment as vehicles created solely to spotlight advertisers, one then presumes, using that same logic that Spin City, Dawson’s Creek and Dark Angel were programs initiated and paid for by the advertisers.

 O              P     !  " # $% &  '    (,/( This linking of advertainment to entertainment vehicles created solely to spotlight specific advertisers does not create a problem, but Kretchmer’s description of the two forms is problematic. It seems reasonable that advertainment should describe a situation where the advertisement is the entertainment, and the Anheuser-Busch’s Bud Bowls example is a good one. However the other form, that the entertainment is the advertisement in the Spin City, Dawson’s Creek and Dark Angel examples are confusing as it suggests that the purpose of these programs is to advertise products15. In the traditional understanding of product placement, the program content is developed independently of the needs of the product or brand, and the product or brand takes advantage of the existing program content. Therefore, there is a key difference between product placement and advertainment in regards to the degree of ownership and control that the advertiser has over the process.

Russell and Belch (2005) also confuse matters with their terminology, using the terms ‘hybrid advertisement’ and ‘sponsorship’ in the context of product placement, and classify sponsor-owned shows (such as the early radio ‘soap opera’) as a form of product placement. But these distinctions are ambiguous and allow all of the following conditions to qualify as product placement:

• A program that is owned and/or strongly endorsed by a brand. Here only one brand is associated with a program, with the programs purchased by a sponsor and featuring only the sponsor’s brands. Early examples include Lux Theatre and the Lucky Strike Program with Jack Benny (both radio programs). The practice is less common in modern programs where multiple advertisers and/or sponsors contribute, although recently in Australia, Wild Turkey sponsored televised celebrity poker matches, featuring in the program name (Wild Turkey Joker Poker), with the brand mentioned and shown countless times throughout each show. Also in Australia, a tabloid magazine (Famous) now sponsors a paparazzi-fuelled show imported from the US, calling it

15 It is probably best to think of these examples as endorsement, with the television ads purposely timed to screen during the endorser’s program, thus heightening the effectiveness and possible priming.  O              P     !  " # $% &  '    (,0( Famous presents Hollywood Uncensored, with the magazine advertised throughout. Online, Anheuser-Busch has developed a seven channel television at www.bud.tv.

• A program developed to support the brand. Though a number of terms abound, the terms advertainment and advergaming are of most relevance to this idea. Advergaming sees brands inserted into computer games and examples include: the Nestle Nesquick and Kellogg’s CocoPops websites; the basketball advergame for Sprite linked to the FIBA World Basketball championship; and the Barclays social advergame The Transaid Challenge (supporting charity group Transaid). Advertainment can also include mini-movies with varying levels of mentions of the advertiser’s name. Examples include BMW’s The Hire, starring British actor Clive Owen and Skyy Vodka’s development of www.skyy.com which houses several mini-movies.  • A program where brands are used as props to enhance/support program content (at the discretion of the program creator). Brands used this way can be selected by the program creator for payment or non payment, although the former is increasingly more common.

However, the new definitions and distinctions presented here would classify these scenarios as three separate activities. Using our classification, the first condition may better be considered as sponsorship, the second a form of advertainment, with the third example being product placement in its purest form.

Since advertainment and product placement are different, but related, activities, it could be helpful to have an ‘umbrella’ label for this evolved practice. Product placement was coined at a time when the practice was geared towards products as props and the strategic value was not fully realised by brand managers. This realisation, coupled with the emergence of advertainment suggests that it may be useful to create a new term and a new definition now that these newly established parameters are in place.

This research offers the term ‘brandcasting’ as an umbrella term for both concepts, reflecting the notion that the brand is deliberately cast in the story. It should be noted that since this term was first coined by us in 2005, it has started to appear in the popular press,  O              P     !  " # $% &  '    (,1( sometimes with slightly different meanings16. However in this research, brandcasting relates to the inclusion of products - branded and/or unbranded - in entertainment story content17, and it can take two forms – advertainment or product placement (Scott and Craig-Lees 2006). More specifically, advertainment relates to entertainment story content designed to support a brand. The inclusion of a brand in an advertainment is always purposeful (as this is its raison d’être) and hence paid for, and the advertiser has more creative control. Product placement on the other hand relates to a product or brand used to support content and may or may not be deliberate or paid for (see Table 2.7.2). Furthermore, while both are forms of brandcasting, the purpose, function and viewer’s processing of the brand is likely to be different.

Table 2.7.2 – Key differences between Product Placement and Advertainment Product Placement Advertainment Advertiser control Low(er) High(er) Program development Independent of Developed for/by the advertiser needs advertiser Paid for by advertiser Not necessarily Always Degree of brand integration with story Variable High Inclusion of brand has commercial intent Not necessarily Always Purpose of underlying story Entertainment Advertising

The relationship between advertainment and product placement may be viewed as a continuum (Scott and Craig-Lees 2005b) (see Figure 2.7.2). American Express’ webisodes starring Jerry Seinfeld and Superman are clear advertainment. Nokia being used by key characters in Charlie’s Angels is product placement. Somewhere in the middle lie movies such as Herbie: Fully Loaded (where a Volkswagen Beetle is the central character and hero) or Where the Heart Is (which is set in a Wal-Mart store). Here, the brand takes a

16 For example, Wikipedia defines it as a form of digital broadcasting where brands pay to create instructional and informational videos that are useful and relevant to their target audience. A recent Brand Republic article defines it as self-expression by brand owners (Kolah 2006). 17 N.B. As with the definition for product placement, the definition for brandcasting has also been slightly altered. Our original definition was ‘the inclusion of products – branded and/or unbranded – in entertainment programs’ (Scott and Craig-Lees 2006). However, the word “program” has been removed and replaced with the more general “entertainment story content” to reflect the breadth of media that brandcasts can be found in.  O              P     !  " # $% &  '    (,2( critical role in the film, but it is still just supporting the story; the story is not secondary to the promotion of the brand (but the advertiser’s degree of control may be).

Figure 2.7.2 – The Relationship between Brandcasting18, Advertainment19 and Product Placement20

BRANDCASTING The inclusion of products – branded and / or unbranded – in entertainment

Advertainment Product Placement Content designed to Product or brand used to support a brand support content

2.7.3 - Measuring effectiveness Central to measuring effectiveness is the need to have a clear idea of what effects are possible, and how they should be measured. Yet, identifying these effects remains a significant issue within the product placement literature. As discussed in Sections 2.4 and 2.6, this debate is magnified as there is little consensus or knowledge as to what effects practitioners are hoping to gain from product placement, what effects could theoretically arise from product placement, and what effects have actually been demonstrated through empirical studies.

For example, Balasubramanian, Karrh and Patwardhan (2006) adopt the C-A-C pattern of response to stimuli (Lavidge and Steiner 1961) suggesting that several cognitive, affective and conative effects are possible. However some of these do not seem operational or overly useful (e.g. brand typicality, brand portrayal rating), and the conative effects may be

18 Brandcasting – the umbrella term to describe both advertainment and product placement, whereby a product – branded and/or unbranded – is included in entertainment story content. 19 Advertainment – the inclusion of products – branded and/or unbranded – whereby the entertainment story content supports the brand 20 Product placement - the inclusion of products – branded and/or unbranded – to support entertainment story content  O              P     !  " # $% &  '    (-3( considered too difficult to isolate to product placement alone (e.g. identification with the brand, purchase intention, brand usage behaviour). In particular, purchase and sales are influenced by a range of factors - some controllable by marketers, others not - so the most that one can expect from advertising efforts, including brandcasts, is that preference for the brand is created and/or sustained, whether this be by building or sustaining brand awareness, or slight changes in brand attitude or image. Further complicating this goal is the fact that the entwined nature of joint promotional campaigns makes it difficult to tease apart the effects of brandcasts from other forms of marketing efforts. For this reason, to gauge brandcast effects we need to understand what audiences do with them at the point of exposure.

Basic models of advertising effectiveness, all of which are based on the Hierarchy of Effects model (Lavidge and Steiner 1961; Palda 1966), assume that for any stimulus (i.e. an advertisement, a product placement) to be processed, the audience must be exposed to it and pay attention to it. Whilst criticisms of the model’s linear nature abound (e.g. Palda 1966), most agree with awareness (either conscious or unconscious) being an initial assumption underlying all other possible effects (such as preference and purchase). In this way, the first stage of the hierarchy, awareness, must take place for the other effects to happen (Barry 2002; Weilbacher 2002). So unless audiences are aware of, remember, and store brandcast information, there is no way that these brandcasts can be thought to be effective, or to reliably correlate with any level of preference or intention formation. Therefore, retention (conscious or unconscious) should be the basic effect in all processing models, with all other effects flowing on from that.

As stated in Section 2.6, there are two possible forms of memory testing that can be used to assess these effects - explicit and implicit memory tests. Conscious memory taps into the deliberate recollection of facts and past experiences, and there are two basic types of explicit memory tests to capture this - recall tests and recognition tests. Such tasks require an individual to consciously think back to a prior exposure episode (in this case, a brandcast), and intentionally attempt to access the information that was presented (Shapiro  O              P     !  " # $% &  '    (-*( and Krishnan 2001). More specifically, recall tests require people to retrieve memories without the benefit of any hints or cues and thus rely exclusively on conscious memory to identify the memory trace. This can take a number of forms – free recall, serial / ordered recall or cued recall. Recognition tests require people to examine a list of items and identify those they have seen before, or to determine whether they have seen a single item before. Such tests may rely on both conscious and unconscious memory as it is often used to identify sensory memory content (Baddeley 2007; Cohen and Conway 2007).

Whilst the focus of advertising research has been on conscious storage and retrieval, there is a growing interest in unconscious associations, with implicit memory testing now being discussed. Specifically, Law and Braun-LaTour (2004) argue that recall and recognition are not capable of detecting the more subtle effects of product placement, and thus advocate the use of implicit measures. Unconscious memory sees an individual change their performance on a task due to a prior exposure episode, or perhaps by constant or repeated exposure, but they do not deliberately attempt to recollect this previously encoded information (Schacter 1987). This is evident in indirect tests (e.g. sentence completion, word association, projective tests) where consumers do not use conscious memory retrieval (Krishnan and Chakravarti 1999). Such retrieval represents unconscious learning, and these memories have been shown to lead to a response bias in which there is a greater likelihood of using the previously seen information to complete a task without the awareness of doing so (Shapiro and Krishnan 2001; Lee 2002). In this way, implicit memory tests can measure changes in attitude towards brands, namely changes in preference and choice, which are good predictors of purchase intention and even sales. This infers that failure to consciously remember exposure to a brandcast does not preclude the possibility that it has still affected consumer behaviour processes such as brand consideration (Auty and Lewis 2004b). Consequently, these researchers argue that implicit tests should be considered more and utilised in product placement (and presumably all brandcast) research. Certainly, the criticism that the tools used to evaluate product placement effectiveness are still relatively unsophisticated and hindering the advancement of knowledge most likely relates to researchers assuming explicit measures are suitable in all instances and not recognising that  O              P     !  " # $% &  '    (-+( to measure certain objectives, implicit measures may be more relevant. This might explain why no explicit memory tests have found effects on brand attitudes or purchase intention (e.g. Karrh 1994; Ong and Meri 1994; Vollmers and Miserski 1994; Baker and Crawford 1996).

2.8 – Research Gaps After reviewing the literature, four aspects of empirical work on product placements are worth highlighting. These have driven the design of this research and justify its importance to the continued development of this research area. Firstly, only a few studies exist, especially relative to the level of real-world commercial activity. Secondly, some of the empirical findings do not converge, meaning that there is a strong need for further validation and generalisation. Indeed, because of the relative “newness” of the research area, and the absence of a clear research focus or strategy, researchers have tended to conduct ad-hoc and disparate studies and not made it a priority to validate previous results. Furthermore, inferences about product placement fail to converge not only across empirical studies, but also across practitioner sentiments and insights from qualitative inquiries (Balasubramanian et al. 2006).

Thirdly, much of the empirical research to date concerning product placement has focused only on one or two executional variables (i.e. product placement characteristics) under the control of the sponsor and program creator (e.g. placement modality, prominence). In doing so, these studies have ignored how different product placement characteristics impact on consumer reactions and memory differently and how they interact with each other. Moreover, past studies have also ignored how characteristics of the audience member themselves may make them more or less likely to remember, counter-argue or be persuaded by the message. We cannot assume that all people watching the same program will process the same product placement in the same way. Indeed, product placement effectiveness may well be a function of individual processing styles, engrossment with the program and brand familiarity, but this is yet to be fully explored.

 O              P     !  " # $% &  '    (-,( Finally, previous studies generally restrict attention to a few measures of message effectiveness (namely recall), meaning that the research emphasis has remained uneven across the effectiveness variables that could be relevant (Gupta, Balasubramanian and Klassen 2000), and has often used an inappropriate form of memory test to test for an impact.

What appears to be lacking is a strong conceptual foundation and a clear understanding of the possible effects and viable tools to measure these. There remains no theoretical framework that describes how brandcasts are processed. Do we know what factors affect the processing of brandcasts? Does this processing differ between individuals, even those exposed to the same placements in the same program? Do all brandcasts have the same effects? The answers to all these questions is likely to be no, yet researchers and practitioners have almost blindly run ahead without properly considering these issues.

This chapter has in part already addressed some of these issues. It has established a strong definition of product placement and parameters that distinguish product placement from other related activities. It has also identified possible effects and discussed the range of methods to appropriately measure them. However, there still needs to be a better understanding of the processing that underlies brandcast effects, in particular, the interaction between brandcast characteristics and audience characteristics.

The rest of this thesis attempts to address this issue by developing a conceptual model of brandcast processing and ensuring that the factors identified in this model are operational. This study is anchored from the perspective of the individual audience member to show that each individual processes the same brandcast differently depending on certain characteristics. To do this, a new scale to measure an individual’s engrossment with an entertainment story will be developed. Indeed, the shortcomings of previous research may well be in part due to the absence of the conceptual foundations and measures that the remainder of this research is seeking to develop and promote.

 O              P     !  " # $% &  '    (--( 2.9 – Conclusion This chapter provided an analysis of the current state of product placement research, and has demonstrated that research in the area needs to be continually updated to capture the dynamic nature of the technique and the changing practice in its use. Several gaps and shortcomings of present product placement literature have been identified, namely the practice’s poor conceptualisation and the lack of focus on the role of the audience member in making a product placement (brandcast) successful. Chapter 3 offers a framework of audience processing, and examines one of these characteristics in more detail – the level of engrossment an audience member has with the entertainment they are consuming. It also furthers the argument that the success of a brandcast lies with the audience. Indeed, the question of brandcasting should be not what brandcasts do to audiences, but what audiences do with them (DeLorme and Reid 1999).

 O              P     !  " # $% &  '    (-.( CHAPTER 3: ADVANCING PRODUCT PLACEMENT RESEARCH

“Counting eyeballs is not enough, because product placement is so subjective. It is taking a look at the context of how a product is used to determine the impact of whether a consumer noticed it, whether it changed their perception of the brand, whether it influenced their decision on whether to purchase the brand. Those questions haven’t been asked before."

(Sauer 2004, p4)

3.1 - Introduction Before useful research and effective practice can occur, and valid and reliable ways to measure the value of brandcasts developed, the factors that affect their processing need to be identified, and a model of brandcast processing developed. These tasks are all achieved in this chapter. Chapter 2 concluded by arguing for a greater understanding of the role of the audience in the success of a brandcast, especially since most previous research regarding product placement has focussed on examining the impacts of executional aspects of product placement quality, and not the characteristics of the audience member. This chapter furthers the argument for a focus on the individual. It re-introduces the concept of ‘audience engrossment’ and promotes the development of a specific Audience Engrossment scale since there are no existing scales which describe an individual’s interaction with an entertainment program

3.2 – The Need for a Better Conceptual Model As noted in Chapter 2, a solid conceptual framework to direct product placement research is yet to be developed. Recently Balasubramanian, Karrh and Patwardhan (2006) developed a model framework which they claimed filled this gap (see Figure 3.2.1). Whilst their framework does attempt to identify a range of components that an ideal model should

 O              P     !  " # $% &  '    (-/( include, and also considers the role of the audience member, it is difficult to operationalise and as discussed next, it cannot really be considered a processing model.

Any processing model reflects a flow of effects from an initial stimulus, with this relationship being mediated or moderated by any number of factors. The problem with the Balasubramanian, Karrh and Patwardhan (2006) model is that it does not start with this stimulus, but rather a group of execution factors (which may or may not be descriptions of the stimulus or factors inherent in it) and a group of individual-difference factors (which may be better considered as moderators or mediators – and are therefore depicted in the model incorrectly).

Figure 3.2.1 – Balasubramanian, Karrh and Patwardhan’s (2006) Proposed Model Framework

Expanding on these observations further, many of the items this model terms “execution factors” (which they claim are stimuli-based) are not actually inherent in the product placement itself. These incorrectly classified factors are program type, induced mood,

 O              P     !  " # $% &  '    (-0( execution flexibility and priming. Whilst they are all important and need to be considered when managing the placement’s execution, they cannot be used to describe the placement itself or classify the parameters of the stimulus in the same way that other elements such as modality and opportunity to process can. Execution flexibility and priming (as defined by the authors) may be better described as factors external to the product placement itself that may enhance its usability and effectiveness respectively. Furthermore, program induced mood is not fully under the control of the advertiser or even the entertainment content producer, as each audience member may process the story and the product placement differently. It is not fixed like the other components are and may actually be better classified as an individual-difference factor.

Secondly, it is not actually possible to measure “processing type” – one just presumes it happens in order for these effects to occur. Therefore this box is redundant. The significant issue that processing type relates to is identifying the outcome and using the correct test to measure that outcome (see Section 2.7.3). That said, this model not only offers a shopping list of all the potential factors that could impact processing, but as discussed in Section 2.7.3, it proposes eleven different effects, many of which are difficult to measure or are not very useful. In this way, the model becomes un-manageable and un-operational. Furthermore, by featuring all these effects in the same box, no hierarchy of effects or natural flow of effects is depicted, thus suggesting that all effects are equally likely.

A better approach may be to begin with a more basic model that contains the “essential” factors. This model would describe the stimulus (i.e. those factors inherent in the product placement) and flow through to possible effects, with only a core set of factors moderating this relationship. If all the factors in that model appear to have an effect, other variables may then be added to see their impact. As discussed in Section 2.7.3, the first effect that should be tested for is an initial memory trace which may then lead to formed (i.e. awareness) or altered (i.e. preference) perceptions. It should be a priority to test for these effects first before expanding the model too widely.

 O              P     !  " # $% &  '    (-1( 3.3 – Developing a Model of Brandcast Processing Brandcasting effects are dependent on the audience processing the embedded brand, and this occurs simultaneously with their processing of the entertainment story content. When watching programs, limited processing capacity must be shared by the storyline and peripheral information such as embedded placements, with the main focus going to the program. Hence, processing of these embeds requires parallel information processing. Therefore, the issue is to determine how factors related to the processing of an entertainment program affect the concurrent processing of a brand featured within it. The answer may depend on the relative balance of three potentially countervailing forces – the program context, the brandcast itself, and the characteristics of the individual audience member. It is this concurrent processing and interaction of numerous factors that makes brandcasts qualitatively different to other forms of advertising and promotion and an interesting avenue of academic inquiry.

To understand the effects of a brandcast, it is necessary to understand how audiences respond to the activity, and specifically, how they process the brand exposure. Indeed, since the audiences, as processors, are part of the process, they must also be incorporated into the theory. However, until now, past product placement research has generally ignored audience characteristics. But audiences bring to each processing their own abilities and interests which are bound to interact with the brandcast quality. Some dominant and common factors are able to be identified and studied, such as the level of engagement with the program (necessary because a brandcast must be processed concurrently with all other elements of the program content) and the level of familiarity with the product and/or brand (since research shows that people pay attention to things that are of personal interest and familiar to them). Connectedness21 may also be relevant when studying these effects in television programs, but it is generally not relevant for other entertainment vehicles such as film (see Section 3.7). Therefore it is proposed that a basic brandcast processing framework

21 Connectedness refers to the intense relationship between the audience and a television program that extends beyond the television watching experience into an individual’s personal and social life (Russell et al. 2004). It will be discussed in further detail in Section 3.7 and was introduced in Section 2.6.  O              P     !  " # $% &  '    (-2( (which is equally relevant to both product placements and advertainment) would include: the quality of the brandcast; the quality of audience engagement (engrossment); levels of product/brand familiarity; and depending on the context, connectedness, as these are key factors that will impact every processing episode. Furthermore, basic to any model is the identification of the possible outcomes and how these may be conceptualised, so these must also be established and feature in the model (see Figure 3.3.1)22. Other factors may be added to this basic model later23.

Figure 3.3.1 – Basic Theoretical Model of Brandcast Processing

Brandcast Quality Memory Brand Modality Awareness Prominence Conscious / Star presence /usage Unconscious Preference Method of Depiction Quality Temporal Quality Plot Connection

Audience Product / Engrossment brand familiarity

Connectedness Feelings Arousal Cognitive effort Appraisal

22 For this study, audience characteristics such as age, education, gender and media usage and frequency were controlled for and compared. Since delivery context was limited to film, and product placements contained within them as the stimulus, any results will only hold for this context. Future research should test different contexts in order to develop generalisations. 23 Other factors not mentioned here which may impact brandcast memory and brand preference include the audience member’s level of consumerism and materialism (e.g. Moschis and Churchill 1978; Richins and Dawson 1992), skepticism to advertising (e.g. Obermiller and Spangenberg 1998), attitudes to the practice of product placement (e.g. Nebenzahl and Secunda 1993; Gupta and Gould 1997; Gould et al. 2000; Gupta et al. 2000; Karrh et al. 2001; McKechnie and Zhou 2003), attitudes to a particular product placement and executional style (e.g. humorous, comparative). That said, these were omitted from this study because they can all be operationalised using existing tools. This thesis focuses on developing the new audience engrossment measure. Measuring all these other factors as well would have made the task un-manageable.  O              P     !  " # $% &  '    (.3( 3.4 – Measuring Effects As outlined in Chapter 2, a significant issue for brandcasting is determining what audience effects are possible. This debate is magnified as there is little consensus or knowledge as to what effects practitioners are hoping to gain from brandcasting, what effects could theoretically arise, and what effects have actually been demonstrated through empirical studies.

In traditional advertising, exposure-effects measurement has been conducted from an information processing perspective structured around the Hierarchy of Effects model (Lavidge and Steiner 1961). While most hierarchy of effects research focuses on advertising, Barry (2002) suggests that it is relevant for all marketing communications tactics, so it should also apply to brandcasting. To determine what is useful to them, and worth paying attention to, audiences use selective perception. This in turn will govern their interest in, and preference for, the product that is placed. Therefore, brandcasting planners must carefully match the products they place with the likely target audience of the entertainment program in order to maximise awareness (memory) for the placed brand24. Indeed, a product placement or advertainment sends information to individuals in the hope of them having some conscious or unconscious retention of them and then changing or solidifying their preference or attitude towards the product, or reinforcing and/or legitimising the acceptance of the product or brand in society. So with awareness and preference considered desirable effects, the central issue for brandcasting research is to find a viable way to measure these effects

Our model of brandcast processing assumes that the creation of a memory trace is the fundamental goal of brandcasting and allows for the measurement of both conscious and unconscious memory. It also allows for two levels of effect to be measured – memory and preference. As discussed in Chapter 2, unless a memory trace is created the brandcast can be deemed to have failed. This memory trace can be conscious or unconscious, and the

24 This further justifies the need for an examination of brand familiarity as a moderating factor in this model.  O              P     !  " # $% &  '    (.*( nature of this trace can range from awareness to a change in product/brand perceptions (positive or negative). It is for these reasons that we have retention (conscious or unconscious) as the basic effect in our model of brandcast processing, with awareness and preference for the brand flowing on from that memory trace. Explicit tests capture conscious memory traces and implicit tests capture unconscious traces. Which test is used is directed by the research objectives.

This research will measure recognition as the memory outcome since it can indicate both conscious and unconscious processing as it is often used to identify sensory memory content (see Baddeley 2007). Whilst a product placement does not have to be processed consciously, it needs to be retrieved to be effective. This retrieval is easier if conscious processing occurs. It is for this reason that Percy (2006) argues that to be effective, the product placement (as with any other effective marketing communication) must be attended to consciously, and positively linked to appropriate associations in non-declarative emotional memory and explicit memory. Given that increased brand awareness is one of the primary goals for product placement, explicit memory-based measures have been further justified as an appropriate way to test effectiveness, based on the premise that effectiveness measures should conform to the goals of the advertiser (Rossiter and Percy 1998; Nelson 2002).

Another reason for measuring recognition is that Singh, Rothschild and Churchill (1988) suggest that testing for recognition (as opposed to recall) is “enough” – that developing learning and memory to the point of achieving good recall results (the more difficult task of the two) is unnecessary for most brand choice decisions. Recognition is also a more sensitive measure and results generally show greater variance and discrimination than do results from recall tests (Singh and Rothschild 1983; Singh et al. 1988).

3.5 – Brandcast Quality Any pre-existing attempts to establish a typology of product placement were based on executional elements (e.g. Gupta and Lord 1998; Russell 1998; d'Astous and Seguin 1999).  O              P     !  " # $% &  '    (.+( This research study considers the role that six different factors of brandcast quality have on memory for the brandcast – modality; prominence; star presence and use; method of depiction; temporal quality; and plot connection. Because it is the brandcast that the audience member is exposed to, the quality of this stimulus (resulting from its mix of characteristics) is depicted as having a direct effect on the audience’s stages of processing as depicted in Figure 3.3.1. The specific effects that these particular elements had on memory and attitudes in previous studies were discussed in Section 2.6 and will only be revisited here briefly. In short however, we expect all six of these factors to have positive impacts on recognition (see Hypotheses 1 and 2 in Section 6.2).

3.5.1 - Modality Modality refers to how the brandcast is communicated to the audience. There are three main ways this can occur. Visual-only placements show a product, logo, billboard, or some other visual brand identifier without any relevant message or sounds. In contrast, audio- only placements exist when the brand name is mentioned or heard without showing the product. Audio-visual placements show a brand at the same time as mentioning the brand name or conveying a brand related message (Gupta and Lord 1998).

Combined audio-visual placements generally accrue higher recall than single mode placements (Steortz 1987; Sabherwal et al. 1994; Russell 1998; Law and Braun 2000). However Gupta and Lord (1998) found no significant difference between placements that were audio-visual compared to those that were only visual. The results comparing visual- only and audio-only placements are even more conflicting. Hence, further investigation exploring whether any differences in conscious memory exist is necessary to aid product placement planners in making better decisions. However, we would expect that dual modality will enhance product placement recognition.

3.5.2 - Prominence Prominence refers to the extent to which the product placement possesses characteristics designed to make it a central focus of audience attention. Creative (or subtle) brandcasts  O              P     !  " # $% &  '    (.,( occur when a brand appears as a prop in the background of a shot, is small in size, or is lost in an array of multiple products or objects (Murdock 1992). In contrast, on-set brandcasts are displayed more prominently, by virtue of their size and/or position on the screen. Both unaided and aided recall (Gupta and Lord 1998) and recognition (Brennan et al. 1999) for on-set / prominent placements have been found to be significantly higher than for creative / subtle placements, and this is the result we expect to find in this study. Such findings endorse the film studio practice of charging a premium for prominent placements (Darlin 1995).

3.5.3 – Star presence and star use Prominence may also result from the fact that the product (or other brand identifier) is consumed or mentioned by one of the leading actors (Babin and Carder 1996). Such a brandcast is termed ‘use by star’ (Scott and Craig-Lees 2004). A lesser form of use by star is star presence, which we define as when the product or brand is in the same shot as the star. Star presence is limited to audio-visual media such as television, film, plays, music videos and video games, since presence cannot really be captured in a novel or magazine.

Given that much of the product placement literature suggests that the audience associations linked to the placement are an important feature in memory storage and attitude formation, then the impact of use by the star (the celebrity endorser) needs to be considered (see Section 1.1). In doing so, this research seeks to address limitations from Brennan, Dubas and Babin’s (1999) study by investigating the specific effect that product use by a star or the mere presence of a star in the same scene as a product may have on memory, whilst considering the respondent’s level of preference for that star. Whilst d’Astous and Chartier (2000) found that star presence had positive effects on both recognition and recall, Scott and Craig-Lees (2004) found no significant effect of use by star on recognition. That said, given the strong theory underlying product placement (e.g. social learning theory, role modelling), we would expect both star presence and star use to enhance product placement recognition.

 O              P     !  " # $% &  '    (.-( 3.5.4 – Method of Depiction In keeping with the definition of brand, and as highlighted by Rössler and Bacher (2002), brandcasts can also feature services, ideas, countries, personalities and music, as well as products. Taking a broad perspective, there are two main types of brands that can appear. A corporate brand, for example, a Prada logo on a shopping bag can appear in a scene (such as in Legally Blonde), or be described in text. Alternatively, a specific product bearing the name or unique brand identifier may appear in some way (for example, in Legally Blonde, one character says to another - “Don't stomp your little last season Prada shoes at me, honey”). These more specific brandcasts may not only improve brand image and awareness but also encourage use and we expect them to enhance product placement recognition. Indeed, a product’s appearance in a movie represents a brand contact that can be used to shape impressions and ultimately motivate consumer action. Thus, positive product portrayals in movies can contribute to a consumer’s decision to use a product, while negative or unflattering portrayals can discourage use (Morton and Friedman 2002).

Whether a specific brand is featured or not, it must be kept in mind that the product category as a whole is also being promoted as a result of the brandcast. Whilst brandcasting is most often referred to as creating favourable effects on the specific brand that features, it may also be viewed as a primary demand strategy, designed to increase the level of demand for a product form or class. Such a strategy is particularly advantageous to market leaders such as Coca-Cola, who have the most to gain from expanding the size of the market, as they are inclined to capture the largest share of any new customers. Alternatively, depicting product category usage in a film may encourage present users of the category to increase their current rate of usage.

There are a number of ways that a product or brand can feature in a story line (which to date, have not been examined separately). The actual product could appear, or an advertisement, brand name or logo may feature (DeLorme and Reid 1999). Such appearances, depending on the media they feature in, and their function to the plot, may appear visually, aurally, audio-visually, prominently or subtly.  O              P     !  " # $% &  '    (..( 3.5.5 – Temporal Quality The fifth characteristic relates to the temporal quality of the brandcast, and is comprised of two (somewhat related) components which we expect to enhance recognition.

Frequency of exposure is measured by the number of times the product category or brand appears throughout the story. Social learning theory suggests that the more frequently images of product consumption, activities and lifestyles prevail in films, the more likely that modelling of such behaviours will prevail (Bandura 1977). Furthermore, repetition helps slow down the process of forgetting or decay, and increases the ability to retrieve information from the memory store (Schiffman, Bednall, Watson and Kanuk 1997).

Total length of exposure is the cumulative time of all brandcast exposures. Quite simply, the more screen time a product is given, the greater the chance it has of being noticed and subsequently recalled. Indeed, the amount learnt is a direct function of the time devoted to learning (Baddeley 1990). Because of this, product placement contracts routinely include a minimum airtime clause (Turcotte 1995). That said, Gupta and Gould (2007) failed to find a significant impact of exposure time on either aided or unaided recall of prizes shown during game shows

3.5.6 – Plot Connection Plot connection relates to the degree that the brandcast is integrated with the plot of the story (Russell 1998). Whereas lower connected plot brandcasts do not contribute much to the story, higher connected plot brandcasts constitute a major thematic element (Holbrook and Grayson 1986).

Research shows that product placement only works if it is in context; if it is used inappropriately, viewers will be aggravated, so a brand or product must fit naturally into a scene to be successful and not break the audience’s focus. A recent Australian example of poor execution was the second season of The Block, where one episode had 143 product logos and brand names, with speculation that the drop in ratings for that series was because  O              P     !  " # $% &  '    (./( of product placement overkill (Taffel 2004). This is consistent with research by both Russell (1998) and d’Astous and Chartier (2000) who found that integration and congruence of the placement with the plot had a positive impact on consumer liking and acceptance of the placement (but a negative impact on recall).

Whilst conceptually the idea of plot connection is easy to understand, operationalisation of the concept has not been reported by Russell (2002) or d’Astous and Chartier (2000), so we developed our own method to determine a product placement’s degree of plot connection which related to how essential it was to the scene. To determine this, we asked the question – if the product was not in the scene, could the scene still exist in the same way? For example, a car chase scene could not exist without cars, but two characters having a conversation as they walk passed some parked cars would reflect a less essential placement, even if the placement was prominent or there was a close-up of the car’s badge. Using this operationalisation, plot connection became less associated with prominence. Based on its aforementioned operationalisation, and the strong linkage of a brand to the plot, we would expect plot connection to have positive effects on recognition.

3.6 – Audience Characteristics Consumption objects, including entertainment vehicles, are typically consumed in a variety of ways by different groups of consumers (Holt 1995). The uses and gratifications research stream is an audience-centred perspective that builds on this idea, emphasising individual choice in explaining media effects. It looks at what people do with the media rather than what the media does to them (Wakefield, Flay, Nichter and Giovino 1998). An essential assumption of this perspective is that mass media use is controlled by the individual audience members, thus taking a user-level view as opposed to a mass exposure perspective in understanding media use (Stafford, Stafford and Schkade 2004). In this model, the audience selects and uses content that will best meet their needs, suggesting that media behaviour can be purposeful and goal directed, and that the same program may gratify different needs in different audience members. Similarly, levels of utility, intention, selectivity and involvement vary when media are encountered by people under different  O              P     !  " # $% &  '    (.0( circumstances (Blumler 1979; Kim and Rubin 1997; Wakefield et al. 1998). Therefore, it seems sensible to argue that the memory of a brandcast will be moderated by an audience member’s characteristics.

3.6.1 - Level of Product / Brand Familiarity There is empirical evidence for the role of brand familiarity in relation to formal advertising messages (e.g. Alba and Hutchinson 1987; Rao and Monroe 1988; Coupey, Irwin and Payne 1998), but not in the context of product placement in films, except for Scott and Craig-Lees (2003) and Brennan and Babin (2004). Brand familiarity provides consumers with a superior ability to accumulate, integrate and judge the relevance of product information, thereby creating a sophisticated memory schema (Alba and Hutchinson 1987; Rao and Monroe 1988). With familiarity affecting what, how and how much brand information is retained, audiences are more willing to process personally relevant messages and find it easier to process information about products and brands that are familiar to them and then recall them or discriminate between similar products, two very important tasks relevant to this research (Petty et al. 1983; Celsi and Olson 1988; Babin and Carder 1996; DeLorme and Reid 1999). In the context of brandcasting, the perceived personal relevance may allow faster recognition and facilitate concurrent processing of the embedded brand, hence impacting on the direct relationship between the brandcast characteristics and retention.

There is some evidence that audiences pay attention to, and remember brands placed in media programs. DeLorme and Reid (1999), in a qualitative study exploring how moviegoers interpret brands placed in movies, found that "moviegoers were particularly attuned to familiar branded products and services that they themselves had previously purchased and consumed in their everyday lives" (DeLorme and Reid 1999, p78). This finding also has roots in DeLorme, Reid and Zimmer’s (1994) earlier qualitative study which found that moviegoers notice and like familiar brands in movies, and that product placement led to relation with characters due to shared brand use.

 O              P     !  " # $% &  '    (.1( That said, it must be noted that there are competing theories for how brand familiarity may influence memory. The other theory suggests that perceived novelty, including the unusual nature of a brand, may cause it to stand out (Wallace 1965). Indeed, the Von Restorff effect (Wallace 1965) suggests that since unfamiliar or unexpected stimuli are incongruent with prior expectations, they attract greater attention and produce superior cognitive outcomes (e.g. recall) than familiar stimuli.

However, the stance taken in this research is that outlined initially - that familiar brands are easier and quicker to process and retrieve (see Hypothesis 3 in Section 6.2). They also facilitate audience identification with characters in the program. In other words, placements involving familiar brands are more diagnostic to viewers in terms of quickly understanding complex meanings in program content (Balasubramanian et al. 2006). Research shows that accessible attitudes play a role in directing attention to certain objects in the visual field and that those accessible attitude objects automatically attract attention (Roskos-Ewoldsen and Fazio 1992). Prior research considering this in a product placement context found that familiar brands maintained higher recognition levels than unfamiliar brands in studies of product placements in movies (Scott and Craig-Lees 2003; Brennan and Babin 2004) and video games (Nelson 2002; Nelson, Yaros and Keum 2006). However, Redondo (2006) found the targeting of product placements in movies to be poor (i.e. brands placed in the film were not necessarily known to the audience watching that film), except for those with high plot connection. We attribute that to better planning for those placements.

3.6.2 – Level of Audience Engrossment The variable use of entertainment media makes audience measurement an especially difficult task. But for many reasons (such as the buying and selling of media space), it is important to distinguish audiences in terms of the intensity of their relationships with the media. Yet understanding must come before measurement (Clancey 1994). Simply reporting on audience sizes implies that all viewers relate to media in the same manner, and are therefore equally affected by what they see (Clancey 1994; Russell and Puto 1999).

 O              P     !  " # $% &  '    (.2( Similarly, the analysis of a program’s content does not allow us to predict with certainty how that program will be interpreted by different individuals (Livingstone 1990).

Watching is a subjective term, that when applied in the real-world setting, is open to many different interpretations by those who measure, those who are measured, and those who use the information obtained from the measure. Indeed, a measurement system that uses “watching” as its operational definition will allow two people who are in the presence of an entertainment program to be reported as equal, regardless of their level of engagement with the program (Clancey 1994). Simply because a person is counted as being in the audience for a particular program does not mean that any given advertisement, product placement or program segment has been seen. The environment in which an ad appears may have a significant effect on the viewer, and this effect may be caused by viewer attitudes towards the vehicle, mood engendered by it, the degree of involvement of the audience member and how the message content is processed (Lloyd and Clancy 1991; Norris and Colman 1993).

Understanding the impact of context and audience state on brandcast effectiveness is important and raises two issues that this research seeks to address. One is the theoretical issue – the role of the viewer in the success of a brandcast. The second is a methodological problem – how can we investigate different reactions to the same content and how might these cause different effects? This is particularly significant given that a viewer’s engrossment with a program’s content may influence the effectiveness of its embedded placements (Bhatnagar et al. 2004).

As a variable, audience activity needs to be defined, operationalised, tested and refined (Livingstone 1990). However audience activity is multi-dimensional and audiences can exhibit varying degrees of activity both within and between programs. Intensively focused viewing is only one way of watching. Most people have different intensity levels of viewing which vary with the content of what they are watching, what else they are doing, time of day, motivation for viewing, and who else is watching with them (Lee 2002). For

 O              P     !  " # $% &  '    (/3( this reason, it may be more accurate to think of watching not as a binary condition, but a continuum (Kim and Rubin 1997; Lee 2002).

Perhaps a better term to capture this idea of ‘viewing intensity” or “activity” is audience engrossment. Engrossment means to occupy exclusively; to absorb, to give complete attention, concentration and intense mental effort; it is the mental state of being preoccupied by something. Synonyms include captivation, concentration, absorption, enthrallment, fascination, immersion, intentness, involvement, raptness and preoccupation. Applying this to media processing and interaction, audience engrossment represents the degree to which individuals are engaged - affectively, cognitively and behaviourally - with the entertainment content they are consuming, at the time of consumption (Scott and Craig- Lees 2005a, p2).

3.7 - Inadequacy of existing concepts and measures which could apply to Audience Engrossment As reflected by its definition, audience engrossment is a multi-dimensional and holistic measure describing an individual’s emotional and cognitive engagement with entertainment content. However extensive reviews of the literature revealed that there was no adequate existing measure that could be applied to this new audience engrossment concept, although some did measure related concepts. These include involvement (e.g. Zaichkowsky 1994), pleasure and arousal (e.g. Mehrabian and Russell 1974), program liking (Murry et al. 1992), flow (Csikszentmihalyi 1990), transportation (Green and Brock 2000) and connectedness (Russell et al. 2004). Each of these will now be discussed in turn.

The term involvement is used to denote an 'interest' in something and results from an individual's inherent needs, values and interests (Greenwald and Leavitt 1984; Zaichkowsky 1985). However, for it to be a truly operational and more useful construct, quality and context needs to be considered. In marketing, the involvement concept has been linked to purchase behaviour (enduring and situational) and to the processing of

 O              P     !  " # $% &  '    (/*( advertisements, and scales have been developed to assess these forms of involvement. However, since these scales were developed to capture ‘interest’ in a different context, they are not equipped to capture the audience engrossment construct (and other related concepts such as transportation, flow and connectedness). Indeed, audience engrossment is multidimensional and complex, and represents how a person might interact with and react to, both emotionally and cognitively, entertainment content.

In marketing, the most cited involvement scale is Zaichkowsky's (1985) Personal Involvement Inventory, which captures enduring involvement using semantic differential scales. Although Zaichkowsky asserts that this scale is context free and appropriate for measuring all sorts of involvement (product, advertising and purchase), this scale is geared towards involvement with a product and not towards the advertisement (or in our case, the storyline) that is communicating the message or visually portraying the product. It is therefore not necessarily transferable to engrossment with a storyline (whatever the platform, whether it be a film, sitcom, novel, etc). Moreover, items such as 'needed/not needed' and 'essential/inessential' are irrelevant for describing one's engagement with a specific storyline or plot (although they may be useful for describing one's relationship with a particular medium).

Involvement with advertising messages has been examined, but these have been examined via proxy measures such as program context, program-induced viewer mood, program- advertisement congruity, program-induced viewer excitement, attitude / liking of the program, program-induced viewer drive for closure; program-induced emotional arousal or pleasure; program impact or appeal; and program-induced viewer involvement (Gunter, Furnham and Beeson 1997). Krugman (1967) measured involvement as an individual’s direct personal experience during message reception. However, brandcasts do not interrupt the programs in which they are embedded, and nor do they aim to give direct brand information, so there is a critical difference between them and commercials.

 O              P     !  " # $% &  '    (/+( The Mehrabian and Russell (1974) Pleasure, Arousal, Dominance scale measures emotional response, and has been used extensively in regards to assessing emotional response to situations. However, it does not capture any information regarding cognitive response. Furthermore, in the context of experiencing an artistic event such as listening to music, Ritossa and Rickard (2004) suggest that the circumplex theory of emotions more fully captures the emotional response to an aesthetic experience.

Murry, Lastovicka and Singh (1992) focused their study on program liking, assessing the summary evaluation of the viewing experience and the intrinsic satisfaction that viewers derive from viewing. However, this Program Liking scale was not deemed wholly appropriate for capturing engrossment since it focuses on liking and does not capture the arousal or cognitive aspects that we believe comprise audience engrossment.

Other concepts, not yet used in a marketing communication context include flow theory (Csikszentmihalyi 1990), which is based on a symbiotic relationship between challenges and the skills needed to meet those challenges. The flow experience is believed to occur when one’s skills are neither over-matched nor under-utilised to meet a given challenge. When this balance is disrupted, apathy (i.e. low challenge, low skill), anxiety (i.e. high challenge, low skill) or relaxation (i.e. low challenge, high skill) are likely to be experienced (Csikszentmihalyi 1997). But when the challenges encountered in an environment are matched above some critical threshold to a person’s ability, that person feels more active, alert, concentrated, happy, satisfied and creative (Csikszentmihalyi and LeFevre 1989). Although flow and engrossment both describe a state of intense concentration or deep absorption in an activity that is intrinsically enjoyable, the interaction quality is distinct. The role of challenge and skill is not relevant to consuming most entertainment (except for video games and advergames), and the fact that flow invokes a growth principle whereby a more complex set of capabilities is sought after and developed means that it is qualitatively different to engrossment (Shernoff, Csikszentmihalyi, Schneider and Steele Shernoff 2003). Indeed, the application of flow to the consumption of

 O              P     !  " # $% &  '    (/,( a predetermined story would be dubious as audiences do not necessarily seek more complex movies to watch as their experience with movies increases.

Another high level construct similar to flow (Csikszentmihalyi 1990), presence (Lombard and Ditton 1997) and escapism (Mathwick, Malhotra and Rigdon 2001) is transportation, a tripartite formulation (attention, imagery and feelings) of persuasive communication that sees the reader lose access to some real world facets in favour of accepting the narrative world that the author has created. It can occur on both a physical and psychological level (Green and Brock 2000; Green, Brock and Kaufman 2004). As with flow, individuals who are transported are fully concentrating on the task (the story) and often lose track of time or fail to notice events occurring around them (Green et al. 2004). The authors claim that transportation into a story causes people to be less motivated (or less able) to disbelieve any particular conclusion, becoming so absorbed in the story that they are unlikely to stop and critically analyse propositions presented therein. Furthermore, since stories are generally presented as entertainment rather than as vehicles for attitude change, they provide few explicit triggers for critical thinking and counter-arguing. For example, attachment to a protagonist may be an important determinant of the persuasiveness of a story, with the experiences or beliefs of those characters having an enhanced influence on readers’ beliefs (Green and Brock 2000).

On face value, transportation describes a mental shift from reality to the feeling that one has actually entered this narrative world. Therefore, transportation may be an extension of engrossment but it is not in itself equivalent to engrossment. Some individuals, while engrossed (i.e. carrying out intense processing of the program) may also become ‘one’ with the story (i.e. be transported), whilst others may engage in intense processing but not be transported. For these reasons, audience engrossment may actually be an antecedent to transportation, and even flow.

However, examination of the items comprising transportation suggests that the similarities between engrossment and transportation might not be so strong, with three key differences  O              P     !  " # $% &  '    (/-( identified. The first is to do with the different story-telling environments. Transportation has only been tested in literature, and its scale items reflect this written text skew, with items attempting to capture whether respondents could create images of the events taking place in the story and images of the characters. Film and television provide a different storytelling environment, providing this rich visual imagery to its viewers. Green, Brock and Kaufman (2004) acknowledge this, claiming that further exploration of how we interact with different media is a potentially fruitful direction for empirical research. Secondly, many of the transportation items focus on elaboration of the story (e.g. ‘I found myself thinking of ways the narrative could have turned out differently’, ‘The events in the narrative have changed my life’). Engrossment does not aim to capture these thoughts, but rather just how the respondent reacts whilst consuming the entertainment, not how they feel or what they do afterwards. Finally, the items comprising transportation do not capture the breadth and depth of emotion and cognition that engrossment does (despite claims to the contrary). Transportation has single items measuring whether the ‘narrative affected me emotionally’ and ‘I was mentally involved in the narrative’.

One concept closely related to engrossment (and which appears in the conceptual model in Figure 3.3.1) is Russell, Norman and Heckler’s (2004) idea of connectedness. They use the term primarily in the context of television programs and define connectedness as the intense relationship between the audience and a television program that extends beyond the television watching experience into an individual’s personal and social life. This relationship can be seen to lie on a continuum from no involvement to fanaticism, and is manifested by identification with characters, commitment to the show, and finding the show personally relevant and involving (Russell and Puto 1999). Extending the usefulness of this scale to product placement, Russell et al (2004) found that peripheral elements of the story and set (i.e. product placements) were noticed by highly connected viewers. Furthermore, these viewers may not necessarily perceive any commercial intent behind brand usage in the program.

 O              P     !  " # $% &  '    (/.( The fundamental difference between engrossment and connectedness is that engrossment captures an audience’s reported state during media consumption. Connectedness relates to the more global enduring feelings felt by the audience pre- and post-exposure and about the program in general. Hence, connectedness may be a measurement factor in engrossment, as the degree of connectedness can influence how an audience approaches the consumption of a program and may function as an antecedent of liking and mood state. Alternatively, audience engrossment may over time, and after multiple exposures, lead to connectedness, especially towards a star, television program, or movie franchise. This is why in our model, there is a two-way arrow connecting connectedness and audience engrossment.

Not only can audience engrossment lead to, or be caused by connectedness, but it may also lead to, or alter, mood state. Mood refers to a transient state of mind that whilst generally not tied to a specific event or object, can both affect, and be affected by, the consumption experience, thus becoming the outcome of an emotional experience (Gardner 1985). In relation to audience engrossment, the emotions (i.e. feelings and arousal) generated by the story can either sustain or alter a particular mood state felt by the audience member, with this mood state likely to last beyond consumption of the story. Mood however is different to connectedness as connectedness, whilst more enduring, is directed only towards the program and/or stars, whereas mood is omnipresent.

Linking connectedness back to the basic idea of involvement, connectedness could be considered akin to enduring involvement as it represents an ongoing interest in a product category or brand (or in this case, a television show). However, connectedness reflects an extreme interest, where one is highly committed and loyal to that program and/or those stars. Continuing this analogy, audience engrossment could be considered similar to situational involvement given that it captures a person’s interest in a program only for the duration of that program (much like situational involvement only lasts for the duration of the decision making process or information search).

 O              P     !  " # $% &  '    (//( Scott and Craig-Lees (2003) considered the role of both emotional and cognitive involvement and star liking on product placement recognition, using a combination of scale items from Mehrabian and Russell (1974), Perse (1990; 1998), Rubin, Perse and Taylor (1988), and adapting Murry, Lastovicka and Singh’s (1992) Program Liking scale to measure star liking. Their research confirmed the findings of Singh and Churchill (1987) who found that program involvement, namely pleasure and cognitive effort, had a positive effect on recognition. Cluster analysis also confirmed that when all types of involvement were present (i.e. pleasure, arousal and cognitive effort), as well as star liking, recognition was at its highest. Relative to pleasure, this effect was consistent with the pre-existing literature (e.g. Isen 1984; Pavelchak, Antil and Munch 1988). However, prior research would suggest that higher levels of cognitive effort expended on the movie would have a negative impact on recall (e.g. Easterbrook 1959). Although these aforementioned scales may be well accepted and tested in their original contexts, it is likely that they are not specific enough to capture the complex nature of audience engrossment, thus explaining the conflicting results of many previous studies (e.g. Singh and Churchill 1987; Pavelchak et al. 1988; Tavassoli, Shultz and Fitzsimons 1995) and highlighting the need for a new, specific measure.

3.8 – Components of Audience Engrossment Audience engrossment needs to be a multi-dimensional construct. Specifically, there needs to be both an emotional aspect (feelings evoked, physical reactions, overall liking) and a cognitive aspect (the degree of mental engagement with the program - or at various times during the program). We hypothesise these four dimensions to be feelings, arousal, appraisal and cognitive effort. How we came to identify these four dimensions will now be discussed.

Emotions are an important factor in the advertising process and are now accepted as an important mediator of cognitive and behavioural consumer responses to advertising, especially since they function as the gatekeeper for further cognitive and behavioural reactions (Holbrook and Batra 1987; Poels and Dewitte 2006). Emotions are complex  O              P     !  " # $% &  '    (/0( reactions involving not only the intense subjective feelings we label as joy, anger, sorrow etc, but also outward emotional expressions such as laughing and crying (Baron 2001). It is not the specific events or physical circumstances that produce emotions, but rather, the unique psychological appraisal made by the person evaluating and interpreting the events and circumstances. Different people can have different emotional reactions (or none at all) to the same event or happening. Moreover, the individual’s own assessments of these emotions can be deliberative, purposive and conscious, or unreflective, automatic and unconscious (Bagozzi, Gopinath and Nyer 1999).

Researchers have operationalised emotions in two primary ways – level of arousal and polarity along the pleasure dimension (e.g. Mehrabian and Russell 1974). Arousal refers to the activity of the autonomic nervous system, and its experience is defined as a feeling state varying along a single dimension from drowsiness to frantic excitement (Mehrabian and Russell 1974). Besides this intensity, affective reactions can be characterised by their valence (i.e. the extent to which the emotion is pleasant or unpleasant). Indeed, this approach has a history of successful applications in consumer behaviour, and has been found by virtually all dimensionally-oriented researchers (e.g. Mehrabian and Russell 1974; Pavelchak et al. 1988).

In regards to emotion in this research, three concepts are identified – feelings, arousal and appraisal (see also Section 4.6.3). Feelings refer to the label given to the felt emotion25, whereas arousal refers to the subsequent physical reactions resulting from these feelings. For example, a person might feel sad (feeling) and this might manifest itself in the form of crying (arousal). High arousal levels (i.e. high intensity) have generally been found to disrupt information processing, particularly when the task is complex (Sanbonmatsu and Kardes 1988). Under these high arousal levels, performance on the secondary task (processing the product placements) deteriorates whereas performance on the primary task

25 In scales such as Mehrabian and Russell (1974), feelings are actually termed pleasure. But they were re- labelled here since many of the items actually refer to specific feelings (which may be positive or negative), and the focus was on intensity of feeling, not valence.  O              P     !  " # $% &  '    (/1( (processing the movie) is unaffected and in some instances, augmented (Eysenck 1982; Sanbonmatsu and Kardes 1988). As with the dual task paradigm, Easterbrook (1959) suggests that this is because the heightened arousal produces an attention narrowing process, restricting the range of cue utilisation. For these reasons, we would expect high levels of feelings and arousal to inhibit product placement recognition (see Hypotheses 4 and 5 in Section 6.2).

In this research, appraisal relates to liking the program, actors and genre, and is an evaluation of the pleasure felt from consuming the program. It has generally been accepted that people process information more efficiently and learn more material when in a positive mood than when in a negative mood because pleasant states facilitate learning by leading to the activation of broad, well integrated cognitive categories that enhance stimulus encoding and subsequent ad recall (Isen 1984). It may also occur because unpleasant states inhibit learning by decreasing motivation and inducing the interjection of negative thoughts (Pavelchak et al. 1988). Therefore, we would expect high appraisal to enhance product placement recognition (see Hypothesis 6 in Section 6.2).

Cognitive effort relates to the relative ease or difficulty experienced in making sense of, and following, a story. This cognitive processing aspect relating to the consumption of an entertainment program is very important, yet somewhat neglected (see also Section 4.6.3). In brandcasts, programs are longer and have storylines with varying degrees of complexity, so the processing of brandcasts is qualitatively different to advertising as it involves parallel processing. That our minds engage in parallel processing has been recognised since the 1960s. Influenced by the ‘shadowing’ or ‘cocktail party phenomenon’ described by Cherry (1953), the notion of unconscious processing captured the attention of cognitive scientists and marketers, notably the concepts of mind/brain multi-level and multi-task processing. There is still consensus that although we can engage in multiple tasks and multi-level processing, as more diverse material enters consciousness, there is a decrement in the amount of processing capacity available, so primary task cues are attended to at the expense of secondary task cues as a form of active coping strategy (Eysenck 1982; Worth and  O              P     !  " # $% &  '    (/2( Mackie 1987). Therefore, the effects of exerting high cognitive effort are similar to those of high arousal (see Hypothesis 7 in Section 6.2). Research by Tavossoli, Shultz and Fitzsimons (1995), Pavelchak, Antil and Munch (1988), Pham (1992) and others suggests that different levels of parallel mental activity can impact on the initial memory storage of specific bits of information. Because of limited cognitive capacity, as involvement or motivation to process (i.e. engrossment) reaches high levels, attention becomes more focused on the relevant sources of information (i.e. the movie’s storyline rather than the product placements) (Celsi and Olson 1988). Thus, highly engrossed people are no longer willing to (consciously) process irrelevant product placements when watching the movie.

It seems therefore that while engrossment due to increased cognitive effort with the movie and an increase in arousal and feelings during the movie may have a negative effect on recall or recognition, appraisal may, on the other hand, have a positive effect26. The complicating factor is that while feelings and arousal can be depicted as uncorrelated because they are conceptually orthogonal constructs (Russell, Weiss and Mendelson 1989), a positive correlation is sometimes observed between them (Pavelchak et al. 1988). This indicates that to treat involvement as a single component may not capture the full story, thus further justifying the operationalisation of audience engrossment as comprising of four separate dimensions: feelings, arousal, appraisal and cognitive effort (see also Section 4.6.3).

Looking at the audience engrossment concept more holistically, when content is boring, incomprehensible, or not providing the gratification that they seek, people may attempt to find something more interesting and mentally involving, thus suggesting that attention to a program is driven by the meaning individuals find in its content (Anderson and Burns 1991; Perse 1998). This means then, that when people are not interested in what they are watching (and audience engrossment is low), they may tune to something else (Perse 1998).

26 Specific hypotheses relating to these anticipated relationships are discussed in Section 6.2  O              P     !  " # $% &  '    (03( Relative to brandcasts, this may indicate that when viewers are not engrossed with the story they are watching, they may pay more attention to peripheral cues such as placed brands.

In order to address the question of whether different levels of engrossment with a story affect memory for product placements contained within them, audience engrossment needs to be quantified. Whilst there are a number of existing scales that capture aspects of audience engrossment, they fail to capture the full essence or complexities of this concept (see Section 3.7). They are either lacking in multi-dimensionality (i.e. capturing both emotional and mental aspects of processing), or are not suitable for a multi-sensory environment. Indeed, no single scale adequately measures all these components in the way that is needed to understand the ‘at-time’ consumption of an entertainment program. Therefore, a new measurement instrument specific to media consumption is necessary. As will be discussed in Chapter 4, despite some limitations, a pen and paper scale was believed to be the best way for this audience engrossment concept to be captured.

3.9 – Conclusion In this chapter, a model of brandcast processing was proposed and its factors identified and discussed. In particular, this chapter argued for the need for a new concept, audience engrossment, to capture media processing. It also demonstrated that due to the absence of suitable existing measures, a new measurement scale must be developed to do this. It is proposed that such a scale should comprise items that adequately measure the degree of cognitive effort, appraisal, arousal and feelings. Chapters 4 and 5 focus on the development and refinement of the Audience Engrossment scale, whilst the brandcast processing model undergoes preliminary testing in Chapter 6.

 O              P     !  " # $% &  '    (0*( CHAPTER 4: DEVELOPMENT OF THE AUDIENCE ENGROSSMENT SCALE

“What does it mean if a finding is significant or that the ultimate in statistical analytical techniques have been applied, if the data collection instrument generated invalid data from the outset?”

(Jacoby 1978, p90)

4.1 - Introduction Chapter 3 argued for the need for a measure of audience engrossment as there is no established scale that suitably measures an audience member’s processing of, or interaction with, an entertainment program during consumption. This chapter provides an overview of the process undertaken to develop this Audience Engrossment (AUDENG) scale, and the development of a verbal self-report scale and film as the research context are both justified. Measurement theory is introduced and shortcomings of Classical Test Theory discussed, with Item Response Theory (specifically the Rasch Measurement approach) presented as an alternative approach to developing this new scale. The process involved in generating the 81 items encompassing the four dimensions of the original AUDENG scale is then outlined. The chapter concludes with results from a pilot study to test the scale and a discussion describing the changes that were made before major testing of a 72-item scale commenced in cinemas (see Chapter 5).

4.2 – The Research Setting Film (and even more specifically, films shown at the cinema) was selected as the research setting since the cinema context provides the researcher maximum ability to control the physical environment and exploit its unique characteristics. It was therefore believed to be the best environment to capture audience engrossment in its most pure form. The cinema context is typically characterised by high levels of anticipation, interest and involvement; the only justification for the time and money invested in this environment is the  O              P     !  " # $% &  '    (0+( entertainment value of the movie watched (Gupta et al. 2000). Going to the movies requires far more effort than watching television as an individual must choose a movie and a cinema, drive there, find a parking spot, stand in line, pay money and find a seat. Once there, the particular exposure context associated with cinema screenings (i.e. lights off, minimal noise and distractions, large screen, difficulty in moving around, inability to zap through parts of the story) commands greater attention than watching television in one’s living room (d'Astous and Chartier 2000). At the cinema there is no competitive advertising at the same time (as opposed to TV, newspapers, magazines), and generally the viewer cannot do other things which may affect the degree of attention paid to the film.

A short-length film may have been a more suitable tool to assess audience engrossment and product placement recognition as it would minimise memory failures and provide a more accurate assessment of reactions and feelings evoked by the film. However, sourcing such a film that also had well known actors and plenty of product placements proved a difficult task, as well as having it screened in this cinema environment.

4.3 – Underlying Research Tradition and Strategy The research approach adopted for this thesis falls under the “behavioural and psychological research stream” (Simon, 1994, p27) since the concept of audience engrossment focuses on respondents’ thought processes, memory and interpretations of an entertainment program (e.g. television show, movie), constructs that are largely unobservable. One of the key decisions to be made was whether to use autonomic measures or self-report to measure audience engrossment. Autonomic measurements concentrate on continuous emotional reactions that are not distorted by higher cognitive processes, whilst self-report measures focus on introspective reflections about the emotions felt (Poels and Dewitte 2006). Whilst physical arousal can be captured via observation and autonomic measures, the interpretation of sensation and cognitive effort cannot.

Specifically, audience engrossment seeks to provide a description of an individual’s feelings, reactions and thoughts during a program. Since emotions are often accompanied by bodily reactions (e.g. facial expressions, sweating, increased heart rate), physiological  O              P     !  " # $% &  '    (0,( tools such as the Facial Action Coding System and facial electromyography (both of which code visible facial muscle movements and link them to specific emotions) and skin conductance (which measures activation of the autonomic nervous system and detects the level of sweat in the eccrine sweat glands) might appear to be suitable (Poels and Dewitte 2006). However, we felt that these tools would be unnerving and intrusive for most people and could affect their focus and enjoyment of the movie, even though they may be effective for detecting lower-order emotional processes (Poels and Dewitte 2006). Furthermore, since the intent was to develop a means of gathering data that can be used in field settings (such as cinemas in this study), such physiological techniques could not be implemented for this research.

Self-report measures register the respondent’s subjective feeling (i.e. their expression of their consciously felt experience of emotions) (Stout and Leckenby 1986) and constitute the most frequently used procedures for measuring emotions in marketing (Bagozzi et al. 1999; Poels and Dewitte 2006). Indeed, the descriptive theories of emotion suggest that the labelling of the subjective experience of emotion provides a means for measuring and distinguishing among emotional states (Hirschman and Holbrook 1982; Holbrook and Hirschman 1982; Holbrook and Batra 1987).

Moment-to-moment self-report ratings were not considered appropriate since they would interrupt the participants during the movie and would detract from the entertainment experience. More suitable appeared to be verbal self-reporting where individuals are asked to express their emotions in words, via open-ended questions, or to rate their emotions on a battery of emotion items using semantic differential or Likert scales (Poels and Dewitte 2006). Completed after watching the film, this approach is simple, cheap, quick, non- interruptive, and prevalent in the marketing and psychology literature (Bagozzi et al. 1999; Stewart et al. 2007).

 O              P     !  " # $% &  '    (0-( That said, verbal self-reporting is not without its limitations: • Emotion scales often consist of a long list of emotion adjectives, and rating all these items may be cumbersome and produce respondent fatigue (Poels and Dewitte 2006). • The cognitive processing required in putting a label to an emotion may distort the original emotional reaction, especially in the case of lower-order emotions (Poels and Dewitte 2006). • Respondents may be unable to report their emotions because they are not aware of exactly how they feel (Zaltmann 2003; Chartrand 2005; Winkielman, Berridge and Wilbarger 2005) • Respondents may be unwilling to report their emotions because of social desirability concerns (King and Bruner 2000). • Verbal self-reporting is retrospective, measuring emotional reactions after the stimulus is shown27 (Poels and Dewitte 2006). Discrete emotions are often short-lived, and once activated, may stimulate other emotional reactions closely related to them (Bagozzi et al. 1999). Therefore, when using verbal self-report, a perception of emotional response may be measured as opposed to the actual emotional response itself (Poels and Dewitte 2006).

Yet despite these limitations, marketers have tended to take an empirical approach to the measurement of emotions and rely on verbal self-reports with the marketing and advertising literature filled with pen and paper scales for labelling emotional responses (Stewart et al. 2007). In the typical application, many items are administered to measure reactions to a stimulus, and methods such as factor analysis, multidimensional scaling or cluster analysis are used to identify the underlying emotional dimensions for the sample (Bagozzi et al. 1999). As will be discussed later in this chapter, this research used Rasch analysis to confirm the underlying dimensions.

27 Therefore, to clarify, in this study, respondents were asked after the film to describe how they felt and reacted during the film, and how much cognitive effort they needed to expand whilst watching the film to follow it, as well as reflect on how much they liked certain aspects of it.  O              P     !  " # $% &  '    (0.( 4.4 – Theory of Measurement The advancement of science depends on measurement. Yet Michell (1999) suggests that the social sciences have concentrated only on the “instrumental task” of measurement, devising procedures and instruments to measure latent constructs. Indeed, the commonly cited definition by Nunnally (1967, p2) states that measurement “involves rules for assigning numbers to objects to represent quantities of attributes”. This assignment of numbers is now accepted as the key to measurement in contemporary marketing research. However, this definition has recently been criticised for presuming measurement rather than defining what has to be fulfilled to constitute measurement. The manifest data are immediately seen as being measures, when in actual fact, these numbers must show a pattern of inter- correlations in order to be accepted as measurement. Consequently, there has been a failure to address the “scientific task” of measurement, namely, whether measurement has been achieved at all (Ewing, Salzberger and Sinkovics 2005).

The quality and significance of any empirical findings depend on the quality and the properties of the measurement theory. We would expect that the theory explains how measurement is accomplished, explaining how the gap between the empirical domain (manifest variables / observations) and the numerical domain representing theoretical constructs (latent variables) is bridged (Molenaar 1995). We also want to be able to test the theory and see whether it corresponds with “reality” (Salzberger 1999). Therefore, in general terms, a measurement theory should describe how a measurement instrument or scale performs its measuring function and how the observations obtained by using an instrument translate to a position on the latent concept purported to be measured by that instrument (Singh 2004).

Developing good measures can be challenging, especially in areas where constructs are difficult to define. A desirable measure is one that is simple and easy to use and is characterised by the high quality of the information that is obtained (usually reported as reliability and validity) (Green and Frantom 2002). It should also yield invariant scores. This means that regardless of the instrument used, the score remains constant, and the  O              P     !  " # $% &  '    (0/( instrument can be used with anyone (for example, regardless of the ruler used, a person’s height remains constant and the ruler can be used with anyone) (Green and Frantom 2002).

Specific objectivity is another desirable characteristic in a measure. This means that a person’s trait is independent of the specific set of items used to measure it. For example, it should not matter which ruler is used to measure a person’s height – any ruler could be used and any ruler that is used would be independent of the person’s height (Green and Frantom 2002). Measures with specific objectivity can be tailored to any given respondent, thus permitting individually administered questionnaires and precluding the administration of items that are not appropriate for a particular respondent (Green and Frantom 2002).

With interest in measurement issues remaining unabated, psychometricians have continued to generate alternative measurement approaches and models which offer more precise measures and introduce rigid rules of measurement (similar to physics) into social sciences (Salzberger 2004; 2007). These alternative measurement approaches have been slow to diffuse into the marketing literature, despite marketers’ inherent interest in measurement issues. Instead, most measurement approaches in marketing (namely reliability assessment, confirmatory factor analysis and scale development) still rely on classical test theory (Singh 2004). One alternative measurement approach is Item Response Theory (IRT), in particular Rasch modelling (which is a type of IRT model). This will be introduced next.

4.5 – Justification for Rasch Modelling for scale development “IRT can do the same things better and can do more things when it comes to modelling existing tests, constructing new ones, applying tests in non-standard settings, and above all, interpreting the results of measurement”

(Fischer 1995, p4)

 O              P     !  " # $% &  '    (00( 4.5.1 - What is the Rasch model? Item response theory (IRT) offers an alternative to Classical Test Theory (CTT) and is rapidly becoming the dominant paradigm in psychometrics, education and health research (Bond and Fox 2001). These models scale item characteristics (i.e. item parameters) and person characteristics (i.e. person parameters) onto the same latent dimension and focus on the individual response (Salzberger 1999). A powerful one-parameter IRT model is the Rasch (1960) model, which can be used with both dichotomous and scale data. The polytomous model (Andrich 1988a, p366) (see Figure 4.5.1) is an extension of Rasch’s (1960) original dichotomous model, and is suitable in this research given that the data were collected based on polytomous response formats.

Figure 4.5.1 – General Polytomous Rasch model

The Rasch model uses a logarithmic transformation function to model the probability that a person with X ability (or personality characteristic) will pass (or agree with) an item of Y difficulty (or intensity of that characteristic) (Bond and Fox 2001). In other words, with Rasch modelling, a person having greater ability than another should have the greater probability of solving any item of the type in question, and similarly, one item being more difficult than another means that for any person, the probability of solving the easier item  O              P     !  " # $% &  '    (01( correctly is the greater one (Rasch 1960). In a marketing context the term difficulty can be replaced by the concept of endorsability (i.e. how hard it is to endorse the item or how extreme the item is) (Ganglmair and Lawson 2003; Ewing et al. 2005).

The Rasch model tests whether a single latent trait actually underlies a number of questions that are conceptualised to comprise a unidimensional scale, and establishes where respondents are positioned on this latent trait. These person and item locations are based on observable responses (Salzberger, Andrich and Soutar 2001). The result of applying Rasch procedures when the data adequately fit the model is a sample-independent ordering of items by degree of difficulty and persons by ability level along a linear interval log scale. This means that comparisons of items are invariant to people and comparisons of people are invariant to items (Bond and Fox 2001). Rasch models also check that the probability of answering one question correctly does not increase the probability of answering another question correctly within the questionnaire (local independence) (Bhakta, Tennant, Horton, Lawton and Andrich 2005).

Unlike the Guttman (1950) model, the Rasch model is probabilistic rather than deterministic, and recognises that the same total score can be arrived at by different combinations of items, with the Guttman structure being the most probable pattern (Andrich 1985). The Rasch model uses the traditional total score (i.e. the sum of item ratings) as a starting point for estimating response probabilities and then constructs a line of measurement with the items placed hierarchically on this line according to, in this case, how much they capture a respondent’s level of engrossment. The validity of a given test can be assessed through examination of this item ordering (i.e. by assessing whether all items work together to measure a single variable) (Prieto, Thorsen and Juul 2005). Provided that the data fit the model, the total score for each of the dimensions will add up to provide all of the information about the respondent’s engrossment, and thus, “the classification of persons according to their total scores is justified” (Andrich 1988b, p38).

 O              P     !  " # $% &  '    (02( 4.5.2 - Advantages of the Rasch Measurement approach over the Classical Test Theory approach Classical test theory (CTT) has been the predominant measurement paradigm in marketing research, yet despite its widespread use, it has been criticised for being an unsophisticated approach to measurement (Michell 1999). CTT has several shortcomings that should be considered, the most significant being that it does not provide an empirically testable theory of how measurement is accomplished (Salzberger 1999).

To accomplish measurement, Wright and Masters (1981) established seven criteria for creating an interval level scale that measures a variable. Rasch measurement satisfies all these criteria. Classical Test Theory does not.

1. An evaluation of whether each item functions as intended. 2. An estimation of the relative position (difficulty) of each valid item along the scale that is the same for all persons is required. 3. An evaluation of whether each person’s responses form a valid response pattern is checked. 4. An estimation of each person’s relative score (or attitude, achievement etc) on the scale is created. 5. The person scores and the item scores must fit together on a common scale defined by the items and they must share a constant interval from one end of the scale to the other so that their numerical values mark off the scale in a linear way. 6. The numerical values should be accompanied by standard errors which indicate the precision of the measurements on the scale. 7. The items should remain similar in their function and meaning from person to person and group to group so that they are seen as stable and useful measures.

Classical Test Theory relies heavily on the principle of correlation. From a pool of items, those that are retained are those that show high loadings in factor analysis and contribute to reliability and validity, which in turn, assess the quality of a scale (Churchill 1979). The use  O              P     !  " # $% &  '    (13( of these quality indices has received considerable criticism as it encourages the inclusion of items that tap similar aspects of the construct and have equal endorsability (Salzberger 2000). This can have unfortunate effects on the sensitivity of measures and their ability to provide valid scores at the extremes of the construct range since extreme items are generally discarded because too many or too few respondents affirm them (Ewing et al. 2005). Even Churchill and Peter (1984, p370) observed that maximising reliability tended to favour selection of “items so similar (to each other) that they under-identify constructs” (emphasis added).

On the other hand, Rasch modelling requires differences in the items representing the construct in question. Therefore, the researcher must generate items covering different intensity levels which capture the entire breadth of the construct under investigation, including those “extreme items” which extend the range of coverage of the construct (Salzberger 2000; Tennant, McKenna and Hagell 2004). Instead of focusing on item- intercorrelation (like CTT), the Rasch model focuses on the endorsability of items, ordering items in accordance to this endorsability rather than on their intercorrelation (Ewing et al. 2005). The range of different item locations lends meaning to different levels on the dimension of interest, thereby enhancing the interpretation of person measures, and providing insights that other forms of empirical methods cannot (Salzberger 2000).

Covering a wide range of the latent dimension is important for measurement for two reasons. Firstly, this breadth allows for differentiating between response patterns that are expected and those that are not. This is crucial for fulfilling the scientific task of measurement because if all items are equally endorsable, any pattern has the same likelihood (Ewing et al. 2005). Secondly, breadth of items guarantees the existence of items that yield enough information for respondents to ensure a small standard error of measurement. If all items are endorsed too easily (or not easily enough), the targeting of the measurement instrument is potentially suboptimal. In this case, standard errors are large and the model fit cannot be assessed properly (Ewing et al. 2005).

 O              P     !  " # $% &  '    (1*( The Rasch model estimates item and person parameters that best explain the manifest responses, producing scale-free measures and sample-free item “difficulties”. If the measurement instrument works properly, the estimation of item parameters does not depend on the specific sample used, and unbiased estimates of item properties may be obtained from unrepresentative samples (Embretson and Reise 2000). In contrast, when using CTT, the statistics obtained from the analysis only apply to the specific sample who took the test because the sums of scores on the items and the item difficulties are not calibrated on the same scale (i.e. they are sample dependent). Therefore, CTT cannot produce anything better than a ranking scale that will vary from sample to sample, and it cannot separate the attributes of the questions from the attributes (or ability) of the respondent. This makes it difficult to compare the performance of different sets of respondents who complete the same questionnaire at different times or with content variations (Bhakta et al. 2005).

Generally the measure estimated in Rasch modelling should be different from that calculated in CTT since the Rasch model will produce a different set of items contributing to the dominant trait than will CTT measurement (Waugh 2001). Mathematically, in Rasch measurement, when all the items fit the model, there is a predominant single trait underlying all the items, whilst lesser dimensions are reported as misfit to the Rasch model. Conducting factor analysis first and using these initial observations can lead to misleading results because when observations are non-linear, they can generate illusory factors (Wright 1996). Linacre (1998) argues that exploratory factor analysis can report items clustering at different performance levels as different factors and that there is no way of knowing from factor analysis alone whether each factor is a dimension or a slice of a shared dimension. These differences were clearly demonstrated in an example by Scott, Harris and Craig-Lees (2007).

Factor analysis identifies the relationship to the underlying variable, but not the location on it. Rasch analysis, in contrast, provides item and person location on the variable, facilitating the development of a construct theory and interpretation of levels of the construct (McInnes, Griffin, James and Coates 2001). However Chang (1996) demonstrated that  O              P     !  " # $% &  '    (1+( Rasch and factor analysis can produce similar results, but that Rasch results are simpler to interpret and are more stable and informative. For example, factor loadings are correlations of existing data with a latent vector constructed to minimise residuals, but the loadings are not on a linear metric. They constrict between -1 and +1, and any plots they may be cast in are co-ordinates rather than maps of a variable.

In comparison with CTT, the Rasch model provides a means of assessing a range of additional measurement properties, increasing the information available about a scale’s performance (Bond and Fox 2001). Rasch only uses items that fit the measurement criteria to form a valid measure of the variable, and a check is made to see that people respond to the valid items in a logical and consistent manner to form a scale (thus reducing noise) (Waugh 2001). Some problems with measures can be addressed in a Rasch analysis, but others cannot. We can selectively fix misuse of a response scale, identify and delete malfunctioning items, and remove the data of people who fail to respond appropriately to the task. But other problems cannot be readily fixed, such as failure to define a trait continuum or failure of items or people to yield a trait from their responses. Nevertheless, in either case, helpful information can be gained to aid decision making, guide scale improvement, and shed light on the validity of the scales constructed (Green and Frantom 2002). In traditional CTT approaches, groups of items are identified as factors, but no check is made (as is done in Rasch analysis) that the item responses are answered in a logical and consistent pattern to form a scale (Waugh 2001).

It is the chosen measurement theory that has a direct impact on the scale development process, specifically item generation and validity testing. Whilst there are a number of issues with CTT, there is value in Churchill’s (1979) discussion of the early stages of the scale development process using the domain sampling model. It is also more thorough and rigorous than Rossiter’s C-OAR-SE (2002) approach, especially at the back-end with its multiple tests of validity and reliability. Therefore, some elements of CTT were incorporated into this research. We also saw value in Rossiter’s (2002) approach in regards to construct definition, so adopted this too. However, for the most part, the Rasch approach  O              P     !  " # $% &  '    (1,( was applied in developing items with a range of difficulty to tap the breadth of the construct (as opposed to focusing on the development of highly correlated items as per Churchill (1979)) and in the analysis for refining the scale.

4.6 – Item Generation 4.6.1 – Specifying the domain of engrossment One objective of this scale development was to clarify and delineate the dimensions of audience engrossment in a media context, to precisely determine what is included in the definition of audience engrossment and what is excluded. Churchill (1979) advocates that the specification of this domain must deal with all the facets that belong to the construct. Incorporating the Rasch approach, we must also specify the meaning of different levels on the continuum. Therefore, using Rossiter’s (2002) C-OAR-SE approach, the focal object of study was an entertainment program (in this case, a film), the attribute being judged was engrossment (comprised of feelings, arousal, appraisal and cognitive effort), and the rater entity was an audience member. Hence, the construct definition of audience engrossment in this study would be “the combined level of feelings, arousal, appraisal and cognitive effort expended by an audience member during consumption of a film”. Despite the specificity of this definition, the construct must still be operationalised to “bridge the gap between the theoretical-conceptual level and the empirical-observational level” (Nachmias and Nachmias 1976, p17).

4.6.2 – Generating items With the concept of audience engrossment adequately defined, the second step in developing the scale was to generate items which capture the characteristics of a person who is engrossed with a story, and what feelings and behaviours they have whilst consuming it. Adopting Rasch measurement theory to the scale development process meant that item formulation aimed not only to cover all facets of a construct, but also attempted to represent a wide range of levels of the property to be measured (Salzberger 1999). Therefore, in order to operationalise the domain of interest, all items needed to reflect different degrees of the construct. This differs from the classical approach whereby a  O              P     !  " # $% &  '    (1-( suitable, reliable selection of items is found by thinking of different ways an item can be worded (De Vellis 2003) or by including items of slightly different meaning (Churchill 1979).

Audience engrossment may be considered a formed attribute, so multiple items are needed to describe and represent it. This is in agreement with traditional thinking on the use of multiple measures (e.g. Churchill 1979; Peter 1979; Rossiter 2002). Congruent with Rasch measurement theory, Bagozzi, Gopinath and Nyer (1999) also suggest that at least three items should be used for each emotional subcategory.

Three techniques were used to generate these items (as opposed to Rossiter’s approach which suggests using only target raters) - a thesaurus search to identify synonyms for engrossment and the different feelings and arousal items; a literature review of items used in related scales to determine which ones were relevant to the construct of audience engrossment; and focus groups containing teenagers aged 12-17 (to ensure that even the youngest group of possible respondents could understand all items included in the eventual scale).

In the focus groups, participants were asked questions from the general to the specific, such as - Do you know what engrossment means? What other words are like it? Do you ever get bored when watching a movie? What do you do when a movie does not capture your attention? Are there types of movies that do not interest you? What words would you use to describe how you feel and how you react when you watch a scary / happy / sad / funny movie? How do you feel when you watch a movie with a complicated storyline? At the same time, the film and actor preferences of the teenage participants were captured, providing direction for which films could be used as stimuli in the latter stages of this research (see Section 6.5 and Appendix 6.2). Individual responses were then elaborated on in a group discussion, and even more words developed.

 O              P     !  " # $% &  '    (1.( When consulting the existing literature, articles and scales encompassing the proposed factors comprising audience engrossment (feelings, physiological arousal, cognitive effort and appraisal) were considered and were used as a springboard for ideas. Other related streams of literature (e.g. connectedness, transportation, flow, presence) were also consulted for ideas, and any relevant items were allocated to the appropriate dimension.

4.6.3 – Applying Rasch Measurement Theory to item development With countless items developed for each dimension using these three methods, the next task was to reduce these items. Within each dimension, items capturing similar themes were grouped (for example, items capturing a sense of happiness), with obvious redundancies28 or irrelevant29 items removed. The task was then to identify which items captured a similar level of happiness, which ones captured a lesser level (e.g. contentness) and which ones captured a higher level (e.g. elated). If these different levels had not already been captured from the thesaurus search, literature review or focus groups, they had to be created now. So not only was it important to capture the core types of arousal (for example) which traditional scale development would aim to do, but varying degrees of these items also had to be captured. This had the consequence of creating more items than the usual scale would possess, but this is a natural by-product of this process.

As will be seen next, for each base idea that was identified, two other items were attached to it to add levels of difficulty. This resulted in 21 feelings items, 21 arousal items, 6 cognitive effort items and 33 appraisal items. The core feelings that were identified were happy, angry, horrified, distressed, scared, sad and uplifted. Arousal was comprised of crying, feeling relaxed, feeling tense, feeling anxious, fidgeting, feeling sleepy, and laughing. Appraisal considered assessments of the storyline, the actors, the characters, and the movie overall. Cognitive effort focussed on attention and plot complexity. The theory

28 An item was deemed redundant if it was a direct synonym of one of these items, or was perceived to be the same / similar level of intensity (e.g. elated vs joyous). 29 An item was deemed irrelevant if it was not appropriate to that dimension (e.g. sad is a pure feeling, not a reaction to that feeling i.e. arousal) or not a likely reaction to a film (e.g. shame)  O              P     !  " # $% &  '    (1/( that underlies the development of these items will now be discussed, and the items introduced.

Feelings and Arousal (see also Sections 3.8 and 4.3) Many scales and research studies focus on ‘emotions’, but they tend to tangle the concepts of feelings and arousal. There is general agreement amongst psychologists who study emotions that they involve three major components – physiological changes within our bodies (e.g. changes in heart rate, blood pressure), subjective cognitive states (the personal experiences we label as feelings) and expressive behaviours (outward signs of those internal reactions) (Baron 2001). As stated in Section 3.8, for this research, feelings were defined as the labels people give to those subjective cognitive states and sensations, and arousal was to do with the physiological changes and expressive behaviours caused by these feelings. There are two key aspects to emotion – valence (the extent to which the emotion is pleasant or unpleasant) and arousal (the intensity of the emotion) (Baron 2001). Because each item was unipolar, it had a given valence level (i.e. positive or negative). Intensity was captured by having three levels of each ‘basic’ emotion, and also by the type of response scale utilised (a frequency scale asking respondents to say how often throughout the movie they felt that way).

The Mehrabian and Russell (1974) scale is comprised of two separate similar dimensions – pleasure (which are related to feelings in this research) and arousal. This list served simply as a foundation, as it was not as exhaustive as was needed. A number of other existing scales and articles were also examined for relevant items (e.g. Watson and Tellegan 1985; Holbrook and Batra 1987; Storm and Storm 1987; Fischer, Shaver and Carnochan 1990; Richins 1997; Bagozzi et al. 1999; Feldman 1999). Some items were filtered out because they were not likely feelings to be evoked from an entertainment program or story, but rather, from inter-personal interactions (for example, envy, contempt, regret). Similarly, distinctions needed to be made between feelings and arousal, as many scales contained a mixture of both ideas, so these needed to be teased out and assigned to the appropriate dimension. For example, the Watson and Tellegen (1985) two-factor structure of (positive  O              P     !  " # $% &  '    (10( and negative) affect contained items such as active, excited, jittery, calm, happy and sad, but we consider jittery and calm to be forms of arousal, and happy and sad to be feelings. The following items were developed to capture feelings and arousal. They are listed in their triplets, in order of ascending difficulty of endorsability within each triplet.

Feelings I felt good’ ‘I felt annoyed’ ‘I felt dismayed’ ‘I felt happy’ ‘I felt angry’ ‘I felt appalled’ ‘I felt elated’ ‘I felt enraged’ ‘I felt horrified’

‘I felt concerned’ ‘I felt apprehensive’ ‘I felt sad’ ‘I felt distressed’ ‘I felt scared’ ‘I felt miserable’ ‘I felt distraught’ ‘I felt terrified’ ‘I felt totally depressed’

‘I felt comforted’ ‘I felt heartened’ ‘I felt uplifted’

Arousal ‘My eyes welled up with tears’ ‘I was anxious’ ‘I cried’ ‘I was uneasy’ ‘I sobbed’ ‘My stomach felt like it was tied in knots’

‘I was at ease’ ‘I found myself looking around the room’ ‘I was relaxed’ ‘I found myself fidgeting’ ‘I was in a dream-like state’ ‘I couldn’t sit still’

‘I smiled or chuckled to myself’ ‘I yawned’ ‘I laughed out loud’ ‘I had trouble keeping my eyes open’ ‘I laughed so hard that I cried’ ‘I fell asleep’  O              P     !  " # $% &  '    (11( ‘I was tense’ ‘I felt jumpy’ ‘I needed to grab hold of something (e.g. the person next to me, the armrest)’

Appraisal (see also Section 3.8) In this research, appraisal is essentially to do with assessments of the different elements of the movie that may contribute to overall liking and felt pleasure. The aspects considered were the overall storyline, the actors and the characters. Overall evaluative statements of the movie as a whole were also developed, as were some negatively phrased questions to test the reliability of responses.

Again, items from existing involvement scales were used as a springboard for ideas, as most involvement scales tend to focus solely on this liking concept, using liking items as proxy involvement measures. These included a “Program Liking” scale used by Murry, Lastovicka and Singh (1992) (e.g. ‘I am glad I had a chance to see this program’; ‘I would never watch a rerun of this program on television’; ‘I liked watching this program’; and ‘If I knew this program was going to be on television, I would look forward to watching it’) and Zaichkowsky’s (1985) 20-item Personal Involvement Inventory (PII) (e.g. appealing, interested).

The following items were developed and are listed in their triplets, in order of ascending difficulty of endorsability within each triplet.

‘I was interested in the storyline’ ‘I couldn't wait to see what happened next’ ‘I felt completely immersed in the story’

‘I like this type of movie’ ‘This is my favourite type of movie’ ‘I only ever watch this type of movie’  O              P     !  " # $% &  '    (12( ‘My mind wandered at times during the movie’ ‘My mind was constantly on other things’ ‘My mind was totally pre-occupied’

‘I like the actor(s) in this movie’30 ‘This movie starred one (or more) of my favourite actors’ ‘I would always make an effort to see a movie that starred one of these actors’

‘I found one or more of the main characters appealing’ ‘I really liked one or more of the main characters’ ‘I developed a real affection towards one or more of the main characters’

‘I occasionally chose not to watch or listen to the movie’ ‘There were many times that I chose not to watch or listen to the movie’ ‘I barely watched or listened to any of the movie’

‘I enjoyed watching this movie’ ‘I enjoyed watching this movie more than I have most others’ ‘I could happily watch unlimited re-runs of this movie’

‘One or more of the main characters did not appeal to me’ ‘I really disliked one or more of the main characters’ ‘I loathed one or more of the main characters’

‘The storyline had a lot of detail’ ‘The storyline was complicated’ ‘The storyline was extremely intricate and complex’

30 N.B. In the situation where there are no actors (e.g. documentary, cartoon), these items can be either deleted or adapted. In the case of a documentary on Africa for example, the items could be altered to read: ‘I like the country in this documentary’, ‘This documentary was about one of my favourite countries’, ‘I would always make an effort to watch a documentary about Africa’.  O              P     !  " # $% &  '    (23( ‘This movie did not appeal to me’ ‘I disliked this movie’ ‘I hated this movie’

‘I didn’t like the actor(s) in the movie’ ‘This movie starred one (or more) of my least favourite actors’ ‘I would avoid seeing another movie that starred one of these actors’

Cognitive Effort (see also Section 3.8) Cognitive effort relates to the ease with which an audience member can follow and understand entertainment content. Audience engrossment reflects the attention and effort put forth to do this. Research from Perse (1990; 1998) provided the foundation for items to do with attention and plot complexity (e.g. ‘I pay close attention when I watch television’; ‘I put a lot of mental effort into my television viewing’). These are now listed in their triplets in order of difficulty within each triplet.

‘The storyline was unclear’ ‘The storyline was difficult to understand’ ‘The storyline was totally incomprehensible’

‘I needed to pay attention to follow the story’ ‘I really had to concentrate to follow the story’ ‘Following this story was mentally demanding’

Each statement was reviewed so that its wording was as precise as possible (Churchill 1979). The respective sets of items were designed to have internal discriminant capacity, achieved by the items being written in differing levels of anticipated difficulty. Logic in the arrangement of items indicates that the researcher understood the construct, adequately operationalised it with the items written, and successfully communicated it to respondents via the items written to define it (Green and Frantom 2002). However, we could only  O              P     !  " # $% &  '    (2*( hypothesise these levels of difficulty from easy to hard. The data analysis outlined in Section 4.9 and Chapter 5 tested this ordering.

The other issue to consider was the rating scales. Bagozzi et al (1999) recommend the use of unipolar scales that ask respondents to express the extent that each emotion describes their own subjective feelings, rather than bipolar scales that can obscure differences in emotional responses across the various dimensions. Indeed, rating categories within items should form a continuum of less to more, so that only being able to endorse a lower category should represent being lower on the trait. Feelings, arousal and cognitive effort were initially asked according to a five-point frequency scale (never, a couple of times, a few times, often, throughout the movie). Appraisal was formatted into a five-point Likert scale (strongly disagree, disagree, neither agree nor disagree, agree, strongly agree), as such scales are widely used in instruments measuring opinions, beliefs and attitudes (De Vellis 2003).

4.7 – Initial Audience Engrossment Scale The initial scale comprised of four sections. Note that (R) next to an item reflects that it is reverse coded.

 O              P     !  " # $% &  '    (2+( Arousal – 21 items

PART 1

Firstly, we would like you to tell us how you reacted DURING the movie. Often Often Never Item label the movie

For each statement, please circle the number times A few

that best describes your reactions. of times A couple of most Throughout tears My eyes welled up with tears 1 2 3 4 5 cried I cried 1 2 3 4 5 sobbed I sobbed 1 2 3 4 5 ease I was at ease 1 2 3 4 5 relax I was relaxed 1 2 3 4 5 dream I was in a dream-like state 1 2 3 4 5 tense I was tense 1 2 3 4 5 jumpy I felt jumpy 1 2 3 4 5 grab I needed to grab hold of something 1 2 3 4 5 (e.g. the person next to me, the armrest) anxious I was anxious 1 2 3 4 5 uneasy I was uneasy 1 2 3 4 5 knots My stomach felt like it was tied in knots 1 2 3 4 5 look I found myself looking around the room (R) 1 2 3 4 5 fidget I found myself fidgeting (R) 1 2 3 4 5 still I couldn’t sit still (R) 1 2 3 4 5 yawn I yawned (R) 1 2 3 4 5 eyes I had trouble keeping my eyes open (R) 1 2 3 4 5 asleep I fell asleep (R) 1 2 3 4 5 smile I smiled or chuckled to myself 1 2 3 4 5 laugh I laughed out loud 1 2 3 4 5 Lhard I laughed so hard that I cried 1 2 3 4 5

 O              P     !  " # $% &  '    (2,( Feelings – 21 items

PART 2

In this section, we would like you to tell us

how you felt DURING the movie. Often Often Never Item label

For each statement, please circle the number A few times of the movie that best describes your feelings. Throughout most most Throughout A couple of times A couple

good I felt good 1 2 3 4 5 happy I felt happy 1 2 3 4 5 elated I felt elated 1 2 3 4 5 annoyed I felt annoyed 1 2 3 4 5 angry I felt angry 1 2 3 4 5 raged I felt enraged 1 2 3 4 5 dismay I felt dismayed 1 2 3 4 5 appalled I felt appalled 1 2 3 4 5 horrified I felt horrified 1 2 3 4 5 concerned I felt concerned 1 2 3 4 5 distressed I felt distressed 1 2 3 4 5 distraught I felt distraught 1 2 3 4 5 apprehensive I felt apprehensive 1 2 3 4 5 scared I felt scared 1 2 3 4 5 terrified I felt terrified 1 2 3 4 5 sad I felt sad 1 2 3 4 5 miserable I felt miserable 1 2 3 4 5 depressed I felt totally depressed 1 2 3 4 5 comforted I felt comforted 1 2 3 4 5 heartened I felt heartened 1 2 3 4 5 uplifted I felt uplifted 1 2 3 4 5

 O              P     !  " # $% &  '    (2-( Cognitive Effort – 6 items

PART 3

In this section we would like to know how you made sense of the movie. Often Often Never Item label the movie A few times For each statement, please circle the number A couple of times A couple

that best describes your response. of most Throughout diff1 The storyline was unclear (R) 1 2 3 4 5 diff2 The storyline was difficult to understand (R) 1 2 3 4 5 diff3 The storyline was totally incomprehensible (R) 1 2 3 4 5 attn1 I needed to pay attention to follow the story 1 2 3 4 5 attn2 I really had to concentrate to follow the story 1 2 3 4 5 attn3 Following this story was mentally demanding 1 2 3 4 5

Appraisal – 33 items

PART 4

ree

Now we would like to find out your thoughts g about the movie.

Agree Disagree Item label

For each statement, please circle the number disa nor Neither agree that best describes your thoughts. Agree Strongly Strongly Disagree Strongly interested I was interested in the storyline 1 2 3 4 5 wait I couldn't wait to see what happened next 1 2 3 4 5 immersed I felt completely immersed in the story 1 2 3 4 5 Xnot1 One or more of the main characters did not 1 2 3 4 5 appeal to me (R) Xnot2 I really disliked one or more of the main 1 2 3 4 5 characters (R) Xnot3 I loathed one or more of the main characters (R) 1 2 3 4 5 Glik1 I like this type of movie 1 2 3 4 5 Glik2 This is my favourite type of movie 1 2 3 4 5 Glik3 I only ever watch this type of movie 1 2 3 4 5 mind1 My mind wandered at times during the movie (R) 1 2 3 4 5 mind2 My mind was constantly on other things (R) 1 2 3 4 5 mind3 My mind was totally pre-occupied (R) 1 2 3 4 5 Story1 The storyline had a lot of detail 1 2 3 4 5 Story2 The storyline was complicated 1 2 3 4 5

 O              P     !  " # $% &  '    (2.( Story3 The storyline was extremely intricate and complex 1 2 3 4 5 Alik1 I like the actor(s) in this movie 1 2 3 4 5 Alik2 This movie starred one (or more) of my favourite 1 2 3 4 5 actors Alik3 I would always make an effort to see a movie that 1 2 3 4 5 starred one of these actors Mnot1 This movie did not appeal to me (R) 1 2 3 4 5 Mnot2 I disliked this movie (R) 1 2 3 4 5 Mnot3 I hated this movie (R) 1 2 3 4 5 Xlik1 I found one or more of the main characters 1 2 3 4 5 appealing Xlik2 I really liked one or more of the main characters 1 2 3 4 5 Xlik3 I developed a real affection towards one or more 1 2 3 4 5 of the main characters notW1 I occasionally chose not to watch or listen to the 1 2 3 4 5 movie (R) notW2 There were many times that I chose not to watch 1 2 3 4 5 or listen to the movie (R) notW3 I barely watched or listened to any of the movie 1 2 3 4 5 (R) Anot1 I didn’t like the actor(s) in the movie (R) 1 2 3 4 5 Anot2 This movie starred one (or more) of my least 1 2 3 4 5 favourite actors (R) Anot3 I would avoid seeing another movie that starred 1 2 3 4 5 one of these actors (R) Mlik1 I enjoyed watching this movie 1 2 3 4 5 Mlik2 I enjoyed watching this movie more than I have 1 2 3 4 5 most others Mlik3 I could happily watch unlimited re-runs of this 1 2 3 4 5 movie

A fifth section assessed six external factors that may impact on a person’s ability to become engrossed with an entertainment program (see Appendix 4.1). These items are not part of the audience engrossment scale per se, but are useful information to have for later analysis31. A final sixth section captured key demographic information concerning gender, age, highest level of education achieved, movie-watching frequency and degree of critical viewing/expertise, as it was believed that these factors could affect responses to the

31 N.B. Such analysis was not conducted in this study.  O              P     !  " # $% &  '    (2/( questionnaire (see Appendix 4.1). This information would allow for differential item functioning analysis (DIF) to be conducted (see Appendix 4.2).

4.8 – Initial content validity checks Following the finalisation of the list of items for the pilot study, a panel of experts was asked to review the items to ensure that all facets of each dimension were captured and that the levels of difficulty were appropriate. One academic in the field of emotions was sourced, as well as a Rasch expert and a media/advertising industry practitioner. A slight modification was made to a set of cognitive effort triplets to better ensure they were all capturing the one idea. No new items were suggested. All items were believed to be relevant to the audience engrossment construct and to be clear and concise. This expert review assisted in maximising the content validity of the scale.

4.9 – Refinement #1 – Pilot Study – November 2006 The purpose of the pilot study was to get an idea of how the different items and scale categories worked before conducting the major data collection at the cinemas. Conducted over a month, small groups of people (between two and six) saw a movie of their choice and then completed the questionnaire with the researcher present. This allowed them to ask any questions they may have and for the researcher to observe how they answered the questionnaire in regards to speed, body language and their understanding of the instructions. After all participants in a group had completed the questionnaire, the researcher then asked for feedback about readability, layout, and their understanding and perceived relevance of the different items. They were then debriefed.

4.9.1 – Sample and Implementation A preliminary sample of 34 males and 35 females aged 15-67 was obtained using convenience sampling (friends and acquaintances of the researcher). 94% of participants watched ten or fewer films per month (this included at the cinema, on DVD and on television). 84% had either an undergraduate or postgraduate degree.

 O              P     !  " # $% &  '    (20( The data were collected in the field at various cinemas around Sydney. The researcher paid for participants to see any movie of their choice in return for completing a questionnaire immediately after the movie. This method was deemed suitable due to the difficulties in securing a sufficient sample size quickly and the ease with which respondents were willing to participate. The questionnaire was self-administered using pen and paper32 and took less than ten minutes to complete.

The free choice component is a key feature of this entire scale development process – that all participants chose the movie they saw (as opposed to forcing them to see a movie that they may not typically choose to watch) as this most resembles the normal cinema-going experience and maximises the potential of capturing audience engrossment. Indeed, by providing this more natural exposure setting, the methodology removed the problems related to the forced exposure design of typical experiments, which tend to push subjects to attend to the stimuli more than they otherwise would have (Deighton, Romer and McQueen 1989). Indeed, Moorman (2003) found that experiments and real life settings showed opposing effects for the influence of media context on advertising effects, and that the forced exposure situation (which is characteristic of most experiments) hinders realistic evaluation of the effects of medium context. Furthermore, by collecting real-world data from non-student samples based on a real movie (c.f. Russell 2002), external validity was maximised.

The results that correspond to this pilot study stage may lack the depth of subsequent analysis (and are top-line only) due to the small sample sizes and aggregation of the results from all eleven movies into the one sample. Threshold parameters could not be estimated precisely, the power of the test of fit was limited, and DIF analysis could not be conducted. But whilst no final assessment of the scale was possible with this sample, it was believed

32 Both pen and paper and computer based methods were considered for data collection. Whilst having data collected automatically via a computer would have been more time efficient and accurate, this approach was abandoned for logistical reasons and to maximise external validity. If this scale is to be used in the “real world” it is more likely that it will be implemented using pen and paper, so in its development, this same method should be used.  O              P     !  " # $% &  '    (21( that any severe problems with the scale would be revealed. Therefore, this stage was useful for highlighting problems with item ordering and possible redundancies (via category frequencies and item maps), and gave us an insight into what dimensions may emerge in subsequent stages and whether the targeting between person ability and item difficulty was good. Examination of the item characteristic curves (ICCs) allowed examination of whether the actual results represented theoretical expectations. For further information regarding the analysis that was conducted and an explanation of these tests, please refer to Appendix 4.2.

4.9.2 – Use of RUMM2020 software The RUMM2020 software (Andrich, Lyne, Sheridan and Luo 2003) was used to analyse the data. This program was chosen for its ease of use and ability to provide item calibration and differential item functioning information in a single run, as well as its superior graphical displays.

From a functional perspective, RUMM2020 provides a range of information for assessing the quality of items on a scale including several different statistical and graphical tests of fit between the data and the model. It does this by projecting items and respondents onto the same dimension, thus providing indices and visual displays to investigate whether items spread sufficiently along a continuum rather than clustering towards one point of the dimension. This information (in combination with fit statistics) can be used to establish an overall conclusion of the quality of a scale, and suggest possible modifications. Thresholds, fit statistics, item characteristic curves (ICCs) and item maps were examined, whilst differential item functioning (DIF analysis) looked for differences in responses between groups (see Appendix 4.2).

 O              P     !  " # $% &  '    (22( 4.9.3 – Snapshot of AUDENG scale tested at Stage 133 Number Number of Dimension of items response scale Response scale category labels categories Feelings 21 5 Never, A couple of times, A few times, Often, Throughout the movie Arousal 21 5 Never, A couple of times, A few times, Often, Throughout the movie Appraisal 33 5 Strongly disagree, Disagree, Neither agree nor disagree, Agree, Strongly Agree Cognitive Effort 6 5 Never, A couple of times, A few times, Often, Throughout the movie

Total number of item = 81

4.9.4 – Results for Feelings Dimension When all 21 items were analysed together as one construct, the Person Separation Index (PSI) was 0.619. The overall chi square probability was virtually 0 suggesting that there were misfitting items. These were found to be ‘I felt good’ and ‘I felt happy’, both of which had high positive residuals (>2.5) and low chi square probabilities and F stat probabilities (as per the Bonferroni adjustment). Their ICCs also showed that the actual results deviated significantly from the expected results.

The overall item map was fairly sound, with the bulk of items and thresholds matching the respondents. Any mismatch, with no respondents agreeing with a statement (or choosing an extreme category), occurred with the strong feelings or high levels of these feelings (i.e. ‘miserable’, ‘terrified’, ‘depressed’, ‘enraged’, ‘scared’, ‘angry’ etc). However, it is

33 NB. The number of class intervals set for this analysis was two (under the assumption that each class interval should contain at least 40-50 respondents). This is the minimum number allowed by the RUMM2020 program.  O              P     !  " # $% &  '    (*33( plausible that none of the movies evoked any of these feelings, thus skewing the pattern of results.

Due to the low PSI and overall chi square probability, feelings were separated into two dimensions – positive and negative. Although statistically the feelings dimension was found to be unidimensional (t=1.62), separating this dimension does make sense. The group of six positive feelings (‘good’, ‘happy’, ‘elated’, ‘comforted’, ‘heartened’, ‘uplifted’) now had a good Person Separation Index of 0.821, an overall chi square probability of 0.167, and all individual item fit statistics were sound. However, there was some mis-ordering of the ‘comforted – heartened –uplifted’ triplet. Four out of the six thresholds were disordered (tending to happen at Threshold 3 (CT3) and Threshold 4 (CT4) suggesting that respondents could not distinguish between response options 3 and 4 (CT3) or options 4 and 5 (CT4) (i.e. “a few”, “often” and “throughout”) and there was now a gap in item-person matching, with no items to match those people with a low level of positive feeling (although conceptually these people may have felt only negative feelings or no feelings at all).

In regards to the 15 negative feelings (‘annoyed’, ‘angry’, ‘enraged’, ‘dismayed’, ‘appalled’, ‘horrified’, ‘concerned’, ‘distressed’, ‘distraught’, ‘apprehensive’, ‘scared’, ‘terrified’, ‘sad’, ‘miserable’, ‘totally depressed’), the Person Separation Index was good (PSI = 0.845) and the overall chi square probability not significant (p=0.103). All other key summary statistics were within their acceptable ranges, except the person location mean (- 2.136) which suggests that the items were generally difficult to endorse. This is because none of the movies watched by the respondents are likely to have evoked these feelings, as most tended to watch upbeat movies. There was also some mis-ordering of triplets with respondents not answering according to the level of difficulty anticipated by the researcher. Again, many of the thresholds were disordered (13 out of 15), tending to occur at CT3 but also CT2 and CT4, suggesting that people could not reliably report the frequency of their feelings. Whilst one should not make changes to the scale categories from such a small sample, it does flag the possibility that these category labels or number of categories may  O              P     !  " # $% &  '    (*3*( be unsuitable, and this was noted for closer examination in the next stage when the sample was larger.

4.9.5 – Results for Arousal Dimension The overall PSI for all 21 arousal items was 0.379 with a significant overall chi square (p=0.0006). However, all other summary statistics were sound, as were the item map and all the individual item statistics (except for the low chi square probability of ‘relaxed’).

Further investigation suggested that the arousal dimension was not unidimensional (t=- 2.34). Two dimensions emerged – tense / anxious and relaxed / bored / sleepy / laughing / crying. The tense-anxious dimension had an excellent test of fit, with a PSI of 0.924 and an overall chi square probability that was not significant (p=0.57). All individual item statistics were strong, but as with the feelings dimensions, there were disordered thresholds, mainly around CT3 (i.e. between “a few times” and “often”), and some mis-ordering of triplets. Overall however, the item maps looked fairly good with reasonable matching of items and people

The other dimension had a good PSI of 0.728 and overall chi square probability (p=0.21), thus suggesting good fit. All individual items showed good fit except for ‘I laughed out loud’ which had a positive residual value of 2.531. All items showed disordered thresholds around CT3 and CT4, suggesting confusion between “a few times”, “often” and “throughout”. The item map showed a good spread of items to people, with only the crying items not matching any people – presumably because none of the films shown evoked this reaction. Indeed, only the easiest item ‘I was close to tears’ showed any variance in response; the other two crying items were recorded as “never” by every respondent for every movie.

4.9.6 – Results for Appraisal Dimension All 33 appraisal items together had an excellent Person Separation Index of 0.951, but a highly significant overall chi square (p<0.0001) and high standard deviation fit residuals for  O              P     !  " # $% &  '    (*3+( both items and people which suggests that there was a lot of noise in the data. Most thresholds also tended to be disordered, generally at CT3, but also at CT2 and CT4.

Several items were found to be problematic. Items to do with the storyline being ‘detailed’, ‘complicated’ and ‘complex’ had high positive fit residuals, low chi square probabilities and F stat probabilities and visually poorly-fitting ICCs. Upon reflection of these items, it was believed that these items were not suitable for the appraisal dimension as some people might like these story characteristics, others may not, and a story having these characteristics is not necessarily a bad thing. Such ambiguity should be avoided, so these items were marked for deletion.

Another item with high positive residuals was ‘I loathed one or more of the main characters’. When an item has such high positive residuals (i.e. >2.5) it is flagged for deletion because it is under-discriminating and the responses are less predictable than expected. Examination of the item maps showed redundancy with the liking / not liking items (of the actors, characters and the movie) in terms of their locations. In addition to this, it was also felt (in hindsight) that the ‘not liking’ items were harder to interpret and that their inclusion made the questionnaire unnecessarily long and difficult to answer. Therefore, these nine items were also flagged for deletion.

The data from the appraisal dimension was re-analysed with those nine items removed, as well as the three aforementioned storyline items. The Person Separation Index was now 0.944 and the overall chi square probability improved (p=0.02), as did the other key summary statistics, ICCs and degree of disordered thresholds. Item maps revealed that all item triplets were ordered in the expected patterns, and there was good item-person matching. Two items had high positive residuals, but removing these did not improve the model, and for the sake of maintaining items in their triplets, these were kept for further testing in the next stage. Finally, all 21 remaining items were found to form a unidimensional appraisal construct (t=0.4).

 O              P     !  " # $% &  '    (*3,( 4.9.7 – Results for Cognitive Effort Dimension Together, all six items produced an excellent Person Separation Index (PSI = 0.876) and a good chi square probability (p=0.07) to form a unidimensional scale (t=0.27). Individually, item fit was good, with each item’s residuals within the healthy range (between +/- 2.5), and no significant chi square probabilities or F stat probabilities (as per the Bonferroni adjustment) were present, although every item suffered from disordered thresholds.

Despite these positive results, the item map revealed that in general, the items were difficult for the majority of respondents to endorse. This was also reflected by the mean person location (-1.619). Conceptually, this shows that most people found they had to evoke little cognitive effort to understand the story, and suggests that items that capture the more common movie experience whereby high levels of attention and concentration are not needed should be added.

4.9.8 – Modifications to the scale before Stage 2 Items to remove It should be noted that the 12 appraisal items were removed because the statistics highlighted potential problems that in hindsight made conceptual sense. We were reluctant to remove items based only on statistical results because of the small sample size in this stage. That said, the impact of removing these aforementioned items from the scale could be seen immediately as all results improved. It also made sense from an implementation perspective in terms of reducing the length of the scale and making it more user-friendly.

Items to add Analysis of the cognitive effort scale highlighted different concerns, and actually suggested that additional items were needed to capture a more typical movie processing experience – one that did not require a lot of effort. Consequently, three new items were added which dealt with the storyline being easy to follow (i.e. lacks complexity and requires only low levels of attention):

 O              P     !  " # $% &  '    (*3-( • ‘I found the storyline clear and easy to understand’ • ‘I thought the storyline was simple’ • ‘I thought the storyline was too simplistic’

Conceptually, we could not think of different items to tap into this ‘easiness’ for complexity and attention separately, so we relied on subsequent stages to suggest whether they would become a third dimension, whether they would bring all the dimensions into one dimension, or whether they would relate more strongly to either complexity or attention.

Changes to scale categories and items to reword The answering scale for the cognitive effort dimensions was changed from a frequency scale to a ‘strongly disagree - strongly agree’ scale. This decision was made as it was felt that the previous answering system may have been too difficult for people to answer. It was felt that it would be easier for people to agree or disagree with statements regarding complexity and difficulty rather than how often they felt the storyline was complex or difficult to follow. It was hoped that this new method would make future results more reliable.

Following this change, there was a slight rewording of the original six items so that they now incorporated a frequency component in them, as well as a personal evaluation statement (i.e. ‘I thought…….). It also meant synonyms were minimised as degrees of a trait were built via frequency rather than intensity of words.

• ‘The storyline was unclear’ b ‘I thought that parts of the storyline were difficult to understand’ • ‘The storyline was difficult to understand’ b ‘I thought that overall the storyline was difficult to understand’ • ‘The storyline was totally incomprehensible’ b ‘I thought the storyline was totally incomprehensible’  O              P     !  " # $% &  '    (*3.( • ‘I needed to pay attention to follow the story’ b ‘There were times I needed to pay attention to follow the story’ • ‘I really had to concentrate to follow the story’ b ‘Generally I had to concentrate to follow the story’ • ‘Following this story was mentally demanding’ b ‘I found that following this story was mentally demanding’

Since the thresholds relating to the frequency scales were consistently disordered across the different dimensions, a slight change was made to the wording of the scale categories to see if this made it clearer for respondents. The categories were changed from never, a couple of times, a few times, often, throughout most of the movie to never, a couple of times, fairly often, very often, throughout most of the movie.

Item order Finally, it was decided that a random mixing of items within each dimension (as opposed to listing them in order within their triplets) would be done for Stage 2. One requirement of measurement and of fit to the Rasch model is local independence. This means that the response only depends on the location of the item and of the person, and that the probability of answering a question one way does not increase the probability of answering another question that same way. It is represented by uncorrelated residuals (<0.3). Leaving the items in their triplets, in ascending order of difficulty creates extra dependency amongst the items and may result in respondents responding artificially and attempting to respond consistently with the pattern that emerges. The last thing a researcher wants is one item implying the response to another. When multiple measures of a single discrete emotional response (e.g. ‘happy’, ‘pleased’, ‘joyful’ for joy) are interspersed throughout a questionnaire, this tends to reduce correlations between them (Bagozzi et al. 1999). Therefore, randomly ordering the items within their dimension would minimise this effect and ensure greater local independence.

 O              P     !  " # $% &  '    (*3/( 4.10 – Revised Audience Engrossment Scale Following these changes, the following items formed the scale to be tested in the formal scale testing process. Note that (R) next to an item reflects that it is reverse coded.

Arousal – 21 items

PART 1

Firstly, we would like you to tell us how you reacted e es vi

DURING the movie. o im t Never m Item label Very often often Very Fairly often often Fairly For each statement, please circle the number that best of A couple

describes your reactions. the Throughout Lhard I laughed so hard that I cried 1 2 3 4 5 look I found myself looking around the room (R) 1 2 3 4 5 anxious I was anxious 1 2 3 4 5 eyes I had trouble keeping my eyes open (R) 1 2 3 4 5 relax I was relaxed 1 2 3 4 5 laugh I laughed out loud 1 2 3 4 5 asleep I fell asleep (R) 1 2 3 4 5 uneasy I was uneasy 1 2 3 4 5 smile I smiled or chuckled to myself 1 2 3 4 5 tears I was close to tears (e.g. eyes welled up with tears / there 1 2 3 4 5 was a lump in my throat) grab I needed to grab hold of something (e.g. the person next to 1 2 3 4 5 me, the armrest) still I couldn’t sit still (R) 1 2 3 4 5 cried I cried 1 2 3 4 5 ease I was at ease 1 2 3 4 5 jumpy I felt jumpy 1 2 3 4 5 fidget I found myself fidgeting (R) 1 2 3 4 5 yawn I yawned (R) 1 2 3 4 5 knots My stomach felt like it was tied in knots 1 2 3 4 5 dreamlike I was in a dream-like state 1 2 3 4 5 sobbed I sobbed 1 2 3 4 5 tense I was tense 1 2 3 4 5

 O              P     !  " # $% &  '    (*30( Feelings – 21 items

PART 2

e

In this section, we would like you to tell us how you vi

felt DURING the movie. o Never m Item label Very often often Very For each statement, please circle the number that often Fairly Throughout the the Throughout

best describes your feelings. of times A couple

appalled I felt appalled 1 2 3 4 5 depressed I felt totally depressed 1 2 3 4 5 scared I felt scared 1 2 3 4 5 happy I felt happy 1 2 3 4 5 angry I felt angry 1 2 3 4 5 distressed I felt distressed 1 2 3 4 5 heartened I felt heartened 1 2 3 4 5 miserable I felt miserable 1 2 3 4 5 horrified I felt horrified 1 2 3 4 5 annoyed I felt annoyed 1 2 3 4 5 good I felt good 1 2 3 4 5 dismayed I felt dismayed 1 2 3 4 5 comforted I felt comforted 1 2 3 4 5 elated I felt elated 1 2 3 4 5 apprehensive I felt apprehensive 1 2 3 4 5 concerned I felt concerned 1 2 3 4 5 terrified I felt terrified 1 2 3 4 5 enraged I felt enraged 1 2 3 4 5 distraught I felt distraught 1 2 3 4 5 sad I felt sad 1 2 3 4 5 uplifted I felt uplifted 1 2 3 4 5

 O              P     !  " # $% &  '    (*31( Appraisal – 21 items

PART 3

Now we would like to find out your thoughts about ee the movie. r

sag

i d

For each statement, please circle the number that Agree r Disagree Item label best describes your thoughts. Note that these o n Neither agree questions ask you to agree or disagree with each Agree Strongly statement. Disagree Strongly

Alik1 I like the actor(s) in this movie 1 2 3 4 5 Mlik3 I could happily watch unlimited re-runs of this movie 1 2 3 4 5 interested I was interested in the storyline 1 2 3 4 5 mind1 My mind wandered at times during the movie (R) 1 2 3 4 5 Alik2 This movie starred one (or more) of my favourite actors 1 2 3 4 5 notW3 I barely watched or listened to any of the movie (R) 1 2 3 4 5 Glik1 I like this type of movie 1 2 3 4 5 Xlik3 I developed a real affection towards one or more of the 1 2 3 4 5 main characters Mlik1 I enjoyed watching this movie 1 2 3 4 5 Glik2 This is my favourite type of movie 1 2 3 4 5 Mind2 My mind was constantly on other things (R) 1 2 3 4 5 notW1 I occasionally chose not to watch or listen to the movie 1 2 3 4 5 (R) Xlik1 I found one or more of the main characters appealing 1 2 3 4 5 wait I couldn't wait to see what happened next 1 2 3 4 5 Glik3 I only ever watch this type of movie 1 2 3 4 5 mind3 My mind was totally pre-occupied (R) 1 2 3 4 5 Alik3 I would always make an effort to see a movie that 1 2 3 4 5 starred one of these actors immersed I felt completely immersed in the story 1 2 3 4 5 Mlik2 I enjoyed watching this movie more than I have most 1 2 3 4 5 others Xlik2 I really liked one or more of the main characters 1 2 3 4 5 notW2 There were many times that I chose not to watch or 1 2 3 4 5 listen to the movie (R)

 O              P     !  " # $% &  '    (*32( Cognitive Effort – 9 items

PART 4

In this section we would like to know how you made ee sense of the movie. r

sag

i d

For each statement, please circle the number that best Agree r Disagree Item label describes your response. Like the last section, these o n Neither agree questions ask you to agree or disagree with each Agree Strongly statement. Disagree Strongly

easy3 I thought the storyline was too simplistic 1 2 3 4 5 diff2 I thought that overall the storyline was difficult to 1 2 3 4 5 understand (R) attn1 There were times I needed to pay attention to follow the 1 2 3 4 5 story attn3 I found that following this story was mentally demanding 1 2 3 4 5 easy2 I thought the storyline was simple 1 2 3 4 5 diff1 I thought that parts of the storyline were difficult to 1 2 3 4 5 understand (R) diff3 I thought the storyline was totally incomprehensible (R) 1 2 3 4 5 attn2 Generally I had to concentrate to follow the story 1 2 3 4 5 easy1 I found the storyline clear and easy to understand 1 2 3 4 5

4.11 – Conclusion This chapter outlined the steps involved in developing the Audience Engrossment (AUDENG) scale and justified the cinema setting and use of a verbal self-report method to capture this information. It justified the use of Rasch measurement theory as the underlying methodology, arguing that it is a more rigorous scale development procedure than the traditional Churchill (1979) factor analytic approach. The idea of capturing the breadth of the construct was introduced and the item generation process described, both in terms of the sources of information used, and how difficulty was built into the items. The scale was then reviewed by three experts before being subjected to a pilot study to detect any changes that should be made before major data collection commenced. The chapter concluded with an outline of the 72-item scale that would be formally tested at cinemas in the next stage.

 O              P     !  " # $% &  '    (**3( CHAPTER 5: REFINEMENT OF THE AUDIENCE ENGROSSMENT SCALE

“Most of our measures are only measures because someone says that they are, not because they have been shown to satisfy standard criteria”

(Jacoby 1978, p91)

5.1 - Introduction After introducing Rasch measurement theory and outlining the steps involved in developing the initial 81 items encompassing the Audience Engrossment (AUDENG) scale, Chapter 4 then reported on results from a pilot study. This saw some items deleted, some new items created, and amendments made to some of the response scale categories. This chapter outlines the next three stages of data collection, each of which further refined the items and response scale categories comprising the final AUDENG scale. It concludes by discussing the generalisability and the construct, content and convergent validity of the scale.

5.2 – Refinement #2 – March 2007 5.2.1 – Sample and Implementation The revised 72-item questionnaire was administered in March 2007 at an independent beach-side Sydney cinema to 317 actual cinema-goers after they had seen a movie of their choice. Flyers promoting the study were available at the box office and posters were displayed around the cinema, so interest was created even before people went to see the movie. Because of the intimate nature of this particular cinema, the researcher could also personally introduce herself to the customers and vaguely explain what her study was about (without disclosing the true purpose and goals) and encourage their participation following the movie. This priming was a very effective recruitment strategy. Upon exiting the cinema, patrons were approached using an intercept style method by the researcher and her assistants. Participants went in the draw to win a Gold Movie Pass (free movies at the

 O              P     !  " # $% &  '    (***( cinema for one year), and this incentive was found to be very attractive to all the participants. The questionnaire was self-administered using pen and paper and was found to take respondents approximately seven to ten minutes.

Examining Table 5.2.1.1, one can see that the overall response rate was 22% (1456 customers at the cinema during the days and time of the study, with 317 questionnaires completed). Tuesday, Friday and Saturday were the most successful days. 313 of the 317 responses were useable. Those that were discarded were either substantially incomplete or had no variance in response (for example, the respondent had answered 1 for every question).

Table 5.2.1.1 – Response rates at Cinema #1 Day Questionnaires Complete Tickets Sold Response Rate34 Tuesday 67 310 21.6% Wednesday 9 62 14.5% Thursday 31 74 41.9% Friday 61 164 37.2% Saturday 62 318 19.5% Sunday 53 239 22.2% Tuesday 34 289 11.8% TOTAL 317 1456 21.8%

57% of participants were female, compared to 43% who were male. There was a fairly even spread across age groups - 32% were aged 15-34, 32% were aged 35-54 and 36% were aged 55-plus. Respondents tended to be well educated, with 60% possessing an undergraduate or postgraduate degree. 52% see up to five movies each month (either at the

34 N.B. Response rate is the proportion of completed surveys from total attendees, not just from those people who were approached.  O              P     !  " # $% &  '    (**+( cinema, on DVD or on television), with 31% seeing between six and 10, 11% seeing between 11-20, and 6% seeing 21 or more.

Seven movies were part of this stage - Wild Hogs (n=85), Babel (n=75), The Illusionist (n=65), Notes on a Scandal (n=46), Pan's Labyrinth (n=23), Blood Diamond (n=16), and The Pursuit of Happyness (n=3). 98% of participants were seeing their respective movie for the first time.

In developing the optimal scale per dimension, only the specific results for Babel (n=75) and Wild Hogs (n=85) are discussed as these were the two movies with the largest samples (although other movies may be mentioned and were considered in developing the potential scale). In this way, we demonstrate generalisability of the scale across different movies, analogous to a replication. As opposed to the pilot study, statistics could be better relied upon in this stage due to the larger sample sizes per movie, so items were marked for deletion or modification based on a combination of their statistical results and qualitative assessments.

 O              P     !  " # $% &  '    (**,( 5.2.2 – Snapshot of AUDENG scale tested at Stage 235 Number Number of Dimension of items response scale Response scale category labels categories Never, A couple of times, Fairly Feelings 21 5 often, Very often, Throughout the movie Never, A couple of times, Fairly Arousal 21 5 often, Very often, Throughout the movie Strongly disagree, Disagree, Appraisal 21 5 Neither agree nor disagree, Agree, Strongly agree Strongly disagree, Disagree, Cognitive Effort 9 5 Neither agree nor disagree, Agree, Strongly agree

Total number of item = 72

5.2.3 – Results for Feelings Dimension Fit between the data and the model was not as good as anticipated, as reflected by high standard deviations of residual item means (>1) and significant chi square probabilities for both Babel (p=0.002) and Wild Hogs (p=0.00001) (see Appendix 5.1). Thresholds were examined to see if disordering may be affecting fit. Respondents with a particular level of a feeling should most likely choose the option that corresponds to that level (e.g. “fairly often”). In the case of disordered thresholds, respondents have a higher probability of choosing other categories (e.g. “a couple of times” or “very often”) than the category in between (i.e. “fairly often”). Such disordering would highlight that respondents could not

35 NB. The number of class intervals set for this analysis was two (under the assumption that each class interval should contain at least 40-50 respondents).  O              P     !  " # $% &  '    (**-( discriminate between response options, either due to too many response options or poor labelling of options.

As with the pilot study, thresholds were found to be disordered, a likely consequence of respondents being unable to distinguish between the five answer categories relating to the frequency with which the movie made the respondent feel a certain way (never, a couple of times, fairly often, very often, throughout the movie). After assessing precisely where this disordering was occurring, collapsing these to three categories (never, sometimes, throughout the movie) saw this problem resolved and was a recommended structural change for the next stage of scale testing (see Appendix 5.2).

Examination of each of the 21 items in this new response category structure revealed that ‘good’ was problematic for Babel (resid = 4.213) and The Illusionist (chi square prob = 0.001, F statistic probability = 0.001), ‘annoyed’ problematic for Wild Hogs (resid = 3.628) and ‘apprehensive’ problematic for The Illusionist (resid = 2.695). Removing these three under-discriminating items from the unidimensional feelings construct (Babel – t = 0.375; Wild Hogs – t = 1.039) saw a good fit for both movies (Babel – PSI = 0.865, p=0.47; Wild Hogs – PSI = 0.657; p=0.31). ICCs reinforced that most items had good fit to the model, with endorsability of items increasing consistently with an increased presence of evoked feelings. Item maps revealed that the thresholds captured the majority of respondents for both films, although some high categories on some items were not endorsed by any respondents, as reinforced by the negative person mean location which suggests that respondents had difficulty agreeing with some of the items (Babel = -2.59; Wild Hogs = - 2.86). However, this mismatching of items and people does make sense in relation to what film they relate to. For example, two items, ‘totally depressed’ and ‘enraged’ still had disordered thresholds for Wild Hogs, most likely because of near null counts for feeling this emotion at all during the movie, whilst the location of high level categories for feelings such as ‘miserable’ and ‘terrified’ could not be matched to any person location. This is likely because this movie is a comedy, and frequency counts for strong responses on these

 O              P     !  " # $% &  '    (**.( items were negligible. If an item bank strategy36 was used, it is possible that these items not be included for a light-hearted comedy such as this, or alternatively, exclude them at the analysis stage.

5.2.4 – Results for Arousal Dimension Together, all 21 arousal items had poor fit to the model (Babel – PSI = 0.347; chi square prob = ~ 0; Wild Hogs – PSI = 0.324, chi square prob = ~ 0) (see Appendix 5.3). Much of this may have been driven by the high degree of disordered thresholds for most items. After investigating the disordering, response categories were collapsed so the five-point frequency scale became a three-point scale (as described in the aforementioned discussion for the feelings dimension), and fit improved for both movies (Babel – PSI = 0.458; chi square prob = 0.09; Wild Hogs – PSI = 0.468, chi square prob = 0.003) (see Appendix 5.4). Dimensionality was then assessed and for both movies, arousal was found to be a unidimensional construct (Babel – t = 0.47; Wild Hogs – t = 1.55). However, upon reflection, it was thought that arousal might be multidimensional, so this concept was explored by grouping together similar items to see if fit improved. This proved to be an easier task for Babel than for Wild Hogs.

Different combinations of items which made conceptual sense to be grouped together were tested. The best combination of results for Babel saw three separate dimensions developed to form the overall arousal measure: negative arousal (PSI = 0.748, chi square prob = 0.46) (i.e. ‘I was close to tears’, ‘I cried’, ‘I sobbed’, ‘I was tense’, ‘I was jumpy’, ‘I needed to grab hold of something’, ‘I was uneasy’, ‘I was anxious’, ‘my stomach felt like it was tied in knots’), bored (PSI = 0.758, chi square prob = 0.19) (i.e. ‘I found myself looking around the room’, ‘I found myself fidgeting’, ‘I couldn’t sit still’, ‘I yawned’, ‘I had trouble keeping my eyes open’, ‘I fell asleep’), and positive arousal (PSI = 0.675, chi square prob = 0.03) (i.e. ‘I was at ease’, ‘I was relaxed’, ‘I was in a dream-like state’, ‘I smiled or chuckled to myself’, ‘I laughed out loud’, ‘I laughed so hard that I cried’). In comparison, there were

36 Use of an item bank would mean that only those items believed to be relevant for a particular movie would be measured (taken from the master list of all potential feelings that this research will develop)  O              P     !  " # $% &  '    (**/( not such clear dimensions for Wild Hogs. Their comparable results for these dimensions were negative arousal (PSI = 0.621, chi square prob = 0.04), bored (PSI = 0.848, chi square prob = 0.34) and positive arousal (PSI = 0.544, chi square prob = 0.004) with ‘I was in a dream-like state’ removed due to poor fit. The lack of emergence of clear dimensions in this still early stage of scale refinement highlights the importance of continued exploration in the next stage, with a larger sample and more movies to compare results across. Similarly, item maps were limited in detailing how well targeted the items were to the respondents due to the still relatively small sample sizes. Whist the item maps and person mean locations showed that targeting was generally poor, this may also be a function of the movie. For example, the person mean location for Babel for positive arousal was -2.665, but you would expect that respondents would find it difficult to endorse statements to do with smiling and laughing in what was described by participants as a depressing movie. Yet for Wild Hogs, the person mean location was 0.544 and the item maps showed that the bulk of respondents had items they could endorse.

5.2.5 – Results for Appraisal Dimension Initial inspection of all 21 appraisal items revealed some items to do with genre liking and actor liking to be misfitting (see Appendix 5.5). Upon reflection of this, it was felt that these items were not suitable for inclusion as they were not related to appraisal of this movie, but more general existing attitudes that may impact appraisal for any movie. Therefore, these six items were deleted from the appraisal dimension. However, it was important that a general understanding of genre liking and actor liking be measured to gain a greater understanding of what other factors may impact overall audience engrossment, so two new items would be placed in this section from the next stage onwards (but not analysed in the appraisal dimension, but rather as separate factors that may impact or interact with engrossment): ‘This movie starred one of my favourite actors’ and ‘I really like this type of movie’.

Inspection of the fit of the data from all remaining 15 appraisal items to the Rasch model for Babel showed very good fit between the data and the model (PSI = 0.847, p=0.03).  O              P     !  " # $% &  '    (**0( Whilst Wild Hogs had an excellent PSI of 0.913 (which suggests that the data could separate person ability and item difficulty parameters), there was a significant item-trait interaction total chi square (chi square prob = 0.00001). This suggests that there was some degree of misfit between the data and the model, and that respondents were not consistently agreeing upon the difficulties of the items along the scale. This misfit could be caused by misfit to model expectations of respondents, items, or both (see Appendix 5.6).

Thresholds were examined to see if disordering may be affecting fit, and, it was found across both movies that there were problems surrounding the “neither agree nor disagree” and “disagree” categories. Normally one would collapse these adjoining categories together, but these two obviously cannot be collapsed as they are capturing different ideas (no level of agreement/disagreement versus some level of disagreement). Hence, for the next stage, the “neither” category would be removed, and the scale changed from a five- point strongly disagree – strongly agree scale (with a neither category) to a six-point scale – strongly disagree, disagree, slightly disagree, slightly agree, agree, strongly agree. These problems around the “neither” category were also identified in the pilot study, so one can be confident that making this change is necessary. Giving six categories instead of four was deemed an appropriate solution after removing the “neither agree nor disagree” category, since those with little to no preference had a less intense statement to endorse than “agree” or “disagree”.

Next, individual items were examined, and as expected from its non-significant chi square probability and excellent PSI, there were no misfitting items for Babel. However, one item, ‘My mind was totally pre-occupied’ was seen to misfit in Wild Hogs (resid = 2.847, chi sq prob = ~ 0, F statistic probability = ~ 0)37. This was also confirmed by inspection of the poorly fitting ICC. Removing this item from the scale saw an improvement in the results for Wild Hogs (PSI = 0.915, chi square prob = 0.002) and still strong results for Babel (PSI = 0.841, chi square prob = 0.11) (see Appendix 5.7). With all individual items fitting for

37 Note that this item was also misfitting for other movies not discussed here  O              P     !  " # $% &  '    (**1( both movies, it was hypothesised that the still disordered thresholds may be to blame for Wild Hogs’ poor fit. This final solution confirmed appraisal as a unidimensional construct (Babel – t=0.477; Wild Hogs – t=-1.339) with item maps showing very good targeting between respondents and item difficulty.

5.2.6 – Results for Cognitive Effort Dimension Combining all nine cognitive effort items for Wild Hogs showed a good ability for the items to discriminate between people (PSI = 0.787), but the overall chi square probability was low (chi square prob = ~ 0) (see Appendix 5.8). Examination of individual items revealed that two items were significantly misfitting (‘I thought the storyline was too simplistic’ and ‘I thought that parts of the story were difficult to understand’). Although tests revealed that all items formed a unidimensional scale (t=1.16), items were separated into two dimensions to see if results improved (see Appendix 5.9). The three ‘easy to process’ items performed well together (PSI = 0.729, chi square prob = 0.04), as did the six ‘difficult to process’ items (PSI = 0.878, chi square prob = 0.05).

On the other hand, Babel had stronger results when all nine items were examined together (PSI = 0.769, chi square prob = 0.15). Separating the items into two dimensions saw still healthy, but weaker, results (‘easy’ items – PSI = 0.693, chi square prob = 0.003; ‘difficult’ items – PSI = 0.794, chi square prob = 0.20).

As with appraisal, the five-point strongly disagree-strongly agree scale was not working as expected (especially around the “neither agree nor disagree” response category), and this was resulting in disordered thresholds, which in turn affected fit. As opposed to feelings and arousal, we could not collapse categories in this scale and see if this had a positive impact (for reasons outlined earlier). Instead, modifications to response categories could not be assessed until new data were collected in the next stage. As with the appraisal dimension, moving forward, cognitive effort would now be measured on a six-point agree- disagree scale with no mid-point to see if this produced better results. This proposed change to the response scale structure, combined with the still relatively small samples for these  O              P     !  " # $% &  '    (**2( movies, led to the decision to leave all items in and see how they performed in the next larger research stage.

Item maps for the unidimensional scale showed that for Babel especially, the items were well targeted to the people and that the addition of the items reflecting an ease of processing meant that more people could be matched to an item. This was also revealed by the person mean location (-0.173). In contrast, the person mean location for Wild Hogs was -1.762 suggesting that the items were still somewhat difficult for the respondents to agree to. However this does make sense, as being an easy to follow comedy, strong attentional resources and cognitive effort were not required (which is what the bulk of items were still attempting to capture).

5.2.7 – Modifications to the scale before Stage 3 The aim of scale development is to produce a scale that is robust in different contexts, accurately measures the construct in question and distinguishes between people possessing varying levels of the construct. This larger research stage provided us with a good level of information, thus guiding us through the scale refining process.

Items to remove: Rasch analysis highlighted redundant items (e.g. ‘I found one of the characters appealing’; ‘my mind was constantly on other things’, ‘I barely watched or listened to any of the movie’) and provided multiple indicators as to why an item should be removed (e.g. similar locations on item maps, highly correlated residuals, high positive residual values). Furthermore, six appraisal items to do with actor and genre liking were removed when high positive residuals suggested they were actually measuring a separate idea, and conceptual reflection reinforced this.

Items to add: To compensate for the removal of the six appraisal items relating to actor and genre liking, two new items were created to capture these ideas: ‘This movie starred one of my favourite  O              P     !  " # $% &  '    (*+3( actors’38 and ‘I really like this type of movie’. However, these items do not form part of the audience engrossment scale as they relate to more enduring attitudes not caused solely by the entertainment program (although they may change as a result of watching the program). But it will be interesting to investigate whether these items impact or interact with audience engrossment, and it makes sense to collect this information when collecting the other audience engrossment data.

Items to reword: The Rasch model does not tell the researcher how to fix the problem to make an item fit if it fails to fit – all it tells the researcher is whether the particular wording used for the item produces data that can be explained by a single predominant trait. By examining an item’s location on the item map, its uniqueness could be determined, and even if identified for possible removal through other fit statistics, an item may still be retained due to its uniqueness by slightly changing its wording to see if that improved its fit. For example, ‘My mind was totally preoccupied’ was perhaps too ambiguous, so for the next stage, it was re-worded to ‘My mind was totally pre-occupied with things other than the movie’. Similarly, ‘annoyed’ was altered to ‘cross’, and ‘apprehensive’ to ‘on-edge’. The three items to do with feeling at ease were re-worded and rearranged from ‘at ease – relaxed – in a dream-like state’ to ‘at ease – calm – totally relaxed’.

Changes to scale categories: Through analysis of the thresholds, the randomness of respondents’ answering patterns was highlighted, as was the need to modify the number of categories (as discussed earlier). Post- hoc collapsing of the categories corrected the threshold issue for the feelings and arousal dimensions, but this new structure needed to be tested with the collection of new data using these new categories in the next stage. The next stage would also test whether the proposed changes to the agree-disagree scale would lead to improvements in model fit for appraisal and cognitive effort.

38 N.B. If there are no actors present (e.g. documentary), this item may be either deleted or adapted accordingly (e.g. ‘This documentary was about one of my favourite countries / animals etc’).  O              P     !  " # $% &  '    (*+*( 5.3 – Major Scale Testing – April/May 2007 5.3.1 – Sample and Implementation After engendering the support of a nationwide cinema chain, research was conducted at two of their Sydney locations for three weeks. These locations differed markedly from each other in regards to socio-demographic and economic profiles, thus maximising the potential generalisability of any subsequent findings. A similar intercept strategy was used with flyers available at the box office and posters placed around the cinema promoting the study and the prizes on offer (2 double passes to a full Gold Lounge experience and 50 regular movie tickets). In all, 944 valid responses were collected for five different movies - Shooter (n=272), Spiderman (n=262), Mr Bean's Holiday (n=165), Perfect Stranger (n=159) and The Number 23 (n=86). In developing the best overall scale, analysis was conducted on all movies except for The Number 23 due to its relative smaller sample size. Using multiple movies broadened the frame of reference for the scale and allowed a greater comparison of how items work across different movies.

Cinema 2 provided us with 52% of the valid questionnaires in this stage (see Table 5.3.1.1). This sample was made up of an even number of males and females, with two thirds of the sample being aged 34 or under. The respondents from Cinema 2 were less educated than the respondents from Cinema 3. Whilst 46% had either an undergraduate or postgraduate degree, 35% classed high school as their highest qualification. Patrons of this cinema tended to watch more movies than those at Cinema 3, with 66% watching at least six movies a month either at the cinema, or at home on DVD or television.

 O              P     !  " # $% &  '    (*++( Table 5.3.1.1 – Response rates at Cinema #2 Day Questionnaires Complete Tickets Sold Response Rate39 Thursday 19/4 41 257 16.0% Friday 20/4 75 391 19.2% Saturday 21/4 63 444 14.2% Sunday 22/4 55 370 14.9% Tuesday 24/4 96 452 21.2% Wednesday 25/4 86 509 16.9% Sunday 6/5 81 756 10.7% TOTAL 497 3179 15.6%

Cinema 3 provided us with 48% of the valid responses (see Table 5.3.1.2). Like Cinema 2, there was an even gender split, and two thirds of respondents were aged 34 or under. However, as stated, these customers tended to watch slightly fewer movies, with three quarters of respondents watching no more than 10 movies each month. They were also better educated, with 58% having either an undergraduate or postgraduate university degree.

Overall, there were no significant differences between respondents from the two cinemas for gender (02 (1) =0.356, p=0.546), age (02 (3) =1.496, p=0.683), movie-going frequency (02 (2) =1.3, p=0.522), or self-classified critic type (02 (2) =3.592, p=0.166). However, there was a significant different in education levels between the two cinemas (02 (3) =20.144, p=0.000), with Cinema 3 patrons being slightly better educated (as per the aforementioned discussion). However, based on the inherent overall similarities between the samples, the two samples were aggregated and analysis conducted on this combined sample.

39 N.B. Response rate is the proportion of completed surveys from total attendees, not just from those people who were approached.  O              P     !  " # $% &  '    (*+,( For confidentiality reasons, Cinema 3 would not provide us with information regarding ticket sales. However, one can be confident in saying that the response rate was higher at this cinema than Cinema 2. It was a slightly smaller and less busy cinema, and this allowed greater interaction with the patrons and the ability to generate interest in completing the questionnaire before they saw their movie. Furthermore, the catchment area of this cinema provided wealthier and higher educated people (Australian Bureau of Statistics 2007a), and they seemed more interested in participating in a university / PhD project than being in the running to win free tickets. This was also reflected in the quality of their responses, with less questionnaires being discounted from this cinema than Cinema 2 (two from Cinema 3, four from Cinema 2 – due to high levels of missing data or no variance of response).

Table 5.3.1.2 – Responses per day at Cinema #3 Day Questionnaires Complete Thursday 26/4 21 Friday 27/4 47 Saturday 28/4 76 Sunday 29/4 64 Tuesday 1/5 27 Thursday 3/5 75 Friday 4/5 78 Saturday 5/5 68 TOTAL 456

 O              P     !  " # $% &  '    (*+-( 5.3.2 – Snapshot of AUDENG scale tested at Stage 340 Number Number of Dimension of items response scale Response scale category labels categories Feelings 21 3 Never, At times, Throughout the movie Arousal 21 3 Never, At times, Throughout the movie Strongly disagree, Disagree, Appraisal 12 6 Slightly disagree, Slightly agree, Agree, Strongly agree Strongly disagree, Disagree, Cognitive Effort 9 6 Slightly disagree, Slightly agree, Agree, Strongly agree

Total number of item = 63

5.3.3 – Results for Feelings Dimension The change to the scale labels and number of categories to a three-point frequency scale (never, sometimes, throughout the movie) led to a marked improvement in threshold ordering, with no disordered thresholds for Shooter, Spiderman or Perfect Stranger (see Appendix 5.10). However, six items’ thresholds remained disordered for Mr Bean’s Holiday – ‘cross’, ‘angry’, ‘distressed’, ‘distraught’, ‘miserable’ and ‘depressed’. Interestingly, these were all feelings that would not likely be felt whilst watching this movie, and examination of the frequency counts for these items revealed that indeed, the incidence of these items sometimes or throughout the movie was very low. Although null categories do not mean that the model cannot be estimated in RUMM2020, the lack of information to estimate thresholds can lead to muddled results and the premature diagnosis of disordered thresholds. Disordered thresholds are problematic when many respondents

40 NB. The number of class intervals set for the analysis for Shooter and Spiderman was set at five, and for Perfect Stranger and Mr Bean’s Holiday at three (under the assumption that each class interval should contain at least 40-50 respondents).  O              P     !  " # $% &  '    (*+.( choose a category they should not have chosen, but this is not the case here since frequencies reveal that the item was responded to in the expected fashion, thus making any disordering of thresholds random or accidental. For these reasons, these disordered thresholds were noted, but not necessarily acted on.

Initial examination of the 21 feelings found that for all movies, feelings was a unidimensional construct (t<|1.96|). Attempts to separate feelings into two separate dimensions, positive and negative, which may have made conceptual sense, led to lower PSIs and poorer fit, so the unidimensional structure remained. Whilst all movies showed good Person Separation Indexes (all >0.75), overall chi square probabilities were significant (<0.05), which suggested misfit to the model (see Table 5.3.3.1). With thresholds more or less corrected, individual items needed to be closely examined for each movie to see how they were performing, as it was likely that some of these were affecting the model fit.

Examination of individual item fit across each movie revealed that there were misfitting items for each movie. Whilst the overall goal was to identify which items were having the greatest overall negative impact on model fit and discrimination, each movie was examined individually to see which items were misfitting for that movie and then to identify which items could be removed from all movies to create a model of optimal fit across all movies.

 O              P     !  " # $% &  '    (*+/( Table 5.3.3.1– Initial summation of feelings dimension – 21 items Shooter Spiderman Perfect Mr Bean’s Stranger Holiday PSI 0.810 0.813 0.849 0.764 02probability 0.000001 <0.000001 0.002 0.0001 Item fit problems • Good – pos • Appalled – • Appalled – • Sad – neg residual, 02 pos pos residual • Cross – F residual, residual • Appalled – 2 • Miserable - 0 , F 02 F • Distraught - F DIF41 • CI – cross, • CI – • Gender – - miserable appalled, scared • Gender – distraught good, • Gender – scared scared, • Age – terrified, dismayed, sad concerned Unidimensionality t = 0.60 t = 1.35 t = 1.55 t = 0.18

‘Appalled’ was misfitting and poorly discriminating across three of the movies (see Table 5.3.3.1 and Appendix 5.10 for exact values). Removing this item saw a general improvement in PSI and overall chi square probability. It also caused some previously misfitting or poorly discriminating items to perform better and there was now a smaller range of items mis-performing. ‘Cross’ was now misfitting across two of the movies, with ‘good’ and ‘horrified’ misfitting for one movie each. Each of these items was removed one by one from each movie, with the greatest positive impact found by the removal of ‘cross’ from the now 20-item dimension. The further removal of any other items was explored, but after trying countless scenarios, the best common result was achieved by removing ‘cross’ and ‘appalled’ only (see Table 5.3.3.2 and Appendix 5.11). Note that results for Mr Bean’s Holiday improved significantly from those in Table 5.3.3.2 by removing ‘dismayed’ and the remaining items with disordered thresholds (PSI = 0.779, chi square prob = 0.15). In contrast, the improvement in Spiderman by removing ‘distressed’ was barely noticeable

41 CI refers to significant class interval effects. These normally correspond to significant F statistics. Gender, age etc refers to significant main effects related to that person factor. For more explanation, see Appendix 4.2.  O              P     !  " # $% &  '    (*+0( (PSI = 0.812; chi square prob = 0.03). However, as discussed, the goal was to identify a scale that would suit any movie, so the removal of specific items affecting only one movie was done out of curiosity and would not be the final scale that was proposed since removing these items from other movies led to them having lower PSIs and poorer fit.

Table 5.3.3.2– Feelings dimension after removing ‘appalled’ and ‘cross’ – 19 items Shooter Spiderman Perfect Mr Bean’s Stranger Holiday PSI 0.798 0.815 0.855 0.763 02probability 0.0009 0.02 0.02 0.0001 Item fit problems - Distraught - F - Dismayed - 02 DIF • Gender – • Gender – • Gender – - good, scared, scared scared terrified, • Age – sad dismayed, concerned Unidimensionality t = 0.42 t = 1.21 t = 1.35 t = -0.94

Since overall fit was still less than optimal for three of the movies, but item fit and PSI appeared generally sound, further interrogation of the data needed to take place. For the first time in this scale refinement process, differential item functioning (DIF) could be examined due to the larger sample sizes. This step was conducted now since the removal of any other items was not positively impacting results. Table 5.3.3.2 shows that there were some clear gender DIF issues, thus driving the decision to explore the impact that splitting several items had on overall fit and PSI. Gender DIF was the category of greatest interest as it was to be expected that there would be a difference for males and females – both in regards to actual felt emotion, but also their inclination to report such feelings. Gender was also the most reliable category in that each of the class intervals would only be split into two (as opposed to four for age and education, or three for movie watching frequency and critic), thus having larger sub-samples. For this reason, any DIF found in other person factors was noted, but not acted on.

 O              P     !  " # $% &  '    (*+1( Across the four movies, gender DIF problems were found at times with ‘scared’, ‘sad’, ‘terrified’ and ‘good’. These were split across all movies (to promote generalisability of the scale), but the best common result across movies was found by splitting ‘sad’ and ‘scared’ only (see Table 5.3.3.3 and Appendix 5.19).

Item maps for the movies using this solution showed reasonable matching, although negative mean person locations reflected that there were items that were difficult for respondents to endorse (see Appendix 5.19). These were generally the intense feelings, especially high levels of these feelings (for example, ‘miserable’ and ‘horrified’). Examination of item hierarchy demonstrated that there was a consistent difference with what was hypothesised by the researcher to how respondents interpreted the intensity of items such as ‘totally depressed’ and ‘miserable’ (respondents found ‘miserable’ to be harder to endorse), whilst ‘uplifted’ and ‘comforted’ were found to be very similar.

Table 5.3.3.3 – General model for feelings – 21 items42 Achieved by removing ‘appalled’ and ‘cross’ and gender splitting ‘scared’ and ‘sad’ Shooter Spiderman Perfect Mr Bean’s Stranger Holiday PSI 0.797 0.816 0.857 0.770 02 probability 0.003 0.09 0.06 0.01 Item fit problems - - - - DIF • Gender – • Gender – • CI x • CI x critic good terrified education - – • Age – • CI x critic - horrified distraught dismayed, angry concerned Unidimensionality43 t = < |1.96| t = < |1.96| t = < |1.96| t = < |1.96|

42 N.B. Although only 19 separate feelings are measured, two items are created for both scared and sad – male scared, female scared; male sad, female sad – following the item splitting. This leads to a total of 21. 43 Dimensionality tests could not be conducted properly due to the splitting of ‘sad’ and ‘scared’. Residual principal components analysis could not be conducted as there were now four empty item pair combinations. Instead, random groups of items were compared using independent t-tests to show that there were no differences and that the final solution was unidimensional.  O              P     !  " # $% &  '    (*+2( It should be noted that these results do not represent the optimal scale for each movie. Indeed, there is room for improvement by looking at each movie individually and making unique changes for each movie. Following this, the optimal feelings scale for each movie would be identified as those listed in Table 5.3.3.4.

Table 5.3.3.4 – Optimal feelings scale per movie Shooter Spiderman Perfect Mr Bean’s Stranger Holiday Remove due to Appalled Appalled Appalled Appalled poor discrimination Cross Cross Cross Cross or fit Horrified Dismayed Remove due to - - - Angry disordered Depressed thresholds Miserable Distressed Distraught Split (by gender) Scared Sad Scared - Good Scared Terrified PSI 0.803 0.817 0.841 0.779 02 probability 0.02 0.10 0.11 0.15 Item fit problems - - - - DIF • Age – - - - dismayed, horrified

However, as stated earlier, the goal of this research was to provide a general scale that could be used for any movie, and analysis conducted on a broader range of software than just RUMM2020, so in the interests of generalisability, these results were sacrificed for the aforementioned solution featured in Table 5.3.3.3. However, an advantage of continued use of RUMM2020 in future applications of this scale is that the specifically ideal scale for each movie can be determined (from the master list of items that this research has generated) and used as the basis for further analysis for that movie if desired.

At this point, we were fairly confident with this proposed scale, but still sought confirmation in the next and final stage. To allow for synchronisation of data sets from this  O              P     !  " # $% &  '    (*,3( stage and the next, ‘appalled’ and ‘cross’ were not formally removed from the questionnaire at this stage.

5.3.4 – Results for Arousal Dimension Examining all 21 arousal items together revealed that the new response scale structure was successful, with only two items consistently having disordered thresholds – ‘I fell asleep’ and ‘I sobbed’ (see Appendix 5.12). Consequently, both those items were removed. Removing these items did make sense since it was unlikely that anyone would have fallen asleep during the movie, and sobbing is likely to be captured by a person responding that they had cried throughout the movie.

However, removing these items did not lead to improvements in fit. For all movies, PSIs ranged from 0.109 to 0.439, and overall chi squares were virtually zero. Several items still misfitted. Tests for multidimensionality demonstrated that arousal was comprised of two dimensions. For Mr Bean’s Holiday, Spiderman and Perfect Stranger, these could be described as a high and low arousal. For Shooter however, the dimensions seemed to be related to a positive and negative arousal. Forcing Shooter into the same dimension structure as the other movies led to comparable results to theirs, so the high/low arousal dimension structure remained. Low arousal was comprised of ‘I was at ease’, ‘I was calm’, ‘I was totally relaxed’, ‘I found myself looking around the room’, ‘I found myself fidgeting’, ‘I couldn’t sit still’, ‘I yawned’, and ‘I had trouble keeping my eyes open’. High arousal was comprised of ‘I was close to tears’, ‘I cried’, ‘I smiled’, , ‘I laughed out loud’, ‘I laughed so hard that I cried’, ‘I was tense’, ‘I felt jumpy’, ‘I needed to grab hold of something’, ‘I was uneasy’, ‘I was anxious’, and ‘My stomach felt like it was tied in knots’.

Low arousal brought reasonable initial results, with no disordered thresholds for any movie, and Person Separation Indexes between 0.587 and 0.608. However, overall chi square probabilities were significant (p<0.05) for all movies except Perfect Stranger. This dimension also still had items misfitting and suffering DIF. Similar results were achieved with the high arousal dimension. Here, Person Separation Indexes were higher (between  O              P     !  " # $% &  '    (*,*( 0.507 and 0.717), but two items still suffered disordered thresholds (‘I was close to tears’ and ‘I cried’), and there were still misfitting items and DIF. However, unlike in previous instances, systematic removal of items based on statistics or different theoretical reasons failed to lead to improvements in fit. In fact, it made the results worse.

It was therefore decided to split arousal into three dimensions as was done in Stage 2 – positive arousal (i.e. ‘I was at ease’, ‘I was calm’, ‘I felt totally relaxed’, ‘I smiled or chuckled to myself’, ‘I laughed out loud’, ‘I laughed so hard that I cried’), negative arousal (i.e. ‘I was close to tears’, ‘I cried’, ‘I sobbed’, ‘I was tense’, ‘I was jumpy’, ‘I needed to grab hold of something’, ‘I was uneasy’, ‘I was anxious’, ‘my stomach felt like it was tied in knots’), and bored (i.e. ‘I found myself looking around the room’, ‘I found myself fidgeting’, ‘I couldn’t sit still’, ‘I yawned’, ‘I had trouble keeping my eyes open’). This strategy was moderately successful for negative arousal and bored, but brought upon weaker results for positive arousal. Whilst overall chi square probabilities generally improved (but were still generally <0.05), Person Separation Indexes now ranged from 0.243 for Spiderman to 0.53 for Shooter. Compared to the other dimensions in the AUDENG scale, these results were incomparable and disappointing, yet no amount of item deletion or reallocation of items to different dimensions led to an improvement.

After examining correlation matrices in SPSS, it was decided that recoding some items may lead to improved results. The coding for ‘I was at ease’, ‘I was calm’ and ‘I felt totally relaxed’ was reversed, as it was believed that feeling these things may not be interpreted as positive arousal as had been done previously, but rather, not being aroused at all. In contrast, the fidgeting items (‘I found myself looking around the room’, ‘I found myself fidgeting’ and ‘I couldn’t sit still’) were recoded to be positively coded as it was felt that such behaviour was displaying some level of arousal, albeit negative or disinterested. The items to do with falling asleep (‘I yawned’, ‘I had trouble keeping my eyes open’, ‘I fell asleep’) remained negatively coded. It should be noted that all 21 items were included in this recoding (i.e. ‘I sobbed’ and ‘I fell asleep’ were put back into the dimension).

 O              P     !  " # $% &  '    (*,+( Examining all 21 items with this new coding structure saw instantly stronger Person Separation Indexes, ranging from 0.548 to 0.734 (see Appendix 5.13). Overall chi square probabilities remained virtually zero, but this was likely due to many items misfitting, and several items (‘tears’, ‘cried’, ‘sobbed’, ‘trouble keeping eyes open’ and ‘fell asleep’) still having disordered thresholds. Misfitting items were removed one by one, with eventually all fidgeting and sleeping items being removed, as well as ‘I smiled or chuckled to myself’, ‘I laughed out loud’, and ‘my stomach felt like it was tied in knots’. This left the dilemma of only one laughing item (‘I laughed so hard that I cried’) remaining in the scale, especially since it was the most difficult item to endorse from that triplet. Removing it from the unidimensional scale (t < |1.96| for all movies except Mr Bean’s Holiday) had positive impacts on the Person Separation Index for all movies, with the overall chi square probability remaining generally stable across all movies (see Table 5.3.4.1). This new scale was similar to the previous negative arousal scale, but with the addition of the ‘I was at ease’, ‘I was calm’ and ‘I felt totally relaxed’ items reverse coded. This scale made more sense in describing someone who was in a highly aroused and possibly negative state. A low score on this scale would suggest the person was at ease and happy.

This solution saw strong results for each movie – healthy Person Separation Indexes, non- significant overall chi square probabilities and no significant item misfit (see Table 5.3.4.1).

 O              P     !  " # $% &  '    (*,,( Table 5.3.4.1– Arousal Scale – 11 items Shooter Spiderman Perfect Mr Bean’s Stranger Holiday Disordered Cried Sobbed Tears Cried Thresholds Sobbed Sobbed PSI 0.751 0.779 0.751 0.694 02 probability 0.45 0.03 0.23 0.03 Item fit problems - • Cried - F - - DIF • Grab – • Cried - - • Tears - gender CI MW • Uneasy – age • Ease – MW • Sobbed - critic Unidimensionality t = 0.45 t = 2.12 t = 0.78 t = -1.47

As can be seen, ‘I sobbed’ had disordered thresholds for three of the four movies. Considering it was marked for deletion earlier for being too strong a statement and also captured by a respondent selecting a high response category on another similar question (i.e. ‘I cried throughout the movie’), it was decided to remove it and see the impact on the results. These appear in Table 5.3.4.2 and Appendix 5.20 and show that for the most part, the exclusion of this item led to another improvement in results. All ICCs were strong. As per the previous discussion in Section 5.3.3, DIF found in person factors other than gender was noted, but not acted on due to the small sub-samples that these statistics were derived from. The item with gender DIF for Shooter was also ignored since it only affected one movie. Item-person matching on the item maps was generally very good. Some items (e.g. ‘tears’, ‘cried’, ‘uneasy’, ‘jumpy’ and ‘anxious’) had no people matching them, although this made sense given the movies being tested. Similarly to feelings, there were also a group of people for each movie that did not match any items, presumably those people claiming to have not been at all aroused by the movie. Therefore, person location means tended to be negative (-1.302 to -1.866).

 O              P     !  " # $% &  '    (*,-( Table 5.3.4.2– Arousal Scale – 10 items Shooter Spiderman Perfect Mr Bean’s Stranger Holiday Disordered Cried - Tears Cried Thresholds PSI 0.749 0.773 0.746 0.674 02 probability 0.58 0.16 0.18 0.03 Item fit problems - • Cried - F - - DIF • Grab – - - • Tears - gender MW • Uneasy – age • Ease – age, MW • Calm – ed • Relaxed – CI x critic Unidimensionality t = 0.29 t = 0.33 t = 0.64 t = -2.41

Whilst these results seemed fairly stable – a common solution working across four movies with samples of at least 150 each, confirmation in the next stage was particularly necessary since results had changed throughout every stage for this dimension. For this reason, no items were formally deleted at this stage, with the full 21-item scale to be tested in the next and final stage. It should be noted that the disordered thresholds relating to ‘I was close to tears’ and ‘I cried’ were due to low frequencies across the range of response options since those films affected did not evoke such a reaction.

5.3.5 – Results for Appraisal Dimension Combined analysis of all 12 appraisal items revealed that they did an excellent job of discriminating between respondents (PSI > 0.85), but the chi square probability statistic was virtually zero (i.e. highly significant) (see Appendix 5.14).

For this stage of data collection, the response scale was changed from a five-point Likert strongly disagree – strongly agree scale (with a “neither agree nor disagree” midpoint) to a six point Likert scale with no midpoint (strongly disagree, disagree, slightly disagree,

 O              P     !  " # $% &  '    (*,.( slightly agree, agree, strongly agree). Examination of the thresholds revealed that there was still disordering and that forcing people to either agree or disagree with a statement (to varying degrees) had not solved this problem. Specifically, respondents seemed to have difficulty making a distinction between the options, particularly in the disagree half. However, in the interests of keeping the scale balanced, both sides were collapsed. Based on an examination of where the disordering was occurring, the ‘strongly’ categories stayed, with the “slightly agree” and “agree” categories being collapsed, as were the “slightly disagree” and “disagree”. Doing this removed most threshold disordering and was a recommended change for the final testing of the scale (see Appendix 5.15).

One item, ‘I barely watched or listened to any of the movie’ (‘notW2’) still had disordered thresholds across three of the four movies. In such an instance, it is recommended that this item be deleted. Furthermore, it was felt that another item still present in the scale, ‘There were times that I chose not to watch or listen to the movie’ (‘notW1’) would sufficiently capture this idea of “not watching”, so the former item was deemed to be a suitable candidate for deletion for both conceptual and statistical reasons.

After collapsing the scale categories and deleting that item, a better picture could be formed as to how the other items were performing and interacting (see Table 5.3.5.1).

 O              P     !  " # $% &  '    (*,/( Table 5.3.5.1– Results after collapsing scale categories and removing notW2 – 11 items Shooter Spiderman Perfect Mr Bean’s Stranger Holiday PSI 0.861 0.884 0.842 0.834 02 probability 0.01 0.004 0.000 0.001 Item fit problems • Mlik1 – • Mind2 – • Wait – F • Xlik2 – neg resid, pos resid, • Mind2 – pos resid, F 02 pos resid, 02 • NotW1 – 02, F • Mlik1 - 02, pos resid • Xlik2 – F • Wait – neg pos resid resid, F • NotW1 – • Mlik1 – pos resid, neg resid 02, F • Mlik1 – neg resid, F Unidimensionality t = -2.88 t = -1.525 t = -0.99 t = -0.14

Several issues became clear at this point. 1. For Shooter, appraisal was a multidimensional construct comprised of one dimension to do with liking and being engaged with the story, and the other dimension describing a person who does not like, or is disengaged with the story. For the other movies, appraisal was found to be unidimensional. However, the considerable misfit for items to do with not being engaged (i.e. to do with mind wandering (‘mind2’) and not watching (‘notW2’)) suggests that results could be improved by forcing the data into two dimensions. It also made sense to separate them out in to two dimensions, as one scale to measure high appraisal and one scale to measure low appraisal (or being disengaged) may provide greater information later on. 2. ‘Xlik2’ (‘I developed a real affection towards one or more of the main characters’) and ‘Mlik1’ (‘I enjoyed watching this movie’) were misfitting and poorly discriminating across all movies, thus marking them as likely candidates for deletion. Conceptually, it was believed that they could be sacrificed as there were other items capturing similar ideas, but with a different intensity. Indeed, the varying degrees of agree and disagree that respondents would use to answer those items still in the scale would compensate

 O              P     !  " # $% &  '    (*,0( for the deletion of these items. For example, strongly agreeing with the statement “I really liked one or more of the main characters” would likely be very similar to agreeing that you “developed a real affection towards one…. of the main characters”. 3. There were no significant DIF issues to be addressed.

The dimensionality issue was addressed first, since items were likely to behave differently when grouped differently. Tables 5.3.5.2 and 5.3.5.3 outline these results. Note that ‘notW2’ (‘I barely watched or listened to any of the movie’) was put back into the second dimension to see if it fitted better when grouped this way.

Table 5.3.5.2– Liking / Engaged dimension – 8 items Shooter Spiderman Perfect Mr Bean’s Stranger Holiday Disordered - - - - Thresholds PSI 0.840 0.895 0.861 0.818 02 probability 0.04 0.13 0.01 0.001 Item fit problems • Mlik1 – - • Xlik2 - 02, • Xlik2 – neg resid, F pos, 02, F F

Table 5.3.5.3– Not liking / Disengaged dimension – 4 items Shooter Spiderman Perfect Mr Bean’s Stranger Holiday Disordered - notW2 notW1 - Thresholds notW2 mind1 mind2 PSI 0.740 0.750 0.782 0.766 02 probability 0.0004 0.00006 0.006 0.15 Item fit problems • Mind1 - 02, • Mind1 - 02, • Mind2 - 02, - F F F • Mind2 - 02, • Mind2 - 02, F F

 O              P     !  " # $% &  '    (*,1( Since the fit was still so poor for the not liking / disengaged dimension and because there were still disordered thresholds, ‘notW2’ was again removed to see if this led to improved results. These appear in Table 5.3.5.4

Table 5.3.5.4– Not liking / Disengaged dimension without notW2 – 3 items Shooter Spiderman Perfect Mr Bean’s Stranger Holiday Disordered - - - - Thresholds PSI 0.656 0.703 0.589 0.706 02 probability 0.01 0.0002 0.36 0.36 Item fit problems • Mind2 - 02, • Mind1 - 02, - - F F • Mind2 - 02, F

As can be seen, both the PSI and overall chi square probability remain dubious. Removing ‘mind2’ (‘My mind was totally preoccupied with things other than the movie’) or ‘mind1’ (‘My mind wandered at times during the movie’) from this dimension led to even poorer results, presumably because of the lack of information the two remaining items were able to provide. Therefore, after careful consideration of the conceptual ramifications, it was decided to delete all the disengaged items, and have appraisal measured purely as a positively positioned construct. A low score on that scale featured in Table 5.3.5.2 would likely reflect someone that was disengaged with the movie and who did not like it.

There were still misfitting items in the eight-item liking/engaged version of the appraisal scale (see Table 5.3.5.2). Removing ‘Mlik1’ provided the best fitting and best discriminating scale for Shooter, Spiderman and Perfect Stranger (PSI > 0.81, chi square prob > 0.08), but gave a poor result for Mr Bean’s Holiday (PSI = 0.777, chi square prob = 0.0001). The best result for Mr Bean’s Holiday came from deleting ‘Xlik2’ (PSI = 0.834, chi square prob = 0.48), but this produced less than optimal results for the other movies. However, removing both ‘Xlik2’ and ‘Mlik1’ offered the best compromise, with all movies having good PSIs (>0.8) and reasonable fit statistics (see Appendix 5.21). It should also be

 O              P     !  " # $% &  '    (*,2( mentioned that attempts were also made to consider both dimensions as one and remove ‘Xlik2’, ‘Mlik1’, ‘notW2’ and then both ‘mind2’ and ‘mind1’ separately, but none of these solutions was strong across all movies. Therefore, it would appear that removing all the disengaged items, as had been done previously, was in fact necessary.

That said, due to the proposed changes of the response scale categories and the conflicting results, it was decided that it would be premature to formally delete any items from the scale at this stage. Hence, for the final confirmatory stage, all 12 items would still appear in case the new response scale altered how people answered the four ‘disengaged’ items and their performance improved.

5.3.6 – Results for Cognitive Effort Dimension All 9 items when analysed together formed a scale with good discrimatory power (PSI > 0.731) but with a low overall chi square probability (p = ~ 0). Further interrogation of the data revealed significant threshold disordering and several individual items that were under or over discriminating and/or misfitting (see Appendix 5.16).

As with the appraisal dimension, threshold ordering could be corrected by collapsing the six-point scale to a four-point scale. However doing this caused no great improvement in other aspects of the scale’s performance (see Table 5.3.6.1 and Appendix 5.17).

 O              P     !  " # $% &  '    (*-3( Table 5.3.6.1– Cognitive effort dimension with thresholds corrected – 9 items Shooter Spiderman Perfect Mr Bean’s Stranger Holiday Disordered - - - Easy1 Thresholds PSI 0.756 0.758 0.754 0.735 02 probability ~ 0.000 ~ 0.000 ~ 0.000 ~ 0.000 Item fit problems • Easy2 – • Easy2 – • Easy2 – • Easy2 – pos resid, pos resid, pos resid, pos resid, 02, F 02, F 02, F 02, F • Easy3 - • Easy3 - • Easy3 - 02, • Easy3 – pos resid, pos resid, F pos resid, 02, F 02 F • Attn 3 – 02, F • Attn1 - 02, • Attn 3 - 02, neg resid, • Attn1 - 02, F F F F • Attn2 - neg • Diff1 - neg • Diff1 - 02, • Attn2 - 02, resid, 02, F resid, 02, F F F • Diff1 - neg • Diff2 – neg • Diff2 - F • Attn 3 – resid, 02, F resid, F neg resid, • Diff2 - 02, • Diff3 – F 02, F F • Diff1 - neg resid, 02, F • Diff3 - 02, F Dimensionality t = 0.846 t = 0.860 t = 0.109 t = 1.722

There was still significant misfit with most individual items. Upon reflection however, it was decided that ‘easy2’ (‘I thought the storyline was simple’) and ‘easy3’ (‘I thought the storyline was too simplistic) were actually evaluations of the story’s complexity, and not statements that reflected the ease or difficulty of processing the story – which is what this dimension was supposed to capture. So for conceptual reasons, these items were the first to be deleted, providing the results outlined in Table 5.3.6.2.

 O              P     !  " # $% &  '    (*-*( Table 5.3.6.2– Cognitive effort dimension with thresholds corrected and easy2/3 removed – 7 items Shooter Spiderman Perfect Mr Bean’s Stranger Holiday Disordered - - - Easy1 Thresholds PSI 0.837 0.839 0.833 0.856 02 probability 0.00001 0.08 0.19 ~ 0.000 Item fit problems • Diff1 – • Diff1 – - • Easy1 - neg, 02, F neg, 02, F pos, 02, F • Diff3 – • Diff1 – pos, 02, F neg, 02, F Dimensionality t = 0.305 t = 0.838 t = 0.252 t = 0.788

Removing these items had a favourable impact on the Person Separation Index for each movie, and fit for Spiderman and Perfect Stranger. However, there was still poor fit for Shooter and Mr Bean’s Holiday. In Rasch analysis, items with positive residuals are deemed more problematic than items with negative residuals. This is because those with positive residuals under discriminate and the responses given are less predictable than what was expected. Therefore, when considering items for deletion, these items are generally the first to be flagged. Removing ‘diff1’ (‘I thought that parts of the storyline were difficult to understand’) was attempted, but as expected (because of its negative residual) results did not improve, so it was left in.

In contrast, removing ‘diff3’ (‘I thought that the storyline was totally incomprehensible’) did lead to improved results. Results improved even further when ‘easy1’ (‘I found the storyline clear and easy to understand’) was also deleted. The removal of ‘easy1’ was only temporary however and it would certainly appear in the next stage of testing. But since it was the only item measuring the ease of processing component, it could not provide all the information necessary to adequately represent this concept. This idea still needed to be measured, so it was recommended that for the final stage, additional items be added to capture this ease of processing and replace the ones that were deleted. But for now, to

 O              P     !  " # $% &  '    (*-+( demonstrate the strength of the other items, it would come out of the proposed scale, for which the results are outlined in Table 5.3.6.3 and Appendix 5.18.

Table 5.3.6.3– Proposed scale for cognitive effort with diff1/2, attn1/2/3 – 5 items Shooter Spiderman Perfect Mr Bean’s Stranger Holiday Disordered - - - - Thresholds PSI 0.832 0.805 0.836 0.858 02 probability 0.73 0.27 0.48 0.01 Item fit problems - - - • Diff1 – neg, 02, F • Attn1 – F • Attn3 – F Unidimensionality t = 1.299 t = -1.161 t = 0.076 t = 1.02

Despite this somewhat optimal solution, ambiguity surrounding this dimension remained for several reasons. Firstly, in the next stage, the new scale category labels and structure needed to be tested. Secondly, new items needed to be created to capture the idea of easy processing, and with ‘easy1’ prematurely removed from the scale at this stage, interactions between items may change in the next stage. Thirdly, Mr Bean’s Holiday achieved its optimal results with the removal of ‘diff1’, not ‘diff3’. Therefore, for the next stage, all items would be retained in the scale to be sure that no item was removed prematurely. Note that there were no problems to do with DIF for the cognitive effort scale.

5.3.7 – Modifications to the scale before Stage 4 Items to remove: Whilst Rasch analysis showed that ‘appalled’ and ‘cross’ should be removed from the feelings scale, they were kept for testing in the final stage so that data from this stage and the next stage could be aggregated. However, it was likely that these items would also perform poorly in the next stage and be removed. Similarly, although four items to do with being disengaged with the film were marked for deletion from the appraisal scale, and at least one from the cognitive effort scale, it was decided to keep them in for this next stage

 O              P     !  " # $% &  '    (*-,( in case the changes being made to the scale categories would affect their performance. Moreover, keeping them in and testing them on an additional film would give greater confidence in our findings.

Two items (‘easy2’ and ‘easy3’) were formally deleted from the cognitive effort scale because they failed to capture the concept of evoking little effort to understand the movie’s storyline.

Items to add / reword: These two deleted ‘easy’ items from the cognitive effort scale needed to be replaced with items that captured low cognitive effort more accurately. In doing so, ‘easy1’ was also reworded. Consequently, In place of the original three items, two new ones were created: ‘I had no problems following the storyline’ (‘easy2’) and ‘I found the storyline very easy to understand’ (‘easy1’).

Changes to scale categories: As stated, there was still disordering of thresholds in the agree-disagree scale used to measure appraisal and cognitive effort. Post-hoc collapsing of the categories corrected the threshold issue and allowed analysis in this stage, but the new four-point strongly disagree – disagree – agree – strongly agree structure needed to be formally tested with the collection of new data using these new categories in the next stage. Therefore, this response scale structure would undergo its third change to see if this would make the task of responding to the questionnaire easier for respondents and provide more reliable data.

 O              P     !  " # $% &  '    (*--( 5.4 – Confirmatory Test in High Schools – June 2007 5.4.1 – Sample and Implementation Following approval from the University's Human Ethics Committee, the Catholic Education Office (Sydney) was approached to obtain their support for schools under their jurisdiction to participate in the final stage of scale refinement / confirmation and testing of the brandcast processing model. Their support was gained, allowing the researcher to contact schools in the Sydney region. Personal and written contact was made to the Principals of these schools, and any questions they had were addressed by the researcher. The schools were asked if they would allow any of their students from Years 10-12 (aged 15-18) to participate in a study concerning how teenage audiences process the information and storylines contained within films, with a focus on the impact of product placement on vulnerable audiences (although this information was kept from the participants). It was at the Principal's discretion as to which year groups or classes (if any) they would allow to participate.

Two high schools agreed to participate – one boys school and one girls school. School A offered approximately 200 girls from Years 10 and 11 (aged 15-17), whilst School B offered approximately 85 Year 10 boys (aged 15-16). After removing questionnaires with high levels of missing data or those where the same response was circled for all questions, the final sample for this confirmatory scale testing comprised 193 girls and 80 boys. The study was run in a single session at each school in order to minimise communication effects. The movie (The Island) (see Section 6.5.2) was shown in both locations in a setting that was as close to real cinema environment as possible, namely a large hall with a cinema-style projection screen and excellent viewing facilities (see Section 6.8.2).

 O              P     !  " # $% &  '    (*-.( 5.4.2 – Snapshot of AUDENG scale tested at Stage 444 Number Number of Dimension of items response scale Response scale category labels categories Feelings 21 3 Never, At times, Throughout the movie Arousal 21 3 Never, At times, Throughout the movie Appraisal 12 4 Strongly disagree, Disagree, Agree, Strongly agree Cognitive Effort 8 4 Strongly disagree, Disagree, Agree, Strongly agree

Total number of item = 62

5.4.3 – Results for Feelings Dimension Together, all 21 items had an excellent Person Separation Index (PSI = 0.867) but very poor fit (overall chi square prob = 0.00002). Specifically, ‘appalled’ under discriminated and had significant chi square and F statistic probabilities, whilst ‘distraught’ also had a significant F statistic probability. All thresholds were properly ordered.

Based on the results from the previous stage, ‘appalled’ was the first item to be removed. Doing so saw the PSI rise to 0.872 and the overall chi square probability rise to 0.004. However, ‘distraught’ still had a significant F statistic probability, and there were several DIF issues regarding gender for ‘good’, ‘angry’, ‘scared’, ‘sad’ and ‘comforted’.

To attempt synergy with the proposed model highlighted in the previous stage, ‘cross’ was removed, even though it was causing no apparent problems to this scale. Following this, results improved again (PSI = 0.870; chi square prob = 0.02), but the DIF issues remained.

44 NB. The number of class intervals set for this analysis was five (under the assumption that each class interval should contain at least 40-50 respondents).  O              P     !  " # $% &  '    (*-/( Next, ‘scared’ and ‘sad’ were split according to gender, as DIF analysis highlighted that they were being responded to differently depending on the respondent’s gender. In keeping with the proposed model from the previous stage, just these two items were split. This led to the Person Separation Index stabilising (PSI = 0.867) and the overall chi square probability improving (p=0.05). ‘Distraught’ was still misfitting, and removing it saw the PSI remain stable at 0.861, but overall fit improve dramatically (overall chi square prob = 0.23). Gender differences also existed for ‘angry’ and ‘comforted’, However, in the interests of generalisability, ‘distraught’ was put back in since the previous solution still had excellent results, and applied to all movies, and similarly, the remaining DIF issues were noted, but they were not acted on (see Appendix 5.19).

Knowing that this scale now fitted five movies from this stage and the previous stage, and that there were no significant violations of local independence (since few residual correlations >0.3) for any of the five movies, this proposed scale was used to revisit the Babel and Wild Hogs data (see Appendix 5.19). It provided excellent results for both, with both Babel and Wild Hogs having a PSI of 0.849 and an overall chi square probability of 0.501. Whilst ‘good’ was under-discriminating in both movies, it should be noted that removal of ‘good’ in both this confirmatory stage with The Island, and in the previous stage with the other four movies saw no improvement in results.

Item maps (see Appendix 5.19) revealed that respondents still generally had difficulty endorsing these feeling items. The person mean location for The Island was -2.191, and there was a group of respondents at the bottom of the map with no items to match them in location. Perhaps this group reflects people who either felt nothing whilst watching the movie, or who could not articulate these feelings whilst completing the questionnaire, or did not want to. That said however, the bulk of people were matched by items, although there were some high level feelings and high intensity categories of feelings that did not correspond to anyone (e.g. highest category of ‘depressed’ and ‘horrified’). Interestingly, the highest category of ‘sad’ and ‘scared’ for males had no people matching them. These results were echoed in the maps of Shooter, Spiderman, Perfect Stranger and Mr Bean’s  O              P     !  " # $% &  '    (*-0( Holiday – negative person mean locations (ranging from -2.380 to -1.871), and high level feelings not being felt by any respondents. Despite this, ICCs for all items across all movies were very good, with all class interval averages generally in accord with model predictions. Item ordering across all movies tended to be as hypothesised, with two consistent exceptions. ‘I felt miserable’ was consistently a more difficult to endorse item than ‘I felt totally depressed’. Furthermore, ordering between ‘I felt heartened’, ‘I felt comforted’ and ‘I felt uplifted’ was muddled, with different ordering for all movies. These items were left in this present model due to excellent fit and unique item locations, but future research might consider re-wording or removing one of these items if item hierarchy is deemed a more important criterion of the research.

 O              P     !  " # $% &  '    (*-1( Figure 5.4.3.1 - Item Map for Feelings dimension – The Island ------LOCATION PERSONS ITEMS [locations] ------3.0 | | | | | IMsca 2.0 | | IMsad | | miser | 1.0 | | | o | dsmay dstra depre o | horif elate 0.0 o | terif raged oooooo | comft dstrs o | uplif ooooo | IFsad heart IFsca oooooo | angry -1.0 oooooo | good oooo | happy ooooooo | oooooooooo | cncer ooooooooooo | edge -2.0 oooooooooo | oooooooooooooo | ooooooooo | ooooo | ooooooooooo | -3.0 ooooo | ooooo | o | ooo | oooo | -4.0 oo | | oooo | o | | -5.0 | oooooo | ooooo | | | -6.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '    (*-2( 5.4.4 – Results for Arousal Dimension Three different solutions were examined here, reflecting the different results that have been obtained or suggested during the three previous research stages. Whilst greater weight was given to the solution arrived at in Stage 3, to demonstrate rigour, the other solutions were also attempted.

Using the original coding system (as per Stages 1 and 2), all 21 items analysed together resulted in a Person Separation Index of 0.536 and an overall chi square probability of virtually 0. Four items had disordered thresholds, with six others showing considerable misfit. Tests for multidimensionality revealed arousal to be comprised of two dimensions – high and low (t=-2.55) and the following results achieved (see Table 5.4.4.1).

Table 5.4.4.1– Results for Low vs High Arousal (as per pilot study) Low Arousal High Arousal Disordered - Tears Thresholds Cried PSI 62.8% 69.9% 02 probability 0.0001 0.02 Item fit problems • Calm – 02, F - • Fidget – F DIF • Ease – seenb4 • Anxious – seenb4 • Calm – CI, seenb4 • Relax – seenb4 • Look – seenb4 Unidimensionality t = -0.68 t = -0.02

These results, especially for low arousal, were disappointing, so the three-dimension solution was attempted, with the following results achieved (see Table 5.4.4.2).

 O              P     !  " # $% &  '    (*.3( Table 5.4.4.2– Results for Positive Arousal vs Negative Arousal vs Bored (as per Stage 2)

Positive Arousal Negative Arousal Bored Disordered Thresholds - Cried - PSI 28.7% 72.3% 70.7% 02 probability 0.0008 0.20 0.02 Item fit problems • Calm – F - • Fidget - F • Lhard – 02, F DIF • Ease – gender - • Fidget – CI • Calm – CI, • Yawn - gender gender • Relax- gender • Smile - gender • Laugh – gender • Lhard – CI, gender Unidimensionality t =0.55 t = 2.77 t = 0.17

Again, the results relating to the positive arousal items were very poor. No degree of item deletion led to improved results. Hence, the data were recoded as per the latter stages of Stage 3 and the analysis attempted again.

As with Stage 3, the results improved dramatically taking this approach. The final 10-item unidimensional solution (t=1.06) had a Person Separation Index of 0.778 and an overall chi square probability of 0.04 (see Appendix 5.20). There were no misfitting items or items suffering from DIF. Item-person matching on the item maps was excellent, and local independence strong (as seen in the residual correlation matrices for all movies). ‘I was close to tears’ and ‘I cried’ both had disordered thresholds, but this is a result of low frequencies across the range of response categories as this film is not one that would evoke such a reaction.

This 10-item scale was also tested on Babel and Wild Hogs to confirm its generalisability (see Appendix 5.20). Initial results confirmed that ‘I was in a dream-like state’ was still a

 O              P     !  " # $% &  '    (*.*( problematic item, even with this new structure and coding, so one can feel confident that its removal and re-wording of those triplets was correct. Assessing a 9-item version of the arousal scale, results confirm that for both movies, arousal is unidimensional (Babel – t=0.969; Wild Hogs – t=-0.228). There were no misfitting items or disordered thresholds, and overall fit and discrimination was good (Babel – PSI = 0.723; chi square prob = 0.99; Wild Hogs – PSI = 0.569; chi square prob = 0.4).

Overall, there was some mismatching between people and items, as reflected by negative person mean locations (ranging from -3.215 to -1.250). As seen on the item maps, ‘I cried’ and ‘my eyes welled up with tears’ are the likely cause for this (being too intense for people to agree with), combined with some people having no items to endorse (i.e. no items of low intensity). Item ordering was as expected, although ‘I was at ease’, ‘I was calm’ and ‘I was totally relaxed’ often appeared in different orders for each movie. However, all three items remained in this present research due to good fit and reasonably unique item locations.

 O              P     !  " # $% &  '    (*.+( Figure 5.4.4.1 - Item Map for Arousal dimension – The Island ------LOCATION PERSONS ITEMS [locations] ------3.0 | | | | | 2.0 | cried o | | oo | tears o | 1.0 | oo | oo | ooooo | unezy | 0.0 oooooo | grab oooooooo | tense | jumpy oooooooooooooo | oooooooooo | anx calm -1.0 | ooooooooooooo | | relax ease ooooooooooooo | | -2.0 oooooooooooooo | | oooooooooooooo | o | oooooooooooooo | -3.0 | o | | oooooooooooo | | -4.0 | | ooooooo | | | -5.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '    (*.,( 5.4.5 – Results for Appraisal Dimension Examination of all 12 appraisal items together revealed an excellent Person Separation Index (PSI = 0.908) but a significant overall chi square probability (p = ~ 0). The new four- point agree-disagree scale resulted in no disordered thresholds for the first time in any of the research stages and confirmed the post-hoc collapsing of six categories to four in the previous stage. Therefore, any model misfit issues were due to individual items misfitting, and as in previous stages, these were ‘Xlik2’ (‘I developed a real affection towards one or more of the main characters’), ‘Mlik1’ (‘I enjoyed watching this movie’) and ‘notW2’ (‘I barely watched or listened to any of the movie’).

As with most of the movies in the previous stage, appraisal was found to be a unidimensional construct (t = 0.30). Separating it into separate engaged / disengaged dimensions (as per Stage 3) saw the four-item disengaged dimension perform very well (PSI = 0.806, chi square prob = 0.22) with no item fit problems, whilst the eight-item engaged dimension had an excellent Person Separation Index (PSI = 0.888) but poor overall fit (overall chi square prob = 0.0001). This result contradicts those in the previous stage where the disengaged dimension was so problematic that it was recommended that those items be dropped from the final appraisal scale.

Looking at the overall 12 item scale, ‘notW2’ was removed first since it had been the most problematic in previous stages and this led to a rise in overall chi square (p = 0.00012), with the Person Separation Index remaining stable (PSI = 0.906). ‘Xlik2’ and ‘Mlik1’ still misfitted. Removing those two items resulted in a still excellent PSI of 0.886 and an acceptable overall chi square result (p=0.03).

However, to be consistent with the overall findings from the previous stage, greater interrogation of the eight liking/engaged items took place. As stated, together they had an excellent PSI of 0.888 but poor fit, with ‘Xlik2’ and ‘Mlik1’ again being problematic. Removing just ‘Xlik2’ actually resulted in poorer fit (chi square prob = 0.005) and a stable PSI of 0.886. In contrast, removing just ‘Mlik1’ saw fit improve significantly (chi square  O              P     !  " # $% &  '    (*.-( prob = 0.29) and the PSI remain strong (PSI = 0.862). Removing both items also led to strong results (PSI = 0.853, chi square prob = 0.36). Local independence was good (i.e. residual correlations for all films tended to be <0.3).

To identify the best general scale for appraisal, Table 5.4.5.1 was developed. This summarises the results for appraisal from both this stage and the previous stage. A seven- item scale without ‘Mlik1’ provides very good results for all movies except for Mr Bean’s Holiday. Mr Bean’s Holiday’s best fitting model comes from removing ‘Xlik2’. However, in the interests of developing a scale which will best fit any movie, the final six-item solution (excluding both ‘Mlik1’ and ‘Xlik2’) should be the best (see Appendix 5.21). All movies achieved good results using this scale (although the fit for Shooter is marginal). Crucially, it excludes both the items that were potentially problematic, so is likely to cause no problems in future applications. To test this idea further, this scale was applied to Babel and Wild Hogs (the movies from Stage 2) (see Appendix 5.21). Neither movie had misfitting items using this model, and both had strong Person Separation Indexes (Babel = 0.762; Wild Hogs = 0.880) and good overall fit (Babel’s chi square prob = 0.42; Wild Hogs’ chi square prob = 0.78) despite disordered thresholds (as these could not be converted to the new scale structure).

 O              P     !  " # $% &  '    (*..( Table 5.4.5.1– Potential general appraisal scales Shooter Spiderman Perfect Mr Bean’s The Island Stranger Holiday All 8 ‘liking’ PSI 0.840 0.895 0.861 0.818 0.888 items 02 probability 0.04 0.13 0.01 0.0006 0.01 2 Item fit probs • Mlik1 – neg - • Xlik2 - 0 , F, • Xlik2 – pos, • Mlik1 – neg, resid, F, DIF DIF 02, F, DIF 02, F, DIF • Xlik2 – pos DIF - - • Mlik3 - age • Immersed – • Wait – CI x ed seenb4 7 liking items PSI 0.840 0.890 0.863 0.834 0.886 (without 02 probability 0.008 0.03 0.31 0.48 0.005 Xlik2) Item fit probs • Mlik1 – neg • Wait – neg, - - • Mlik1 – neg, resid, F, DIF F, DIF 02, F, DIF 2 • Mlik2 - 0 , • Mlik1 – F, F, DIF DIF DIF • Immersed - - • Mlik3 - age • Immersed – • Wait – age CI x ed seenb4 7 liking items PSI 0.812 0.877 0.837 0.777 0.862 (without 02 probability 0.53 0.08 0.21 0.0001 0.29 Mlik1) Item fit probs - - - • Xlik2 – pos, • None 02, F, DIF DIF - - • Interested – • Immersed – • Wait – gender CI x ed seenb4 • Mlik3 - age 6 liking items PSI 0.807 0.868 0.837 0.799 0.853 (without Xlik2 02 probability 0.01 0.50 0.82 0.88 0.36 or Mlik1) Item fit probs - • Wait – F, - - - DIF DIF • Immersed - - - • Immersed – • Wait – age CI x ed, seenb4 critic

 O              P     !  " # $% &  '    (*./( Significantly, analysis of The Island suggested that whether the respondent had seen the movie previously affected their response to ‘I couldn’t wait to see what happened next’ (‘wait’). This was the only movie for which this “seen before” variable could be reliably tested, as the other movies were tested in a real-life cinema setting, so few people had seen the movie previously, thus making any DIF results unreliable (see Table 5.4.5.2). In contrast, The Island was released in 2005 so many respondents had had the opportunity to see it either at the cinema or on DVD.

Table 5.4.5.2 – Previous exposure to the movie Movie Number of Respondents who had seen the movie before Mr Bean’s Holiday 6/150 Perfect Stranger 1/145 Shooter 3/256 Spiderman 4/262 The Island 89/273

N.B. Base sample figures relate to number of people who answered the question, not how many actually saw the movie.

Differences between those that had seen the movie before and those that had not makes sense in regards to this item. Therefore, it is recommended that in future use of this scale, these two groups be split, and tested to see whether significant differences in their results exist.

Looking more deeply into the six-item scale’s performance across all the movies, the person mean locations and item maps revealed that matching between the items and respondents was very good. All person mean locations were between 0.031 and 1.247, suggesting that if anything, items were easily agreed to, with a few extreme people for each movie not having items strong enough to endorse. However, it is important to note that the bulk of people were matched and that there was an excellent spread of both people and

 O              P     !  " # $% &  '    (*.0( items. Only in a couple of cases did two item thresholds sit at the same location. Perhaps the most interesting insight to emerge from the item maps and locations was the order of difficulty of the three items ‘I was interested in the storyline’, ‘I couldn’t wait to see what happened next’ and ‘I felt completely immersed in the story’. It was anticipated that these items would appear in this order on the maps, in ascending order of difficulty. But for all movies except for Wild Hogs, Shooter and The Island, this order was muddled. Certainly from a frequency perspective, ‘interested’ most commonly appeared as the easiest item (five movies out of seven), followed by ‘wait’ (three movies out of seven) and then ‘immersed’ as the hardest item (five movies out of seven), but more often than not, the actual hypothesised order failed to eventuate.

 O              P     !  " # $% &  '    (*.1( Figure 5.4.5.1 - Item Map for Appraisal dimension – The Island ------LOCATION PERSONS ITEMS [locations] ------6.0 | | | oo | | 5.0 | | | | oo | 4.0 | | o | oooo | | 3.0 | | ooooooooooo | | ooooooooooo | 2.0 oo | | ooooooooooooo | | Mlik3 oooooooooooooooo | Mlik2 1.0 o | | oooooooooooooo | | oooooooooooooooooo | 0.0 | imers oooooooooooo | | | Xlik1 ooooooo | wait -1.0 | ooooooo | | inter oooooooo | | -2.0 ooo | | oooo | | oo | -3.0 | | oo | | | -4.0 oo | | | | ooo | -5.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '    (*.2( 5.4.6 – Results for Cognitive Effort Dimension Assessing all eight items together, the revised cognitive effort dimension was found to be unidimensional (t = -0.05), with a very good Person Separation Index (PSI = 0.832) but a significant overall chi square probability (p=0.00002) (see Appendix 5.22). As with appraisal, the revised scale category structure and labels were successful, with thresholds for seven of the eight items ordered. Significantly, the one item whose thresholds remained disordered, diff3 (‘I found the storyline totally incomprehensible’) was also the only item to have poor fit (resid = 5.532, chi square prob = ~ 0, F stat prob = ~ 0). It was felt that there could be several reasons why this item was misfitting. Firstly, when administering the scale in the schools, there was some confusion as to what the word ‘incomprehensible’ meant. If respondents did not know its meaning, it makes sense then that the responses were haphazard. Secondly, the word ‘incomprehensible’ can be interpreted in several ways – extremely difficult to understand, unfathomable, unrealistic, unbelievable, or even immoral. These alternative definitions are all relevant to this movie, and if interpreted in this way would make this item capture a different idea to what was intended. Therefore, because of its ambiguity, and because it had also been problematic in the previous research stage, the final cognitive effort scale will not include this item. Indeed, results improved following its deletion (PSI = 0.845, chi square prob = 0.94) and the scale remained unidimensional (t = - 0.37) (see Appendix 5.23). Item maps revealed that there was an excellent match of items and people, and the person mean location (-0.546) was well within the healthy range of +/- 1. The residual correlation matrix showed that were no major violations of local independence.

It is impossible to confirm these results using movies from other stages since this cognitive scale features new and amended items. What examination of these maps does show however is the need for items that people who had no difficulty processing the movie can endorse. Without these items, all person mean locations were negative, ranging from -0.35 to -2.259. Moreover, many respondents were not in similar locations to items. Therefore we can see that the addition of the two new items to capture easy processing dramatically

 O              P     !  " # $% &  '    (*/3( improved the distribution of items and people. As per the last stage, there were no differences in responses between sub-groups (i.e. there was no DIF present).

Figure 5.4.6.1 - Item Map for Cognitive Effort dimension– The Island ------LOCATION PERSONS ITEMS [locations] ------5.0 | | | | | 4.0 | | | | | 3.0 | | | | o | 2.0 | ooo | | ooooo | | 1.0 ooooooo | | diff2 ooooooo | easy2 o | ooooooo | 0.0 | easy1 diff1 attn3 ooooo | | attn2 oooooooooooooo | | -1.0 | ooooooooooo | | attn1 ooooo | | -2.0 oooooo | | oooooo | | oooo | -3.0 | oo | | | ooo | -4.0 | | | oo | | -5.0 | | o | | ------o = 3 Persons  O              P     !  " # $% &  '    (*/*( 5.5 – Final Audience Engrossment scale This development and refinement of the scale confirmed that audience engrossment is comprised of four dimensions – feelings, arousal, appraisal and cognitive effort. The final scales for each of the four dimensions appear below. Note that (R) next to an item reflects that it is reverse coded.

Arousal – 10 items PART 1

Firstly, we would like you to tell us how you reacted DURING the movie. Never movie For EACH STATEMENT, please circle the number that best describes your At times

reactions. the Throughout I was anxious 1 2 3 I was calm (R) 1 2 3 I was uneasy 1 2 3 I was close to tears (e.g. eyes welled up with tears / there was a lump in my throat) 1 2 3 I needed to grab hold of something (e.g. the person next to me, the armrest) 1 2 3 I cried 1 2 3 I was at ease (R) 1 2 3 I felt jumpy 1 2 3 I was totally relaxed (R) 1 2 3 I was tense 1 2 3

 O              P     !  " # $% &  '    (*/+( Feelings – 19 items PART 2

In this section, we would like you to tell us how you felt DURING the movie. Never movie

At times For EACH STATEMENT, please circle the number that best describes your Throughout the the Throughout feelings. The movie made me feel totally depressed 1 2 3 The movie made me feel scared 1 2 3 The movie made me feel happy 1 2 3 The movie made me feel angry 1 2 3 The movie made me feel distressed 1 2 3 The movie made me feel heartened 1 2 3 The movie made me feel miserable 1 2 3 The movie made me feel horrified 1 2 3 The movie made me feel good 1 2 3 The movie made me feel dismayed 1 2 3 The movie made me feel comforted 1 2 3 The movie made me feel elated 1 2 3 The movie made me feel on edge 1 2 3 The movie made me feel concerned 1 2 3 The movie made me feel terrified 1 2 3 The movie made me feel enraged 1 2 3 The movie made me feel distraught 1 2 3 The movie made me feel sad 1 2 3 The movie made me feel uplifted 1 2 3

 O              P     !  " # $% &  '    (*/,( Appraisal – 6 items PART 3

Now we would like to find out your thoughts about the movie.

Agree For EACH STATEMENT, please circle the number that best describes Disagree your thoughts. Note that these questions ask you to agree or disagree Agree Strongly Strongly Disagree Strongly with each statement. I could happily watch unlimited re-runs of this movie 1 2 3 4 I was interested in the storyline 1 2 3 4 I couldn't wait to see what happened next 1 2 3 4 I felt completely immersed in the story 1 2 3 4 I enjoyed watching this movie more than I have most others 1 2 3 4 I really liked one or more of the main characters 1 2 3 4

Cognitive Effort – 7 items PART 4

In this section we would like to know how you made sense of the movie. Agree

Disagree

For EACH STATEMENT, please circle the number that best describes Agree Strongly your response. Like the last section, these questions ask you to agree Disagree Strongly or disagree with each statement. I thought that overall the storyline was difficult to understand 1 2 3 4 There were times I needed to pay attention to follow the story 1 2 3 4 I found that following this story was mentally demanding 1 2 3 4 I had no problems following the storyline (R) 1 2 3 4 I thought that parts of the storyline were difficult to understand 1 2 3 4 Generally I had to concentrate to follow the story 1 2 3 4 I found the storyline very easy to understand (R) 1 2 3 4

 O              P     !  " # $% &  '    (*/-( 5.6 – Assessing Generalisability of the Audience Engrossment scale This Audience Engrossment scale demonstrated various levels of generalisability. Not only does this prior discussion show invariance between viewers of the same movie, but it also identified items, dimensions and a model that could be used to effectively measure the level of audience engrossment of viewers across seven different movies.

Generally the items had similar item locations and similar ordering of difficulty regardless of the film, particularly for arousal45 and appraisal (see Appendix 5.24)46. This is demonstrated by the clustering of a particular item across the five movies47 around a particular item location range. However, some items, most notably in the feelings dimension, behaved differently for different movies, and covered a wide range of item locations. This is not altogether unexpected when one considers the content and themes of different movies. For example, the item ‘I felt happy’ was much easier to endorse in Mr Bean’s Holiday (item location = -3.611) than in Babel (where it was considered a very difficult item to agree with and generated a higher item location of 1.345). Since DIF analysis revealed that performance of items did not differ according to person factors48, item location is not dependent on any of these factors. It is therefore likely that any differences in responses are dependent on the film. Such findings reflect that there are qualitative differences between genres and films, most notably in relation to evoked feelings. They also suggest that for the feelings dimension, a model that fits all people and all films may actually be difficult to achieve. Further research will confirm this.

45 The varied locations for ‘tears’ and ‘cried’ (in the arousal dimension) may be attributed to low frequency counts for these items and the disordered thresholds that resulted from these low counts. 46 Plots could not be done for cognitive effort since the best items and response scale categories were only determined in Stage 4 (The Island) so there was no data to compare these to. 47 Because of the changes made to the response scale category structures, this analysis could only really be conducted on the movies from Stage 3 and 4 to maximise comparability of the data (i.e. Shooter, Spiderman, Mr Bean’s Holiday, Perfect Stranger and The Island). 48 Person factors that were tested for in Chapter 5 were: seen the movie before, age, gender, movie-going frequency, education and degree of critical expertise. There was some minor gender DIF for two feelings items, but this was corrected for.  O              P     !  " # $% &  '    (*/.( Further insights into the generalisability of the scale were gained by plotting item locations for movies against each other (see Appendices 5.25 and 5.26) A “worst case” (the two most different movies in terms of item locations) and “best case” (the two most similar movies in terms of item locations) approach was taken for the plots. Under the “best case” situation (The Island versus Shooter) practically every item in all three dimensions fell within (or close to) the 95% confidence interval (see Appendix 5.25) indicating a similarity in location for each item. Even under the “worst case” situation (i.e. Mr Bean’s Holiday versus Perfect Stranger), the majority of items fell within a 95% confidence interval range (see Appendix 5.26) especially for arousal and appraisal49. Extending these confidence intervals to 99% saw even better results. These findings provide greater confidence in the data and provide a greater level of generalisability than simply looking at the results of each movie in isolation.

5.7 – Validity of the Audience Engrossment scale For a new test to serve as a valid measure of anything, it is necessary that one first demonstrates reliability (Nunnally 1978). Examination of the Person Separation Indexes confirms that each dimension of the AUDENG scale is reliable (see Table 5.7.1).

49 Again, these plots could not be done for cognitive effort since the best items and response scale categories were only determined in Stage 4 (The Island) so there was no data to compare these to.  O              P     !  " # $% &  '    (*//( Table 5.7.1 – Person Separation Indexes per movie, per dimension Feelings Arousal Appraisal Cognitive Effort50 Babel 0.849 0.72351 0.76252 Cannot compute Wild Hogs 0.849 0.56953 0.88054 Cannot compute Shooter 0.797 0.749 0.807 Cannot compute Spiderman 0.816 0.773 0.868 Cannot compute Perfect Stranger 0.857 0.746 0.837 Cannot compute Mr Bean 0.770 0.674 0.799 Cannot compute The Island 0.867 0.778 0.853 0.845

Since validity is concerned with the meaning of measures of a latent variable estimated from observed manifest variables, the measurement theory used cannot be separated from validity. The very process of using Rasch measurement theory to develop the Audience Engrossment scale has ensured it has met standards of validity much higher than other methods (Salzberger 2007), especially compared to that tested when using the C-OAR-SE procedure which “relies totally on content validity” (Rossiter 2002, p236). Indeed, Rasch analysis offers a “comprehensive view of validity which integrates considerations of content, criteria and consequences into a construct framework for empirically testing rational hypotheses about score meaning and utility” (Messick 1995)

In this study, content validity is assumed due to the exhaustive efforts involved in construct definition and item generation, including the use of expert reviews (Wolfe and Smith 2007). Furthermore, Rasch analysis provides a superior foundation for assessing content and construct validity as it gives insight into what various levels of the construct actually mean by telling us which items are indicating the construct and also what levels of the construct they represent (i.e. where they are located on the dimension) (Salzberger 2000;

50 This statistic cannot be computed using the optimal cognitive effort scale for any movie except for The Island, due to changes in scale response categories between Stage 2 and 4 and the deletion and rewording of items. 51 This statistic is based on a 9-item version of arousal (see discussion in Section 5.4.4) 52 This figure cannot be corrected for disordered thresholds due to changes in scale response category structure between Stages 2 and 4. 53 This statistic is based on a 9-item version of arousal (see discussion in Section 5.4.4) 54 This figure cannot be corrected for disordered thresholds due to changes in scale response category structure between Stages 2 and 4.  O              P     !  " # $% &  '    (*/0( Salzberger et al. 2001; Wolfe and Smith 2007). In contrast to the traditional understanding of measurement, content validity and construct validity are much more inter-related because the item content determines the item location (Ewing et al. 2005).

Indeed, the basis of construct validity is a strong theory which proposes relationships between constructs. If the empirical relationships corresponding to the theoretical relationships exist in the data, then this is evidence of construct validity. Therefore, fit to the Rasch model demonstrates both construct and substantive validity (Wolfe and Smith 2007). Fit also means that all model properties hold and that comparisons of respondents and items are specifically objective within the frame of reference, thus providing generalisability (Salzberger 2007). As stated in Section 5.6, generalisability was further demonstrated through the use of multiple samples and multiple movies. Substantive validity was also demonstrated through response scale analysis, with changes to these cross-validated both post-hoc and then with new data (Wolfe and Smith 2007).

Traditional scale analyses rely on inter-item correlations and qualitative reasoning in terms of content validity. Substantial correlations with other scales and related constructs (such as the antecedents and consequences of the construct under scrutiny) represent additional evidence of construct (convergent) validity (Churchill 1979). However, in Rasch measurement theory, there is no particular emphasis on correlations between manifest item scores since as stated, construct validity is a direct consequence of model fit and sufficient variation in terms of item locations.

That said, convergent validity was assessed to see whether the person scores relating to audience engrossment correlated highly with the factor scores for Program Liking (Murry et al. 1992) – a scale it should be theoretically related to (see Section 3.7 and Appendix 6.8). Whilst audience engrossment is a higher level and more encompassing measure, one would assume that if individuals are engrossed in the entertainment program they are consuming, they would also like it, and that they want to be engrossed. Therefore, a high

 O              P     !  " # $% &  '    (*/1( correlation between program liking and the appraisal dimension of audience engrossment should occur, but other audience engrossment dimensions may also correlate.

Correlation analysis of each respondents scores on the Program Liking factor and the four Audience Engrossment dimensions revealed that Program Liking was significantly and positively related to appraisal (r=0.86, p=0.000), feelings (r=0.42, p=0.000) and arousal (r=0.27, p=0.002). It was also negatively related to cognitive effort (r=-0.17, p=0.052). Therefore, it can be said that there is strong convergent validity between Program Liking and aspects of the Audience Engrossment Scale, particularly appraisal and feelings.

5.8 – Conclusion This chapter described the steps in refining the Audience Engrossment scale from 81 items with hypothesised dimensions to a 42-item scale comprising four clear dimensions – feelings, arousal, appraisal and cognitive effort. It explained how and why different items and response scale categories were deleted or re-worded and demonstrated that these changes were effective in capturing the underlying dimensions. ICCs reinforced that items had good fit to the model, with endorsability increasing consistently with an increase presence of each characteristic. This model fit gives the scale both construct and content validity. Convergent validity was also demonstrated.

 O              P     !  " # $% &  '    (*/2( CHAPTER 6: TESTING THE BRANDCAST PROCESSING MODEL

“Plans are only good intentions unless they immediately degenerate into hard work”

Peter Drucker

6.1 - Introduction With Chapter 5 detailing the steps in refining the Audience Engrossment scale, we are now in a position to begin exploring the impacts of audience engrossment (via its four underlying dimensions) on product placement processing, thus testing part of the model presented in Chapter 3. This model testing is a preliminary investigation of the moderating role of both audience engrossment and brand familiarity on memory for a product placement, and to understand whether they behave as predicted. Specifically to audience engrossment, this involves investigating the effects of each dimension on product placement recognition. The direct effects of product placement characteristics on recognition will also be examined.

The chapter starts with a brief statement of hypotheses (based on the theory described in Chapter 3). The research strategy will then be discussed, and the use of a quasi-experiment and teenage sample justified. Next, the various stages of the research process are explained, namely how the film which served as the stimulus underpinning the study was selected, development of the questionnaire instruments, and the implementation of the pre- and post- film testing. Following this, the results of the analysis and hypothesis testing are presented.

6.2 – Research Questions and Hypotheses Up until now, product placement research has tended to focus on the effects that product placements have on the audience (e.g. Babin and Carder 1996; Gupta and Lord 1998; Brennan et al. 1999; d'Astous and Chartier 2000; Russell 2002). Are they remembered? Do

 O              P     !  " # $% &  '    (*03( changes in attitude or preferences occur? Which executional variables optimise the positive effects of product placement?

As stated previously, the argument that this thesis takes is that it is not what product placements (or indeed any form of brandcast) do to audiences, but what audiences do with them that governs their success or failure. Whilst practitioners can have some control in the executional factors of the brandcast, it is harder for them to control what the audience does with these brandcasts. Therefore, there is an important role for researchers to consider the role the audience member has in processing brandcasts, and share this knowledge with practitioners in a useful way. Hence, this research has questioned the role that an individual’s level of engrossment with a film has with the processing of product placements contained within it, as well as their level of familiarity with the brands featured in the film.

Specifically, the following research questions will be addressed (see Section 1.4): 1. Are higher quality product placements55 more likely to be remembered than lower quality product placements? 2. Does an individual’s level of brand familiarity affect their conscious memory for a product placement of that brand contained within a film? 3. Do viewers experiencing different levels of engrossment have the same ability to process product placements? Will their conscious memory for the placed brands be the same?

Hence, based on the theory outlined in Sections 3.5 and 3.8, the following broad hypotheses may be formed:

• H1: Each of the product placement characteristics will lead to higher recognition of the brands featured in the film.

55 By high quality, we refer to placements that may be more prominent, have high plot connection, appear aurally and visually, or be used by the star. Prior research and theory suggests these should all have significant impacts (see Section 3.5).  O              P     !  " # $% &  '    (*0*( 56 o H1a: Higher temporal quality will have a positive impact on recognition of a brand featured in the film.

o H1b: Recognition will be higher if a brand’s placement has high prominence than if a brand’s placement has low prominence.

o H1c: Recognition will be higher if a brand is used by a star than if a brand is not used by a star.

o H1d: Recognition will be higher if a star is present alongside a brand than if a star is not present alongside a brand.

o H1e: Recognition will be higher if the brand has high plot connection than if a brand has low plot connection.

o H1f: Recognition will be higher if the actual brand is depicted than if the actual brand is not depicted.

o H1g: Recognition will be higher if the brand's logo / advertising / signage is depicted in the film than if the brand's logo / advertising / signage is not depicted.

o H1h: Recognition will be higher if a brand's placement is by dual modality than if a brand's placement is by single modality.

• H2: Brandcasts with high brandcast quality will be better remembered than those with low brandcast quality.

• H3: When brand familiarity is high, each product placement characteristic will have a stronger positive impact on recognition than when brand familiarity is low.

• H4: When feelings are high (i.e. more intense), each product placement characteristic will have a stronger negative impact on recognition than when feelings are low (i.e. less intense).

• H5: When arousal is high, each product placement characteristic will have a stronger negative impact on recognition than when arousal is low.

56 By temporal quality, we refer to both “total length of time on screen” and “frequency of exposure” of the brand within the movie.  O              P     !  " # $% &  '    (*0+( • H6: When the appraisal of the movie is high, each product placement characteristic will have a stronger positive impact on recognition than when the appraisal is low.

• H7: When cognitive effort is high, each product placement characteristic will have a stronger negative impact on recognition than when cognitive effort is low.

6.3 – Research Strategy The focus of this research on familiarity and recognition essentially directed the structure of this stage as it required a questionnaire-based, multi-stage matching procedure using the same respondents. The first stage assessed the degree of familiarity of each respondent with a set of brands. The second stage required that the same respondents be exposed to a film (in as naturalistic a setting as possible) and then tested to see what brands they could recognise from a list as being in the movie they just saw. The Audience Engrossment scale and some questions measuring their liking of the film were also administered. To avoid contamination, a period of four weeks between the first and second stage was imposed.

Since the study required multiple tightly controlled field exercises to take place in a context that approximated a natural setting, a field-based, quasi-experimental research design was adopted57. Field research without the constraints and controls of a (quasi) experiment, although maximising generalisability, would be a less effective approach. Such data collection would have to take place at a cinema complex and the movie would have to have been seen in advance by the researcher and the product placements coded beforehand. Furthermore, by using a field questionnaire, it may be harder to secure a relatively homogenous sample (which is important from a theory testing perspective) (Calder, Phillips and Tybout 1981). By conducting a quasi-experiment, suitable respondents were committed in advance and a film selected which was known to contain product placements

57 Quasi-experimental designs, as opposed to true experiments, are appropriate for situations in which the researcher lacks the full control over the scheduling of the experimental stimuli which make a true experiment possible (Campbell and Stanley 1963; Cook and Campbell 1979). Quasi-experimental designs are therefore commonplace in behavioural sciences since many research designs involve extraneous variables beyond the control of the researcher. Indeed, most prior research looking into recall and recognition of product placements has conducted a quasi-experiment of sorts.  O              P     !  " # $% &  '    (*0,( relevant to the sample and which should provoke differing levels of engrossment among that sample. The film could also be accurately coded prior to implementation.

6.4 - The Sample 6.4.1 – Procurement of sample The subjects used in this stage were male and female teenagers aged 15-17 years inclusive (i.e. Years 10 and 11). As outlined in Section 5.4.1, they were sourced from two Sydney Catholic schools – one boys school and one girls school – after gaining the support of the Catholic Education Office (Sydney). A detailed letter describing the aims of the study (but not revealing its true purpose), how it would be conducted, what information would be sought and contact details of the researcher were provided to both the students and their parents / guardians (see Appendix 6.1). In order for a student to participate, both the student and the parent had to sign a consent form and return this to their teacher prior to participating in the research.

6.4.2 – Appropriateness of a teenage sample Although this sample means the results of this study represent only a narrow slice of the Australian population, this skew is a justifiable starting point, since teenagers are a primary audience for many film-makers and television producers. In 2005/06, 93% of Australians aged 15-17 and 85% of those aged 18-24 went to the movies, compared to 76% of 25–34 year olds and 69% of 35-44 year olds (Australian Bureau of Statistics 2007b) and 55% of people aged 50+ (Australian Film Commission 2008). Furthermore, the typical 14-24 year old went to the movies 8.9 times in 2006 (Australian Film Commission 2008). Yet despite their high cinema attendance, research on product placement targeted toward children and teenagers is sparse (Karrh 1998; Nelson and McLeod 2005). Therefore, there is a pressing need to better understand this audience, as their high cinema attendance makes them prime candidates for exposure to product placements.

Furthermore, adolescents are still at an age where they are developing an increased cognitive capacity to process information and are less mentally mature (Strasburger 1995).

 O              P     !  " # $% &  '    (*0-( They are yet to develop more sophisticated information processing skills or an ability to direct or control their own learning (Roedder 1981). Therefore, as with advertising, it is likely that adolescents are more vulnerable to product placement, especially since product placement strategies place products in movies with attractive characters who may explicitly or implicitly endorse a product (see Section 1.1).

Adolescence is the period in a person's life that is most often associated with identity development. Much of what they want to say to each other and to the rest of the world they communicate through the brands they wear and consume and the entertainment media choices they make (Galician 2004). Indeed, the way many young people think about and understand who they are is tied to consumption patterns. They buy products not only for their functions, but also for their symbolic value, which serves to display and reinforce one’s self concept and identity (Karrh 1998).

When developing norms, adolescents look to the greater social environment for concepts of adult identity, particularly the behaviour of leaders, heroes and film stars (Lynch and Bonnie 1994). Research by Lockwood and Kunda (1997) and Martin and Bush (2000) revealed that although parents remain the strongest influence on adolescent consumer purchase intentions and behaviours and self-views, vicarious “superstar” role models such as athletes and entertainers also wield significant influence. Moschis and Churchill (1978) also found a strong relationship between TV viewing and an individual’s social motivation for consumption and found that social utility reasons for watching television as a means of gathering information about lifestyles and behaviours are strong predictors of consumer levels of materialism and economic motivation for consumption. Similarly, Moschis (1978) found that adolescent consumers frequently use television celebrities to determine how products are to be utilised in every day life. Therefore, product placements can be seen to reinforce product use through models chosen by the individual to help them acquire brand preference and/or consumption behaviours that benefit the sponsor.

 O              P     !  " # $% &  '    (*0.( Whilst teenagers are sophisticated media users and are prone to tune out to standard advertising, they are at a time in their life where they are looking for role models from which to copy behaviours from and form identity characteristics. Therefore, they may be deemed more vulnerable than older audiences who do not seek to perform this same level of modelling. They hold less capacity to distinguish entertainment from commercial persuasion than do adults, so depicting products in a positive light via film is a persuasive method for advertisers to capture the attention of children and teenagers (Solomon and Englis 1994; Karrh 1998). Indeed, this cinema environment is ripe for both learning about brands as markers of individual characteristics and for applying this learning to one’s personal life.

6.5 - The Film Stimulus 6.5.1 – Understanding the film preferences of teenagers Film selection was a crucial element of this research, forming the stimulus from which audience engrossment would be captured and recognition of product placements recorded. To understand the film genre and actor preferences of the sample, during the focus groups that were conducted for scale item generation, the researcher also asked questions about their favourite films and stars (see Section 4.6.2).

Seventeen teenagers aged 12-17 were involved in these three focus groups obtained using convenience sampling (friends and acquaintances of the researcher), at which point this group was expanded using the snowball sampling technique (Burns and Bush 1998). This method was deemed relevant due to the difficulties sought in securing a sufficient sample size, the ease with which respondents were willing to provide the names of other potential respondents, and the temporal and economic constraints of the researcher. Furthermore, this technique was appropriate since age was the only screening criteria for this sample. It should be noted that this sample was distinct from the samples used in the latter part of the study in order to ensure that latter questionnaires and experiments were not contaminated.

 O              P     !  " # $% &  '    (*0/( A simple questionnaire (see Appendix 6.2) was designed in which respondents were asked to list their five favourite movies, their three favourite movie stars, and select their three favourite and three least favourite genres from a list. They were also asked to list their age, gender and postcode in order to confirm that a suitable cross-section of teenagers aged 12- 17 was sourced. Results were then discussed in the group, with the actual questionnaires returned to the researcher for further examination and comparison.

The data were entered into Excel and frequency analyses were run in order to identify which films, genres and stars were the most popular across both genders (see Table 6.5.1).

Table 6.5.1 - Key Findings from Exploratory Study Favourite Films Lord of the Rings (any), Star Wars (any), Anchorman, Zoolander, The Matrix (any) Preferred Genres Comedy, Action, Thriller, Sci-fi / Fantasy Preferred Actors Jessica Alba, Jackie Chan, Johnny Depp, Ben Stiller, Mike Myers, Will Ferrell

6.5.2 – Movie Selection and Analysis Based on the findings from this questionnaire, a number of films that contained a preferred star or were of a preferred genre were sought out for content analysis to identify all the product categories and brands contained within them58. The movie selection was a crucial element of this research as it needed to be a film which would engage the respondents, and have a range of relevant product placements with different executions.

In order for the optimal selection to occur, key tasks had to take place. These are outlined in greater detail in Appendix 6.3. Briefly however, these included:

58 Content analysis is an observational research method that is used to systematically evaluate the symbolic content of all forms of recorded communications (Kolbe and Burnett 1991). In this way, it can assess the effects (cognitive, affective and behavioural) of different kinds of message content on receiver responses and can provide an empirical starting point for generating new research evidence about the nature and effect of specific communications (Kolbe and Burnett 1991).  O              P     !  " # $% &  '    (*00( • Construction of a film coding form and instructions to accompany this form • Recruitment of four independent coders • Training of these coders (two sessions) • Completion of a film coding form for each viewing of a film

During April and May of 2007, 38 films (all on video or DVD – no new cinema release movies – for ease of coding and availability) were viewed by the researcher (see Appendix 6.4 for the film coding form and instructions). The selected films underwent preliminary coding by the researcher, until a shortlist of films fulfilling the criteria of the exploratory study and featuring a number of relevant brands was compiled. Appendix 6.5 provides a summary of all the brands featured in these movies.

Several movies were short-listed for further analysis by the other four coders, with each film analysed by two coders. These coders were asked to work individually and to use their codebook, strictly adhering to its definitions (and not their own definitions) of categories. It was deemed important to have multiple coders in order to note all brands contained in the film and to verify each others’ findings. If any significant discrepancies arose, these were resolved by reviewing the movie again, together with a third coder, until all parties agreed on the coding.

Coders were directed to code, for each scene, the presence of any brands. If these were present, the coders were to note the brand, the number of times it appeared in a scene, the length of time it appeared on-screen, as well as whether it was displayed in the background or forefront of a scene, or was used by a lead character or was in the same shot as the star. Finally, the product placement had to be characterised as being audio, visual or audio- visual in nature, and an assessment was made regarding the importance of the product/brand to the story (i.e. plot connection)59. For products that received multiple

59 To do this, coders had to decide how essential the product was to the scene. Could the scene still exist if that product was not featured?  O              P     !  " # $% &  '    (*01( placements within a film, the total exposure time was recorded. For each product placement, exposure time was measured from the time at which the product’s name, logo, advertising banner and/or distinctive shape was revealed to when the camera cut away to another shot, or the brand stopped being spoken of. Disagreements were resolved through discussion.

An important consideration when selecting the movie stimulus was movie familiarity. Preferably, respondents should not have seen the movie before as this controls for the influence of any unconscious memory or learning from the first exposure to the movie. However, given the prevalence and availability of movies for hire and download, it was thought that this would be difficult to achieve. An existing question at the top of the Audience Engrossment questionnaire would capture this information regarding whether the respondents had previously seen the movie. Differences in recognition between these two groups (those that had seen the movie before and those that had not) could then be assessed, and as long as there were no differences, the entire group could be analysed together. If there were significant differences, separate analyses would have to be done for those that had seen the movie before and those that had not.

It was decided that only one film would be used in this stage. This allowed it to be better demonstrated that when the movie is held constant, the engrossment of different viewers can vary and may in turn have an impact on one’s memory for the product placements carried within it.

Two movies were strong contenders to be this stimulus. Clueless, a 1995-released, teen- romantic-comedy appeared to be the most suitable movie stimulus. It contained a large number of brands ranging from everyday items to the aspirational, and these were clearly littered throughout the film. Furthermore, the “girly” appeal of the film may have proved to be a good manipulator of engrossment with the movie (i.e. the males in the sample may not like the film as much as the females, even though most of the products featured in the film were relevant to both genders). However, upon discussion with the schools, it was learned  O              P     !  " # $% &  '    (*02( that some students were studying this film as part of their English classes, thus creating an unusually high level of familiarity which could have an adverse effect on the results.

The selected film, The Island (rated M), is a sci-fi / action / thriller with popular actors (Scarlett Johansson, Ewan McGregor and Steve Buscemi) and a large number of very relevant brands and a broad range of product categories. When released in 2005, it had only a moderate box office performance and mixed reviews, so it was believed that not many of the respondents would have seen it. It was felt that this film might serve both genders better than Clueless would - the boys would like the car chases and fighting and the girls would like the underlying love story.

6.6 - Content analysis Using the content analysis form (see Appendix 6.4), a detailed content analysis was conducted of the selected film, The Island. The following 41 brands were identified as being present (see Appendix 6.6)60:

• Puma • Cadillac • Cosmopolitan • Speedo • Hummer • KPMG • Reebok • Land Rover • AmTrak • Aquafina • GMC • Johnny Rockets • Apple • Chrysler • Tag Heuer • Xbox • Chevrolet • NFL • msn • Nissan • Calvin Klein • Budweiser • Volvo • Ben and Jerry’s • Michelob • Pontiac • Cisco Systems • Bacardi • Mack • Nokia

60 N.B. Due to an oversight, Speedo was accidentally left off the brand familiarity and recognition tests. Furthermore, Sprint was not tested since this brand is not available in Australia and it is highly unlikely that respondents would have had any exposure to it, even on their travels. Therefore, 39 brands were available for analysis.  O              P     !  " # $% &  '    (*13( • Jack Daniels • Advanced Armor • Samsung • BMW • Popular Mechanics • Sprint • Lexus • Maxim • NBC • Bentley • Esquire

6.7 – Operationalisation of Constructs and Questionnaire Development As discussed in Section 3.4, recognition of the product placements in the film The Island served as the dependent variable in this study. Brandcast quality, specifically, the number of exposures, total seconds on screen, prominence, star presence, star use, modality (audio/visual or dual/single mode), plot connection, and method of depiction (actual product or logo / ad / signage) were the independent variables, and brand familiarity and audience engrossment were the moderators. As stated in Section 3.7, connectedness was not studied in this research as it is not applicable to film. Brandcast quality was captured via the content analysis, however all other variables needed to be captured via direct questioning of respondents. Operationalisation of these is now discussed.

6.7.1 – Brand familiarity Terms such as “familiarity”, “experience”, “expertise”, and “product (category) knowledge” are interchangeably used to describe how knowledgeable consumers are about a particular product (Rao and Monroe 1988). Moreover, Alba and Hutchinson (1987, p411) define familiarity as “the number of product-related experiences that have been accumulated by a consumer”. Therefore, respondents were asked to state whether they had “never heard of”, “had heard of”, “would recognise the brand or logo”, or “had purchased or used the brand”.

Familiarity information was gathered for all brands identified in The Island, arranged by product category (see Appendix 6.7). For each product category, at least two additional brands were included that were not featured in the movie. Other product categories (and their brands) not featured in The Island were also included (e.g. soft drink). This enabled

 O              P     !  " # $% &  '    (*1*( the researcher to gauge the usage of other products and brands that may be used by the respondents and to test whether the respondents believed that these brands also appeared in the film after they had seen it (and therefore give the product a false-positive rating).

6.7.2 – Audience Engrossment As outlined in Chapters 3, 4 and 5, the audience engrossment construct seeks to capture feelings evoked by a movie, physical reactions to the movie (arousal), appraisal of the movie, and cognitive effort expended in watching it. Respondents were presented with the 62-item version of the scale (which became Stage 4 / confirmatory stage of the scale refinement process) (see Section 5.4) after watching The Island (see Appendix 6.8).

6.7.3 – Recognition After completing the audience engrossment questions, respondents were provided with a list of brands (listed per product category) and were asked to select those which they remembered seeing or hearing reference to in the movie (from a list of alternatives) (see Appendix 6.8)61. Measuring recognition in this way is a form of memory testing that simply requires one to differentiate or discriminate the previously encountered stimulus from a set of distracting stimuli (Bettman 1979).

As with the familiarity questionnaire, respondents were required to give an answer for each question. This could either be “definitely”, “no”, or “maybe” in order to gain an understanding of the respondent’s confidence of each answer. For this reason, the “maybe” category was deemed important as it allowed the respondent to express their own confidence in their answer, and not force them to guess that the brand did or did not appear in the film.

61 The product categories and brands in this questionnaire were the same as those appearing on the earlier familiarity questionnaire in order to address the familiarity–recognition portion of the proposed model. Furthermore, by containing a variety of brands, including those not featured in the film, this design could also examine which brands people selected if they did not recognise the correct brand. However for this thesis, this extension was not undertaken.  O              P     !  " # $% &  '    (*1+( 6.8 - Research management 6.8.1 – Pre-film testing The Brand Familiarity Questionnaire (see Appendix 6.7) was administered by the researcher (with the teacher present) during class in both School A and School B in May 2007. Administering the questionnaires in class was the preferred method as it was believed that this would maximise response rates. Furthermore, the researcher could be assured that it was the teenager themselves completing the questionnaire, that it was conducted without discussion with other students, and that it was not lost or not returned.

Although lengthening the research process, the Brand Familiarity Questionnaire was administered several weeks before screening of the film to generate a more accurate account of actual brand usage. If this questionnaire was administered after watching the film, there is the chance that respondents would modify their responses to be more like the actors in the movie, particularly if the movie featured one of their favourite stars. Furthermore, the research design already had the respondents answering one very large questionnaire with three parts62 following the screening of the film, and by including the Brand Familiarity Questionnaire at this stage as well, there may be dangers of respondent boredom or fatigue, thus reducing the potential reliability and validity of these results. Alternatively, whilst splitting the sample into two groups, whereby one group answers the Brand Familiarity Questionnaire before seeing the film, and one group answers the questionnaire after seeing the film may have allowed for differences between the groups to be either identified or dismissed, to do this properly, it would have to have been done within each school sample. This would have made it very difficult to administer, especially in the post-film testing (that is, identifying who had already answered the Brand Familiarity Questionnaire and who had not), but more problematic is the fact that it is difficult to prevent the teenagers from talking amongst themselves as to what the Brand Familiarity Questionnaire was about, and this in turn could have contaminated the results of the post- film testing.

62 Audience Engrossment scale, recognition task, Program Liking scale  O              P     !  " # $% &  '    (*1,( 6.8.2 – Film Screening Four weeks after answering the Brand Familiarity Questionnaire, respondents were shown, in a single session, the full-length version of the movie The Island. The study was run in a single session at each school in order to minimise communication effects. The movie was shown in both locations in a setting that was as close to a real cinema environment as possible, namely a large hall with a cinema-style projection screen and excellent viewing facilities. Although this is not a truly realistic film-viewing experience, having respondents view an entire film in a controlled setting enhances external and internal validity of the results. In particular, external validity is enhanced because viewing an entire film with several placements simulates the real-world clutter currently found in films (Babin and Carder 1996).

Respondents were not told the purpose of the study before watching the film. Fortunately, both schools were able to create a scenario where the students would not suspect anything. School A was shown the movie as part of their regular English / Film classes, and School B was shown the movie during the final week of the school term as a “treat”. This meant that the respondents watched the movie as they normally would – they would not be paying special attention to it knowing that they would have to answer questions afterwards. Since it did not have the formality of a true experiment, it therefore had more of a “real world” feel, and stronger external validity.

6.8.3 – Post-Film Testing After viewing the movie, subjects were administered with a “surprise” recognition test concerning the embedded product placements, as well as the Audience Engrossment scale and the Program Liking scale (Murry et al. 1992)63. Respondents were given 25 minutes to complete the questionnaire (see Appendix 6.8). Respondents were then debriefed as to the purpose of the study, and a brief outline of what product placement is was given. Finally,

63 The Program Liking scale was included to assess convergent validity of the AUDENG scale and had nothing to do with testing the model. These results were discussed in Section 5.7.  O              P     !  " # $% &  '    (*1-( all respondents and teachers were thanked for their time. The entire process took just over two and a half hours.

6.9 – Data Preparation and Method of Analysis With data collected over two separate stages, each respondent was asked to invent a unique code name that they would be known by for the duration of the research64, and it was with this that the questionnaires were manually sorted and matched. This matching resulted in the identification of 191 respondents who had completed all four tasks across the two stages65. The data were entered into SPSS and analysis was conducted in two stages – descriptive analysis and model testing.

6.9.1 – Descriptive Analysis SPSS 15 was used for data preparation and basic descriptive statistics, to gain an understanding of which brands had the highest recognition, were the most familiar, and to look for differences in these results between the genders. For the purposes of testing the model, the four familiarity response options were collapsed into one aggregate “familiarity” measure (“never heard of” versus “had heard of” / “would recognise” / “had used or purchased”). Similarly, the three recognition responses were collapsed so that the “no” and the “maybe” categories were collapsed into the one “no” category66. Since a number of ‘dummy’ brands were used in the questionnaires, a new variable “true recognition” was calculated for each respondent for each brand, with a “1” denoting that the brand was recognised and was in the movie. This new variable served as the dependent variable for the model testing.

64 The code name was derived from the respondent’s initials, birthday and last 4 digits of their phone number (e.g. 3985 - 2104 - JS) 65 The sample used for this analysis is lower than the 273 discussed in the scale testing stage (Section 5.4) due to this matching procedure. A respondent could only be included if all four tasks had been completed – brand familiarity, audience engrossment, recognition testing and program liking. 66 Familiarity and Recognition were dichotomised for this preliminary investigation for ease of analysis. New scoring for familiarity was: have heard of, would recognise, have purchased or used = 1; have not heard of = 0. New scoring for recognition was: definitely = 1; maybe or no = 0. Further insight may be gained in future research by exploring the full range of familiarity and recognition.  O              P     !  " # $% &  '    (*1.( Initial exploration of the content analysis coding sheet (see Appendix 6.6) and the data (via correlation analysis) was conducted to assess whether there was any multicollinearity between product placement characteristics which may cause problems in any future regression analyses. Significant correlations were found in the data, but generally, these made sense (see Appendix 6.9). For example, a placement with low plot connection was also likely to be a creative placement (i.e. a brand with no relation to the plot is often found in the background of a shot). These same placements, creative and low plot connection, also tended to be visual. There were also high correlations between star use and star presence (which make sense since if the star uses the product, they also need to be present in the scene), and between frequency of exposure and total seconds on screen (obviously the more often a brand appears, it has more opportunity to increase its total screen time). Some of these effects could be minimised through recoding (discussed next), but others could not.

The two variables relating to onset (high prominence) and creative (low prominence) placements were combined into one prominence variable, whereby a “1” denoted high prominence and a “0” denoted low prominence. Similarly, high and low plot connection were combined to form a plot connection variable, with “1” denoting high plot connection and “0” low plot connection. Furthermore, if a brand was ever to have high prominence or high plot connection, it was scored “1” for the relevant new variable. Previously, for example, if in some scenes it had been creative and in some scenes it had been on-set, it was scored “1” for each.

The original coding “double counted” for star use, awarding a “1” for star presence and a “1” for star use. The data were recoded so that if the brand was used or mentioned by a star, it scored a “1” for star use only. A “1” for star presence came about only if the brand was in the same shot as the actor, but not actually used or mentioned by them. Obviously if neither of these events took place, the brand scored a “0” for both.

 O              P     !  " # $% &  '    (*1/( A dummy variable was created to reflect dual modality versus single modality. If a brand had scored “1” for a visual placement and “1” for an aural placement, it now also scored a “1” for the new modality variable. A “0” on the new modality variable implied the placement appeared either visually or aurally only (i.e. not both). This variable was created for simplicity of analysis; future research might wish to further investigate which of the single modes had more of an impact.

No changes were made to the temporal data (i.e. frequency of exposure and total seconds on screen) as we were particularly interested to see if one variable was deemed to have more impact than the other. Creating a combined variable of ‘temporal quality’ that combined these two variables would not only be difficult, but would make it impossible to identify which aspect was more important. However, both variables could not be used in any subsequent modelling due to very high correlations (see Section 6.11).

Cross tabulations were used to investigate the incidence of false-positive responses. As can be seen in Table 6.9.1, there was a very low level (2.4%) of false positives (i.e. respondents saying the brand was definitely in the movie when it was not)67. Overall, 56.7%68 of brands across all people were recognised correctly (i.e. seeing the brand when it was there, and not seeing it when it was not there). More specifically to this research, 23.2% of brands across all people were correctly identified as definitely being in the movie when they were69. These findings give us increased confidence in the data and confirm that respondents were taking the task seriously and not guessing.

67 = 481 / 19662 68 = (1704 + 9448) / 19662 69 = 1704 / 7350  O              P     !  " # $% &  '    (*10( Table 6.9.1 – Accuracy of responses Saw brand in the movie Definitely Maybe No Total Brand was in No 481 2383 9448 12312 the movie Yes 1704 1522 4124 7350 Total 2185 3905 13572 19662

Independent t-tests were conducted to investigate whether there were differences in recognition between the genders or between those who had seen the movie previously. These results are discussed in Section 6.10.

6.9.2 – Model Testing The modelling of recognition was conducted on the 39 brands in the movie (not all the 104 brands that featured in the questionnaires). This is justified given the low level of false- positives, plus the variation in recognition rates of the brands in the movie. Furthermore, these were the only brands that offered any variance in the product placement characteristics.

The modelling was performed at the individual level due to the multiple measures per each respondent (i.e. each respondent had familiarity and recognition data for each of the 39 brands), with the data set akin to panel data. These multiple measures also allowed the modelling to take into account the heterogeneity across respondents. A random effects binary choice logit model was fitted to the data using LIMDEP 9.070.

Direct effects of product placement on recognition were examined through the significance (or lack thereof) of the independent variables. Moderator effects were explored by examining the significance of the interaction terms between each of the audience engrossment dimensions and the product placement characteristics and the interactions between brand familiarity and the product placement characteristics. Significance was

70 It is acknowledged there are a number of ways to account for heterogeneity (see Ailawadi, Gedenk and Neslin 1999), however random effects is a suitable measure for a preliminary investigation of the model.  O              P     !  " # $% &  '    (*11( tested at the 5% level, and model fit was assessed in terms of log likelihoods (LL), Bayesian Information Criteria (BIC) and predictive classification (i.e. the percentage of cases correctly classified by the model), whereby lower LLs and BICs and higher predictive success denoted better fit.

6.10 – Results: Descriptive statistics Table 6.10.1 shows the 39 brands featured The Island, ranked in order of recognition. Familiarity levels of these brands are also shown.

Table 6.10.1 – Recognition and Familiarity of brands in the movie Familiarity Recognition (Have heard of, would (Was brand in the movie?) recognise, have used or purchased brand) Brand Definitely Maybe/No Was familiar with msn 93.2% 6.8% 100.0% Puma 77.0% 23.0% 99.5% Calvin Klein 69.9% 30.1% 98.4% Nokia 63.5% 36.5% 100.0% NBC 63.0% 37.0% 82.0% Microsoft Xbox 61.3% 38.7% 99.0% Mack 52.1% 47.8% 43.1% Maxim 37.6% 62.3% 31.6% NFL 32.4% 67.5% 92.6% Hummer 29.4% 70.6% 86.4% Bentley 25.5% 74.4% 90.5% Chrysler 24.5% 75.6% 79.4% Apple 22.3% 77.6% 100.0% Ben & Jerry's 21.8% 78.2% 20.9% Advanced Armor 21.0% 79.0% 16.5% Popular Mechanics 20.5% 79.5% 7.5% BMW 18.9% 81.0% 100.0% Esquire 18.8% 81.2% 14.4% Cadillac 18.0% 82.0% 88.4% Chevrolet 17.6% 82.5% 72.1%

 O              P     !  " # $% &  '    (*12( Aquafina 16.8% 83.2% 40.4% Land Rover 13.2% 86.8% 99.0% Budweiser 11.2% 88.8% 57.8% Lexus 11.2% 88.8% 97.4% Amtrak 11.1% 88.9% 11.7% GMC 8.0% 92.0% 37.2% Volvo 7.4% 92.7% 97.9% Tag Heuer 6.9% 93.1% 30.5% Jack Daniels 5.3% 94.7% 91.1% Nissan 5.3% 94.8% 100.0% Reebok 4.2% 95.8% 96.8% Samsung 3.2% 96.8% 100.0% Johnny Rockets 2.6% 97.4% 7.4% Pontiac 2.6% 97.4% 51.3% Cisco Systems 2.1% 97.9% 8.4% Cosmopolitan 2.1% 97.9% 90.0% KPMG 2.1% 97.9% 4.7% Bacardi 1.6% 98.4% 93.2% Michelob 1.1% 98.9% 16.6%

These results suggest that the targeting of brands featured in the movie to match the audience was fairly successful. Two thirds of brands were familiar to at least 50% of the audience, with only four out of 39 having familiarity levels below 10%. Furthermore, a general pattern can already be seen that those brands with higher recognition levels also tend to have higher familiarity levels. This superficial analysis begins to offer support for Hypothesis 3, as does the fact that a correlation analysis showed a significant positive relationship between recognition and brand familiarity (r=0.33, p=0.04).

Significant differences (t=3.88, p=0.00) were found between the genders and their recognition levels, with males (mean = 10.7) recognising more brands than females (mean = 8.0). Furthermore, significant differences were also found between gender and familiarity (t=2.34, p=0.02), with males being familiar with more brands than females (26.5 vs 24.8). This, plus the previously stated correlation between recognition and familiarity indicated that it may be the difference in familiarity, not gender, driving the significant differences in  O              P     !  " # $% &  '    (*23( recognition, and that familiarity may actually be a surrogate for gender. More detailed investigation on a brand-by-brand basis revealed that significant differences in familiarity were found between the genders for 13 of the 39 brands, and of these, nine also experienced significant differences in recognition. Many of these differences were logical (for example, the boys were more likely to be familiar with, and thus recognise, the 14 car brands). Therefore, gender was not considered to be an issue, as it appeared to be captured by brand familiarity, and the sample was not split according to gender.

However, there were significant differences in recognition between those that had seen the movie previously and those who had not (t=-2.38, p=0.02), with those who had seen the movie before having higher recognition rates (mean = 10.0) than those who had not (mean = 8.4). This meant that in the second stage of analysis (the modelling), these two groups (seen before and not seen before) would need to be analysed separately, as these differences could not be attributed to differences in brand familiarity (as gender could)71.

6.11 – Results: Direct effects Direct effects of each individual product placement characteristic on recognition was initially explored through a series of simple models (i.e. each model had only one product placement characteristic). These were run separately for the two groups (those who had seen the movie before and those who had not) (see Section 6.10). These results are summarised in Table 6.11.1.

71 There were no significant differences between those who had seen the movie before and those who had not for brand familiarity (p=0.17).  O              P     !  " # $% &  '    (*2*( Table 6.11.1 – Direct effects of each product placement characteristic on Recognition

Not seen before Seen before Variable b b/SE p LL BIC b b/SE p LL BIC Model coeff value coeff value 1 Plot 1.39 18.77 0.000 -2311.8 0.95 1.40 14.48 0.000 -1318.2 1.03 connection 2 Prominence 1.49 17.14 0.000 -2299.3 0.95 1.87 15.16 0.000 -1261.4 0.99 3 Star use 1.02 10.13 0.000 -2392.6 0.99 1.12 9.25 0.000 -1357.6 1.06 4 Modality -1.34 -7.92 0.000 -2432.7 1.00 -1.50 -6.07 0.000 -1384.6 1.08 5 Total seconds 0.01 4.29 0.000 -2469.7 1.02 0.01 2.58 0.010 -1417.6 1.11 6 Freq of 0.02 3.79 0.000 -2472.2 1.02 0.01 2.53 0.012 -1417.8 1.11 exposure 7 Actual brand 0.49 3.54 0.000 -2470.9 1.02 0.40 2.16 0.031 -1416.1 1.11 shown 8 Logo / ad / 0.28 2.89 0.004 -2476.2 1.02 0.17 1.24 0.214 -1419.7 1.11 signage 9 Star presence -0.19 -1.29 0.198 -2481.7 1.02 -0.49 -2.08 0.038 -1416.7 1.11 Predictive Success = 76.84% Predictive Success = 73.82%

When variables were examined individually, overall results for the two groups were fairly similar, with plot connection, prominence, star use and modality having the greatest significant effect on product placement recognition in each set of models (as reflected by their higher b/SEs and lower BICs and log likelihoods). For those that had not seen the movie before, the star’s presence alongside the brand was not found to significantly impact recognition (although it did have a significant negative effect on those that had seen the movie before). In contrast, depiction of a logo, ad or signage had no significant impact on recognition for those that had seen the movie before, but a significant positive impact for those that had not. The models for those that were seeing the movie for the first time had a marginally higher predictive success compared with those that had seen it before (76.84% versus 73.82%).

Due to the collinearity between frequency of exposure and total seconds on screen (r=0.96, p=0.00), only one variable could be used in the modelling of recognition. Total seconds on screen was chosen as the variable to report on as it had a greater range of values which

 O              P     !  " # $% &  '    (*2+( would provide more variance in the data, and is most likely driven by frequency of exposure (as opposed to total time on screen impacting frequency of exposure)72.

When all placement variables were entered into the one model to capture the effect of the placement as a whole, the models that emerged for the two groups (“seen before” and “not seen before”) had a better fit than previously (as reflected by their predictive success and BICs) (see Table 6.11.2). The relative importance of each variable remained more or less constant in both models, as reflected by their relative ordering. The major difference was that total seconds on screen and star presence were not seen to contribute to recognition if the respondent had already seen the movie (although they were significant at the 10% level).

Table 6.11.2 – Direct effects of all product placement characteristics on Recognition Not seen before Seen before Interaction b coeff b/SE p value b coeff b/SE p value Prominence 1.45 12.62 0.000 2.08 13.55 0.000 Plot connection 1.43 12.41 0.000 1.49 10.10 0.000 Modality -2.41 -11.94 0.000 -2.66 -7.36 0.000 Logo / ad / signage 1.26 8.08 0.000 0.99 4.69 0.000 Star use 0.84 5.92 0.000 0.63 3.69 0.000 Actual brand shown 1.12 5.29 0.000 0.81 2.81 0.005 Star presence 1.15 6.53 0.000 0.53 1.91 0.057 Total seconds 0.01 3.48 0.001 0.01 1.75 0.080 Log likelihood = -1924.3 Log likelihood = -1039.4 Predictive success = 80.41% Predictive success = 81.00% BIC =0.81 BIC = 0.84 Random effects heterogeneity was significant for both conditions (p=0.00).

It was hypothesised that each of these nine independent variables (product placement characteristics) would have a positive impact on recognition (see Hypotheses 1a to 1h). For the most part, all hypotheses were satisfied, especially when respondents had not seen the movie before (see Tables 6.11.1 and 6.11.2). Unexpectedly however, modality (Hypothesis

72 N.B. Analysis was done using each of these two variables in turn. Results were very similar in terms of significance levels and relative ordering of items.  O              P     !  " # $% &  '    (*2,( 73 1h) had a negative impact on recognition . Examination of the brands that were placed both aurally and visually74 (Cadillac, Jack Daniels, Cisco Systems, Budweiser and Reebok) revealed that they had very low rates of recognition. Whilst past research suggests that dual modality should aid recognition, perhaps these results suggest that on its own, this characteristic is not enough to lead to enhanced recognition.

It is also likely that product placement quality is not a sum of its characteristics as modelled here, but rather an interaction or combination of these characteristics. This highlights the need for an overall product placement quality measure which may see weightings given to different product placement characteristics, and one product placement score for each brand calculated (see Section 7.5.3). That said however, it is expected that a highly prominent placement, the brand being highly essential (connected) to the scene, and / or the brand being used by the star would be very important factors and given a high weighting. This is reflected by their high (and positive) b coefficients and strongly significant p values75.

Therefore, the hypothesis (H2) that high quality placements (i.e. those placements with these characteristics) will be better recognised than low quality placements appears generally supported.

6.12 – Results: Indirect effects Even though most product placement quality characteristics were found to have a direct significant effect on recognition (p<0.05), when the audience characteristics of brand familiarity and audience engrossment76 were added in as moderators (via interaction terms),

73 It should also be noted that when the impact of visual placements and the impact of aural placements on recognition were assessed separately, their co-efficients were also negative. 74 N.B. These placements were not necessarily both audio and visual simultaneously. They may have appeared in one scene visually, but then been spoken of in another scene. 75 Whilst logo / ad / signage also had a strong direct impact when examined together with the other characteristics, its individual impact when looked at alone was low for those seeing the movie for the first time (and was not significant for those who had seen the movie before). Furthermore, we shall also see later that it had a weak impact when moderated with the audience engrossment dimensions (see Section 6.12). 76 Note that the four dimensions of audience engrossment were examined individually as it was hypothesised that they would have different impacts on recognition.  O              P     !  " # $% &  '    (*2-( many were no longer significant. However, fit of the models improved in many cases, as reflected by lower log likelihoods and higher predictive success.

6.12.1 – Brand familiarity as a moderator The interaction of familiarity with each of prominence, modality, plot connection, use by star, total seconds on screen and the depiction of the actual brand had a significant effect on recognition, thus providing evidence that brand familiarity does moderate the product placement characteristic – recognition relationship. Not surprisingly, results were very similar for both groups (those who had seen the movie before and those who had not) in terms of fit, sign and size of impact. The interaction of prominence and brand familiarity had the strongest impact, followed by modality, plot connection and use by star. These interactions generally had a positive impact on recognition. For example, prominence has a greater impact on recognition if the audience is already familiar with the brand than if they are not familiar.

Table 6.12.1 – Moderating effects of Brand Familiarity on Recognition Not seen before Seen before Interaction b coeff b/SE p value b coeff b/SE p value Familiarity x prominence 1.67 13.79 0.000 2.21 13.91 0.000 Familiarity x modality -2.39 -9.78 0.000 -2.65 -8.11 0.000 Familiarity x plot connection 0.99 6.96 0.000 1.03 6.97 0.000 Familiarity x use by star 0.80 5.63 0.000 0.62 4.23 0.000 Familiarity x seconds on screen 0.01 3.61 0.000 0.01 2.64 0.008 Familiarity x actual brand shown -0.50 -3.59 0.000 -0.52 -2.76 0.006 Log likelihood = -1923.1 Log likelihood = -1036.9 Predictive success = 81.97% Predictive success = 81.90% BIC =0.80 BIC = 0.83 Random effects heterogeneity was significant for both conditions (p=0.00).

Contrary to expectations, the interaction of brand familiarity and depiction of the actual brand was found to have a negative impact on recognition. Since no simple explanation could be found for this, further investigations were made. If the interaction of depiction of the actual brand and familiarity was the only variable in the model, its impact was positive – as expected. However, when any other brand familiarity x product placement characteristic variables were present in the model, its effect became negative. This suggests  O              P     !  " # $% &  '    (*2.( a possible interplay between depiction of the actual brand and other product placement characteristics, and the need to consider groups of product placement characteristics together rather than as individual items (see Section 7.5.3.3).

The negative impact of the brand familiarity x modality variable was explored in a similar fashion. However, its impact was consistently negative, which is not surprising given the negative direct impact of modality found in Tables 6.11.1 and 6.11.2 (but is surprising given Hypothesis 1h and prior research).

Despite these anomalies, there is partial support for Hypothesis 3 – that when brand familiarity is high, each product placement characteristic has a more positive impact on recognition that when brand familiarity is low.

6.12.2 – Feelings as a moderator The interaction of feelings with the product placement characteristics resulted in a number of these interactions having a significant impact on recognition, however there were differences between those who had seen the movie before and those who had not in terms of the relative ordering of the variables, their significance and the strength of their impact (see Table 6.12.2). The interaction of feelings and plot connection had a strong impact for those who had not seen the film before, and this was closely followed by prominence (b/SE of -13.07 vs -12.16). In contrast, for those that had seen the film before, prominence had a significantly stronger impact compared with that of plot connection (b/SE of -15.43 vs -9.41).

 O              P     !  " # $% &  '    (*2/( Table 6.12.2 – Moderating effects of Feelings on Recognition Not seen before Seen before Interaction b coeff b/SE p value b coeff b/SE p value Feelings x plot connection -0.51 -13.07 0.000 -0.56 -9.41 0.000 Feelings x prominence -0.47 -12.16 0.000 -0.75 -15.43 0.000 Feelings x modality 0.73 8.90 0.000 0.92 6.64 0.000 Feelings x use by star -0.29 -5.58 0.000 -0.18 -3.16 0.002 Feelings x star presence -0.31 -4.14 0.000 Not sig Feelings x seconds on screen -0.0030 -3.76 0.000 -0.0032 -2.50 0.013 Feelings x logo/ad/signage -0.15 -3.26 0.001 -0.14 -2.29 0.022 Log likelihood = -2151.7 Log likelihood = -1135.5 Predictive success = 76.29% Predictive success = 76.07% BIC = 0.90 BIC = 0.91 Random effects heterogeneity was significant for both conditions (p=0.00).

With the exception of modality (which consistently acts as an anomaly throughout the analysis), all interactions had the hypothesised negative impact on recognition. This means that when feelings were high (intense), each of these product placement characteristics had a stronger negative impact on recognition than when feelings were low (less intense). These results therefore generally support Hypothesis 4.

6.12.3 – Arousal as a moderator The interaction of arousal with prominence, plot connection, modality, depiction of the actual brand, total seconds on screen and use by star all had a significant impact on recognition (see Table 6.12.3). As with brand familiarity, prominence had the strongest impact, with interaction of plot connection next. If the respondent had seen the movie before, the impact of prominence was much greater than that of plot connection (b/SE of - 13.42 vs -8.54), although there were no other real differences between the two groups.

 O              P     !  " # $% &  '    (*20( Table 6.12.3 – Moderating effects of Arousal on Recognition Not seen before Seen before Interaction b coeff b/SE p value b coeff b/SE p value Arousal x prominence -0.57 -9.53 0.000 -0.75 -13.42 0.000 Arousal x plot connection -0.49 -8.04 0.000 -0.48 -8.54 0.000 Arousal x modality 0.59 5.33 0.000 0.85 7.55 0.000 Arousal x actual brand shown 0.25 2.89 0.004 0.15 2.16 0.031 Arousal x seconds on screen -0.003 -2.45 0.014 -0.0030 -2.34 0.019 Arousal x use by star -0.16 -2.35 0.019 -0.14 -2.55 0.011 Log likelihood = -2282.3 Log likelihood = -1165.2 Predictive success = 77.11% Predictive success = 76.65% BIC = 0.95 BIC = 0.93 Random effects heterogeneity was significant for both conditions (p=0.00).

With the exception of the interactions of modality and depiction of the actual brand with arousal, the other interactions had the hypothesised negative effect on recognition, thus generally supporting Hypothesis 5. The contrary impact of depiction of the actual brand was investigated in the same way as it was for familiarity (see Section 6.12.1), and once again, results indicated that its unexpected sign was likely due to interactions with other product placement characteristics.

6.12.4 – Appraisal as a moderator Of all the interactions between an audience engrossment dimension and the various product placement characteristics, appraisal was found to have the greatest differences depending on prior exposure to the movie (see Table 6.12.4). The model for those seeing the movie for the first time had marginally better predictive success than the model for those who had seen the movie before (76.96% versus 73.82%). For those that had not seen the movie before, prominence, plot connection, depiction of the actual brand and modality were found to have strong impacts on recognition. As with all previous models, depiction of the actual brand and modality performed contrary to expectations. Yet once again, when a model looked only at the interaction of appraisal and actual brand on recognition, this variable performed as expected, thus supporting the idea that its interaction amongst the other product placement variables alters its behaviour. Therefore, for those who have not seen the movie before, there is general support for Hypothesis 6 - that when the appraisal is high,

 O              P     !  " # $% &  '    (*21( each product placement characteristic will have a stronger impact on recognition than when the appraisal is low.

However, appraisal has a minimal moderation effect for those who have seen the movie before. Furthermore, the one variable that is significant (appraisal x use by star) has an unexpected negative impact (relative to the hypothesised impact and the direct and indirect effects of use by star in the aforementioned analysis). No explanation could be found for this anomaly – it may be a peculiarity of the current data – particularly since use by star on its own has a positive impact on recognition. Given this anomaly, a more accurate interpretation in this case may be to assume that appraisal does not moderate in this situation at all (although this should be examined in future research to confirm whether this is in fact the case).

Table 6.12.4 – Moderating effects of Appraisal on Recognition Not seen before Seen before Interaction b coeff b/SE p value b coeff b/SE p value Appraisal x prominence 0.26 10.02 0.000 Not sig Appraisal x plot connection 0.22 5.76 0.000 Not sig Appraisal x actual brand shown -0.14 -3.36 0.001 Not sig Appraisal x mode (audiovisual) -0.24 -3.21 0.001 Not sig Appraisal x use by star Not sig -0.07 -2.27 0.024 Log likelihood = -2428.5 Log likelihood = -1419.93 Predictive success = 76.96% Predictive success = 73.82% BIC = 1.01 BIC = 1.11 Random effects heterogeneity was significant for both conditions (p=0.00).

6.12.5 – Cognitive effort as a moderator The moderating role of cognitive effort on recognition produced a simpler model with fewer significant interactions for those seeing the movie for the first time, than the model for those who had seen the movie before (see Table 6.12.5). For both scenarios, prominence, modality and plot connection again had the greatest impact, however all interactions had a much greater impact for those who had seen the movie before, especially prominence and plot connection (b/SE = -15.55 and -9.05 for those who had seen the movie before versus -6.81 and -2.89 for those who had not seen the movie before). For those that

 O              P     !  " # $% &  '    (*22( had seen the movie, depiction of the actual brand, or its logo / ad / signage were also found to have significant positive impacts on recognition (both in the opposite direction of what was expected). However, when looked at in isolation, depiction of the actual brand was found to have a negative impact on recognition (as expected), whereas depiction of a logo / ad / signage did not have any significant effect on recognition. Again, these results highlight that for certain product placement variables, the interaction of them in the model with other product placement variables significantly impacts their behaviours. Modality continued to have a positive impact on recognition, when looked at in both isolation or as part of the final model. In general however, partial support was found for Hypothesis 7, with each product placement characteristic reducing the probability of recognition to a greater extent when cognitive effort was high as opposed to when cognitive effort was low.

Table 6.12.5 – Moderating effects of Cognitive Effort on Recognition Not seen before Seen before Interaction b coeff b/SE p value b coeff b/SE p value Cognitive effort x prominence -0.26 -6.81 0.000 -0.77 -15.55 0.000 Cognitive effort x modality 0.30 3.92 0.000 0.85 5.46 0.000 Cognitive effort x plot connection -0.12 -2.89 0.004 -0.43 -9.05 0.000 Cognitive effort x seconds on screen -0.003 -2.39 0.017 -0.004 -3.11 0.002 Cognitive effort x actual brand shown Not sig 0.36 4.22 0.000 Cognitive effort x logo/ad/signage Not sig 0.23 2.48 0.013 Log likelihood = -2440.8 Log likelihood = -1258.4 Predictive success = 77.19% Predictive success = 75.37% BIC = 1.01 BIC = 1.00 Random effects heterogeneity was significant for both conditions (p=0.00).

 O              P     !  " # $% &  '    (+33( 6.13 – Summary of Hypothesis Testing Support Hypothesis Not seen Had seen before before

H1: All product placement characteristics will have a positive Partial* Partial* impact on recognition of the brands featured in the film. H1a: Temporal quality (i.e. length of time on screen) will have a  Partial positive impact on recognition of a brand featured in the film. H1b: Recognition will be higher if a brand’s placement has high   prominence than if a brand’s placement has low prominence. H1c: Recognition will be higher if a brand is used by a star than if   a brand is not used by a star. H1d: Recognition will be higher if a star is present alongside a Partial Partial brand than if a star is not present alongside a brand. H1e: Recognition will be higher if the brand has high plot   connection than if a brand has low plot connection. H1f: Recognition will be higher if the actual brand is depicted than   if the actual brand is not depicted. H1g: Recognition will be higher if the brand’s logo / advertising /  Partial signage is depicted in the film than if the brand’s logo / advertising / signage is not depicted. H1h: Recognition will be higher if a brand’s placement is by dual   modality than if the brand’s placement is by single modality.

H2: Brandcasts with high brandcast quality will be better Partial* Partial* remembered than those with low brandcast quality. H3: When brand familiarity is high, each product placement characteristic will have a more positive impact on recognition Partial* Partial* than when brand familiarity is low. H4: When feelings are high (i.e. more intense), each product placement characteristic will have a more negative impact on Partial* Partial* recognition than when feelings are low (i.e. less intense). H5: When arousal is high, each product placement characteristic will have a more negative impact on recognition than when Partial* Partial* arousal is low. H6: When appraisal is high, each product placement characteristic will have a more positive impact on recognition than when Partial appraisal is low.  H7: When cognitive effort is high, each product placement characteristic will have a more negative impact on recognition Partial Partial* than when cognitive effort is low.

* These cases have good support except for the anomalous findings relating to modality and depiction of the actual brand.

 O              P     !  " # $% &  '    (+3*( 6.14 – Revised Model of Brandcast Processing The results presented in these previous sections confirm the direct relationship between the product placement (brandcast) quality characteristics and their impact on memory (in this case via recognition). They also confirm that the components of audience engrossment and brand familiarity moderate this relationship (i.e. these factors impact product placement processing)77. As discussed in Section 3.7, connectedness was not tested in this study since the concept is not suited to film as it relates to pre-existing and ongoing attitudes to a program’s content. However, it is a highly relevant concept when using television programs as the research context.

Star liking has now been added to the basic brandcast processing model as a fourth possible moderator (see Figure 6.14), and is relevant to all research contexts. Originally believed to be captured by audience engrossment (see Section 4.6.3), it was found not to fit with the appraisal dimension (see Section 5.2.5). Re-examination of the items led to the realisation that star liking is actually a pre-existing attitude and not generated by the film (or any other entertainment vehicle) (although it may change as a result of seeing the star in that entertainment program). Basic star liking information continued to be captured via a single item in this research so that basic analysis could be conducted (see Section 7.2.4), but in reality, a single item is not sufficient to measure this construct, particularly if there are numerous actors and degrees of liking that need to be accounted for. It is suggested that future research develops a sound star liking measure. It is a particularly important construct in brandcast processing given that role modelling, endorsement theory and social learning theory are believed to underlie the effectiveness of brandcasting (see Section 1.1 and Section 6.4.2). As with connectedness, it needs to be determined whether star liking is purely a moderator or whether it also has a direct interaction with engrossment. It also needs to be determined whether star liking drives audience engrossment or whether audience engrossment drives star liking, or both (as denoted by the double-pronged arrow).

77 Although brand familiarity was found to have a significantly positive direct effect on recognition, this direct was not as strong as the moderated effect. No audience engrossment dimension was found to have a direct effect on recognition. These results further strengthen the argument that these factors are all moderators.  O              P     !  " # $% &  '    (+3+( Figure 6.14 – Basic Theoretical Model of Brandcast Processing (revised)

Brandcast Quality Memory Brand Modality Awareness Prominence Conscious / Star presence /usage Unconscious Preference Method of Depiction Quality Temporal Quality Plot Connection

Audience Product / Engrossment brand familiarity

Star liking Connectedness Feelings Arousal Cognitive effort Appraisal

 O              P     !  " # $% &  '    (+3,( 6.15 - Conclusion This chapter has outlined and described the methodology used to gather, measure and analyse the data for a preliminary testing of the theoretical model presented in Chapter 3. The research approach was justified, and important issues regarding the selection of the movie discussed and debated. The operationalisation of measures and construction of questionnaires was discussed, and the research procedure explained. The results of the quasi-experiment demonstrated reasonable support for the hypotheses outlined at the beginning of this chapter, that product placements are recognised by audiences, and that these levels of recognition are moderated by brand familiarity and audience engrossment. These results and their broader implications will be further discussed in the next chapter. Finally, a revised model of brandcast processing was introduced to accommodate the concept of star liking which was found to not be captured by Audience Engrossment

 O              P     !  " # $% &  '    (+3-( CHAPTER 7: CONCLUSIONS AND IMPLICATIONS

“Only the curious will learn and only the resolute overcome the obstacles to learning”

Eugene S. Wilson

7.1 – Introduction Identifying a need for more research to be conducted from the perspective of the audience member, this thesis developed a theoretical model of brandcast processing and tested the role of brand familiarity and audience engrossment on product placement recognition. The model assumes that audiences bring to every brandcast encounter characteristics that determine how the brandcast is processed, with the base model premised on the notion that whilst a number of factors may influence processing, two elements are key to every processing occasion - audience engrossment and brand familiarity. The brandcast itself features as the stimulus, and is comprised of a bundle of characteristics. Realistic effects of this processing are conscious and unconscious awareness and preference for the placed brand.

This chapter revisits the research questions introduced in Chapter 1 and shows how each of these was addressed. In particular, the development of the Audience Engrossment scale and its generalisability are discussed, as are the results of the quasi-experiment testing of the model presented in Chapter 3. The theoretical, methodological, managerial and societal contributions of the research are outlined, and limitations of this present research explained. Finally, directions for future research are proposed.

 O              P     !  " # $% &  '    (+3.( 7.2 – Conclusions relating to the research questions 7.2.1 - Conceptual clarification of product placement and related activities (RQ1, RQ2) As demonstrated in Chapter 2, there is confusion as to what constitutes a product placement as the practice has begun to morph into other activities, is applied to a range of different but related practices, and is used differently to how it was described in earlier definitions. Therefore, clarification of the boundaries of product placement and these other activities was needed. To do this, the broad term brandcasting was developed, which refers to the inclusion of products - branded or unbranded - in entertainment story content. It was then determined that there are two types of brandcasts – product placement (the inclusion of products – branded and/or unbranded – to support entertainment story content) and advertainment (the inclusion of products – branded and/or unbranded – whereby the entertainment story content supports the brand). These two activities can now be thought to lie on opposite ends of a continuum (see Figure 2.7.2), with key differences between the two highlighted in a new typology (see Table 2.7.2).

Product placement was also distinguished from sponsorship, endorsement and plugs in Section 2.7.2.

7.2.2 - Development of a new theoretical framework of brandcast processing (RQ3, RQ4, RQ5) Thorough reviews of the literature revealed that there was no strong operational product placement framework in existence. Whilst the Balasubramanian, Karrh and Patwardhan (2006) model provides a valuable guide to research gaps and possible research propositions, in essence it is not a processing model. Therefore, the gap remained, so this research developed a new model of brandcast processing. The model development in Chapter 3 identifies the core factors that affect brandcast processing, with a revised model introduced in Section 6.14.

 O              P     !  " # $% &  '    (+3/( In particular, the model identified both audience characteristics and product placement characteristics that impact on the respondent’s ability to recognise and remember the brands shown in an entertainment program (in this case, a movie). Although connectedness and star liking may affect processing, this model presumes that audience engrossment and brand familiarity should be included in all test situations as they are factors present in all brandcast processing. Brand familiarity will heighten awareness to the placement and enable it to be processed. Audience engrossment describes the degree of cognitive and emotional response that must be experienced when consuming any entertainment. These factors may either inhibit or enhance the processing of the product placement.

In contrast, connectedness (Russell et al. 2004) is only relevant for certain media vehicles, namely television, where a pre-existing connection is likely given its serial nature. Star liking is more relevant across the range of possible media vehicles. However not all brandcasts, especially advertainment, will feature a celebrity. Therefore, these two moderator variables should be added to the model as appropriate.

Our processing model is premised on the assumption that for a brandcast to be effective, a memory trace should be created following exposure to that brandcast. Indeed, the creation of this memory trace (either conscious or unconscious) should be the fundamental goal of any brandcasting activity since all other effects flow on from this trace, with awareness and preference for the brand being the next logical effects. This in turn may lead to other effects such as altered brand attitude or purchase intention. These effects can be assessed by using explicit or implicit memory tests such as recall, recognition and preference based measures.

7.2.3 - Development of the Audience Engrossment Scale (RQ6, RQ7, RQ8) Extensive reviews of the literature revealed that there was no existing scale that was suitable for measuring audience engrossment - “the degree to which individuals are engaged - affectively, cognitively and behaviourally - with the entertainment content they are consuming, at the time of consumption” (Scott and Craig-Lees 2005a, p2). Therefore, a new scale was developed for this research and was refined using Rasch analysis. The use of  O              P     !  " # $% &  '    (+30( Rasch Measurement Theory allowed us to develop a robust scale which covered the range of the construct and offered precise measurement.

It was hypothesised that audience engrossment was comprised of four dimensions – feelings, arousal, appraisal and cognitive effort. These four dimensions clearly emerged through the Rasch analysis. From an original list of 81 items, the final Audience Engrossment scale was comprised of 19 feeling items, 10 arousal items, 6 appraisal items and 7 cognitive effort items (see Section 5.5).

As discussed in Section 5.6, the Audience Engrossment scale demonstrated various levels of generalisability. Not only did it show invariance between viewers of the same movie, but it also identified items, dimensions and a model that could be used to effectively measure the level of audience engrossment of viewers across a range of movies. Furthermore, the items generally had similar item locations and similar ordering of difficulty regardless of the film, particularly for arousal and appraisal (see Appendix 5.24). Any differences in item location in the feelings dimension are likely due to different types of movies evoking different types and intensities of feeling. Variation of response in this feeling dimension is more likely than in the other dimensions where film type and theme should have less of an impact on response. Plotting the item locations of different movies against each other and seeing that their position generally fell within the 95% confidence interval around an exact fit also demonstrated that the scale had high levels of invariance between movies (see Appendices 5.25 and 5.26). Indeed, these graphical illustrations provide greater confidence in the data and provide a greater level of generalisability than simply focussing on invariance of viewers’ responses for each movie (i.e. that the model fits the data from a movie or even multiple movies).

Once developed and validated, the Audience Engrossment scale allowed us to better understand audiences and how they process entertainment stories and how this may impact the subsequent recognition of any brands placed in that story. This was done by featuring audience engrossment in the theoretical model as a moderating variable. To do this, we  O              P     !  " # $% &  '    (+31( investigated the interactions of audience engrossment’s four underlying dimensions with each of the product placement characteristics.

7.2.4 – Direct effects of product placement quality on product placement recognition (RQ9) All product placement characteristics were hypothesised to have a positive impact on recognition, with certain characteristics (prominence, plot connection, modality and use by star) branded higher quality and anticipated to have a stronger impact. All four of these characteristics were found to have the strongest direct impact on recognition, whilst prominence, plot connection and modality were also found to have strong effects on recognition when moderated by audience engrossment.

The importance of prominence found in this research is consistent with prior research measuring its role in product placement recall and recognition (e.g. Gupta and Lord 1998; Brennan et al. 1999; Law and Braun 2000). However, the results of plot connection appear to contradict past research, with both d’Astous and Chartier (2000) and Russell (2002) finding that integration and congruence led to lower recall (but greater liking and persuasiveness). However, it is not known if it is the operationalisation of the plot connection concept that led to different results. Whilst the idea of plot connection and integration seems theoretically sound, the operationalisation of it is difficult. Neither of these other studies elaborated much on how it was done, except to imply that their decisions were subjective judgements. In this study however, a brand was deemed to have high plot connection if the scene could not have existed without the product (for example, car chase scenes could not exist without cars, although the brand of the cars did not matter). In this way, it is not necessarily related to prominence or use by star (thus minimising collinearity effects) and is a more objective way of characterising a placement. We believe it is this operationalisation that led to the strong positive impact of plot connection on recognition in this study.

 O              P     !  " # $% &  '    (+32( The positive impact of total screen time on recognition was consistent with pre-existing literature (e.g. Bandura 1977; Baddeley 1990), but was not one of the more influential factors. From a practitioner’s perspective, what is significant about these findings is that, as noted by Turcotte (1995), product placement contracts routinely contain a minimum airtime clause, and advertisers pay for the amount of airtime they are to receive. However, what is still useful to know, and what could not be examined in this study because of collinearity issues is whether total screen time or frequency of exposure is more important. For example, when buying 30 seconds of screen time, is it better to have the brand appear just once for the whole 30 seconds, or on three separate occasions for 10 seconds each?

A major finding that was in contrast to that of the pre-existing literature was the effect of modality. Although a strong predictor, its coefficient was consistently negative (i.e. dual modality led to lower recognition compared to single modality), whilst prior research suggests it should be positive (e.g. Gupta and Lord 1998; Law and Braun 2000). Similarly, when examined individually, audio placements and visual placements were both found to have negative impacts on recognition. These results were bewildering, especially since modality is an essential product placement characteristic.

Closer investigation of the actual film coding form revealed that most placements (30 out of 39) were categorised as visual only, with four being aural only (leaving only five with dual modality) (see Appendix 6.6). Therefore, with little variability in the modality variable, it is not surprising that the model testing could not accurately account for effects of modality. A recommendation for future research would be to identify a film that had a greater mix of visual, aural and audiovisual placements in an attempt to better isolate the effects of this factor on recognition. It is a particularly important factor to clarify since it reflects a common decision that needs to be made by advertisers and content producers, and also because prior research has recorded conflicting results regarding the relative impacts of the different modes (e.g. Gupta and Lord 1998; Law and Braun 2000; Russell 2002; Scott and Craig-Lees 2004).

 O              P     !  " # $% &  '    (+*3( The factors relating to star presence and star use also produced interesting results. Star presence rarely had a significant impact on recognition, contrary to prior research by d’Astous and Chartier (2000). When modelled on its own, it had a direct impact on recognition for those that had seen the movie before, but when modelled alongside all the other product placement characteristics, it was only found to have a significant impact on recognition for those who had not seen the movie before. In terms of its interaction with the audience engrossment dimensions, it was only found to have a significant effect when interacting with feelings for those who had not seen the movie before. This may be because it simply has minimal impact, or because only five of the 39 placements had star presence.

Star use on the other hand was found to have strong significant direct effects on recognition. What was particularly interesting about this result was that star liking (‘The movie starred one of my favourite actors’) was relatively low (mean = 2.4 out of 4) and was not found to have a significant relationship with recognition (p=0.156)78. Source attractiveness and endorsement theory suggest that if the audience likes the star, they are more likely to pay attention to those brands the star is associated with, perceiving the product categories and brands in the film to be an accessible avenue to associating themselves with the star’s desirable characteristics. In this way, they may model their behaviour on that of the star. However, these results suggest that individuals will consciously process a brand used by a star even if the star is not one of their favourites. In this study, just having a celebrity use or mention a brand was enough to invoke recognition, and potentially other effects (e.g. enhanced brand preference, usage etc).

The significance of these results relating to use by star is from an ethical / societal perspective. Overall recognition rates for the brands featuring in the movie were quite high, but were enhanced by the star using or mentioning a brand. Therefore, indications are that the star does potentially have significant influence over (teenage) audiences, and for this

78 It should be noted however that star liking did have significant relationships with all four dimensions of Audience Engrossment – feelings (r=0.36, p=0.000), arousal (r=0.30, p=0.000), appraisal (r=0.42, p=0.000) and cognitive effort (r=-0.03, p=0.022). It was also strongly correlated to Program Liking (r=0.44, p=0.000).  O              P     !  " # $% &  '    (+**( reason, it can be argued that celebrities need to use their moral judgement and take responsibility for the products they are implicitly “endorsing” by sharing the screen with, and perhaps even consuming on-screen (especially cigarettes and alcohol). Similarly, they need to be made aware of the potential influence that they have over audiences, especially younger audiences as in this study. Unconscious processing, not tapped into in this research could also be at work, which could be making these effects even stronger.

The patterning on the film coding forms (see Appendix 6.6) revealed that recognition was highest when all or many of the product placement characteristics were present. Once a number of these characteristics become absent, recognition began to fall. For example, msn had the highest level of recognition (93.2%), and was the subject of both on-set and creative placements, was used by the star, had its logo and actual product shown, was highly connected to the plot and appeared visually. Furthermore, it appeared on screen for 16.7 seconds (a relatively high amount of time). In contrast, some signage for Bacardi, which had one of the lowest recognition rates (1.6%), was simply shown in the background of another scene. Looking more generally at the patterns that appeared on the content analysis coding form, it would appear that placements with high recognition rates tended to have at least three exposures (thus accumulating more screen time) and had all of the following – onset prominence, high plot connection, use by star and showed the actual brand. Those with low recognition tended to be creative (i.e. low prominence), have only one or two exposures, have low plot connection, and all were visual.

7.2.5 – Moderating effect of brand familiarity on product placement recognition (RQ10) Qualitative work by DeLorme and Reid (1999, p78) suggests that "moviegoers were particularly attuned to familiar branded products and services that they themselves had previously purchased and consumed in their everyday lives". This was confirmed, with familiarity of a particular brand having a strong significant positive moderating effect on product placement recognition of that brand. The strongest interactions were with

 O              P     !  " # $% &  '    (+*+( prominence, modality and plot connection, but significant interactions were also found with use by star, total seconds on screen and depiction of the actual brand.

This assessment of brand familiarity-recognition matching highlights the importance of placing brands relevant to the specific audience in a particular movie, and confirms past research by Scott and Craig-Lees (2003; 2004) and Brennan and Babin (2004). Indeed, audiences are more willing to process personally relevant messages and find it easier to process information about products that are familiar to them and then recall them or discriminate between similar products (Petty et al. 1983; Celsi and Olson 1988; Babin and Carder 1996; DeLorme and Reid 1999).

Whilst there was a generally high familiarity level of all the brands in The Island, this did not automatically mean that these brands would be recognised (although it did make it more likely). Recognition was more governed by the product placement characteristics, so if the placement was not particularly strong (i.e. low prominence, low plot connection, low amount of screen time etc), brand familiarity alone was not enough for it to be noticed. On the other hand, a clever placement (use by star, prominent, actual product shown) in a funny scene could see a brand with low familiarity (Popular Mechanics, 7.5% familiar) accrue a strong recognition rate of 20.5%. Future research should consider exploring these ideas further. Can a familiar brand still get noticed with a weak placement? Can unfamiliar brands get noticed by strong placements? Such results may reveal that familiar brands do not need to spend extra money to have a higher quality placement (i.e. a less prominent placement might be enough), or may mean that for unfamiliar brands (i.e. new brands or new markets), it might be worth placing your brand in a particular movie if you are willing to spend money on a high quality placement if this means your brand will get noticed and remembered.

 O              P     !  " # $% &  '    (+*,( 7.2.6 – Moderating effect of audience engrossment on product placement recognition (RQ11) The importance of the audience engrossment dimensions to product placement processing was confirmed as each was found to have a significant moderating role with certain product placement characteristics, most notably prominence, plot connection, modality, use by star and total seconds onscreen. At times they were found to moderate with actual brand and logo / ad / signage, but rarely with star presence. Feelings and arousal were found to interact significantly with most product placement elements. In contrast, appraisal did not significantly interact with as many elements, especially if the respondent had already seen the movie before.

Overall, it could be said that high levels of audience engrossment actually have a negative impact on recognition, given that three of the four dimensions (feelings, arousal and cognitive effort) were found to have inhibitory effects. Only appraisal was found to enhance recognition. That said, none of these findings were unexpected, and were all generally consistent with the hypotheses outlined in Section 6.2.

If audiences liked the characters, enjoyed the story and felt interested and immersed in it and that they would like to watch this movie again because they enjoyed it so much, their overall appraisal of the movie was high and product placements were more likely to be recognised. This is consistent with the concept of processing efficiency whereby people learn more material when in a positive mood. This is because pleasant states facilitate learning by leading to the activation of broad, well integrated cognitive categories that enhance stimulus encoding and subsequent ad recall (Isen 1984). Therefore, these audience members are likely to be taking in as much information as they can about the movie, and in this way, unconsciously or consciously, simultaneously process the product placements. Significantly to this research, they are able to recognise these placements in the short-term, immediately following exposure to the movie.

 O              P     !  " # $% &  '    (+*-( In contrast, high arousal levels have generally been found to disrupt information processing, particularly when the task is complex (Eysenck 1982; Sanbonmatsu and Kardes 1988). If individuals scored high on the feelings and arousal dimensions it is interpreted as experiencing many intense feelings and reactions. Similarly, scoring high on the cognitive effort dimension meant that the individual found the story difficult and had to pay a lot of attention to understand it. Under these high arousal and effort levels, processing product placements (a secondary task) is sacrificed in order to focus on the primary task of following the storyline.

7.3 – Contributions of this Research 7.3.1 – Theoretical Contribution Relatively speaking, product placement research is still in its early stages. This research has attempted to add to the base of knowledge in several ways by clarifying and expanding existing knowledge.

Examination of prior research suggested that product placement needed to be better defined, so this research offered a new definition of product placement. In attempts to delineate product placement from other related activities, this research developed the concept of brandcasting, which encompasses both product placement and advertainment, as both relate to the insertion of (branded) products into entertainment story content, but have slightly different purposes. Product placement was also clearly distinguished from sponsorship, plugs and endorsement.

Balasubramanian, Karrh and Patwardhan (2006) suggest that product placement research has been held back by the absence of a strong conceptual model of product placement. Whilst they offer one, their model is not a model of product placement processing. Given that the processing of product placements (and indeed any brandcast) is unique given its concurrent nature (i.e. processing brands simultaneously with entertainment content), it is important to understand this processing in terms of what factors may impact it and what

 O              P     !  " # $% &  '    (+*.( effects could result. Therefore, this research offers a basic operational model of brandcast processing.

To do this demanded a re-examination of the memory and information processing literature in order to determine what effects were actually viable from brandcasts. We then argued that these, and only these, should feature in a model of brandcast processing. Brandcasts may be processed either consciously or unconsciously, with Chapter 2 suggesting that different tests need to be conducted to detect different effects, and that explicit memory tests are unsuitable for studying outcomes such as brand attitude, brand image, and purchase intention which would be better suited to implicit memory testing. That said, explicit measures are suitable for measuring conscious memory of the brandcast, and this was the approach adopted for this research.

The new model of brandcast processing is a key contribution of this research. In terms of effects, it clearly reflects that the most that a brandcast can truly independently impact is memory and preference for a brand. It further extended past research by including both audience factors (audience engrossment, brand familiarity, connectedness and star liking) and executional factors (i.e. the various product placement characteristics). Both perspectives are equally important because if effective product placement decisions are to be made, product placement managers need to have some idea of both types of factors and their relative impact.

Prior to this research, no single study had investigated all the variables that could impact recognition (or any form of conscious memory). Rather, memory based research simply tended to look to see how many brands could be remembered and offered no real explanations for this (e.g. Steortz 1987; Babin and Carder 1996). If different executional factors were considered, generally no more than one or two were included in a single study (e.g. Sapolsky and Kinney 1994; Gupta and Lord 1998; Brennan et al. 1999; Russell 2002). Examining all the product placement characteristics and their interactions offers much more knowledge than only considering one or two, with the effects of the interactions in this  O              P     !  " # $% &  '    (+*/( research demonstrating that these factors should not be looked at in isolation (see Sections 6.11 and 6.12).

Whilst not a new idea, focussing on the audience member is not common (especially in marketing), so this research has contributed to theory by promoting the idea of conducting more research from the perspective of the audience member, giving them an active role in the process, and not just assuming they are passively ‘advertised to’ or ‘entertained’. Indeed, other researchers such as Perse (1998) suggest that cognitive and affective involvement with content may be more strongly linked to individual-level variables than to content variables.

Another key contribution of this research was the development of the Audience Engrossment concept and a scale to measure it. This built on the idea of focussing on the individual, and examined how individual reactions to content affect processing and enjoyment of entertainment programs. The scale was rigorously developed and was found to work across movies of different genres, describing an audience member’s engrossment with a particular film via its four dimensions of feelings, arousal, appraisal and cognitive effort. Furthermore, when tested in a product placement context, audience engrossment was found to have a significant moderating impact on recognition of product placements. Such findings demonstrate the significant role that individual characteristics can have on the success of a product placement, and highlight the importance of continued focus on the individual.

7.3.2 – Methodological Contribution The most significant methodological contribution this thesis offers is the application of Item Response Theory, in particular Rasch modelling, in a marketing context. This approach was selected because it is more closely aligned with the scientific task of measurement and it offers more precise measurement than Classical Test Theory. Traditionally Rasch measurement has been used in education, with recent application in the health sciences and psychology, yet it remains in its infancy in marketing. However, for the  O              P     !  " # $% &  '    (+*0( many reasons outlined in Chapter 4, and the insights it could provide in the scale development and refinement process (as detailed in Chapter 5), and for the rigour it demands in testing for measurement (and not just assuming measurement), further implementation of it in the marketing discipline may help marketers develop stronger measures that can stand the tests of time, different samples, and different contexts.

The issues Rasch analysis highlights in one’s data are often not detected using other methods, leading one to believe that these problems do not exist. A good example of this is the response scale category structure (examined via threshold analysis), something that is generally taken for granted and assumed to work correctly. However, as was seen in the refinement of the Audience Engrossment scale, one cannot assume the standard five-point Likert scale actually works, or that scale categories understood by the researcher will be interpreted the same way or be readily distinguished by the respondent (Scott and Salzberger 2008). This highlights the need to treat response scales seriously, and to consider their wording and the number of categories as critically as one does the items underneath them.

7.3.3 – Managerial Contribution The results of this research aim to provide product placement planners with some guidance when making product placement decisions, and ensure that they get maximum value for money for each placement. Results of this research would suggest the following guidelines for practitioners (although further confirmatory research is still necessary).

In regards to the product placement itself: • Onset placements (i.e. those that appear prominently in a scene by virtue of size or position) are more effective than creative placements (i.e. those that appear in the background of a scene or are small in size). • Find a way to make your brand central to the action so that the story could not exist without it (i.e. high plot connection).

 O              P     !  " # $% &  '    (+*1( • Your brand being present in the same scene or even the same shot as a star is not enough; the star needs to actually use or mention your brand. That said, this research suggests that any major star will do – it does not need to be the audience’s favourite. • Placements are more effective when the audience is already familiar with your particular product category / brand (i.e. it is important to match your product to the movie audience). Therefore, product placement may not be a suitable strategy for launching a new product or introducing it to a new market – unless perhaps you are willing to pay for a very strong placement (see Section 7.2.5). • The highest premium should be paid for those placements that combine all product placement elements (i.e. multiple exposures, high screen time, high plot connection, combination of both on-set and creative placements, use by star).

In regards to film selection: • Avoid films that will generate too many intense feelings. • Avoid films that generate negative arousal (i.e. sad, scared, tense, angry) and opt for films which leave people more calm and relaxed. • Find a film with likeable characters. • The plot should be easy to follow and not require high levels of attention or effort to understand. • The film should be one that the audience really enjoys, where they are interested in the story and cannot wait to see what happens next; one they like so much that they want to see it again. However, this is easier said than done, and is surely the goal of every filmmaker. Creating a magic formula for such a film is another study in itself! Indeed, Krider (2006) highlights the need for a forecasting model comprised of variables such as themes, characters, dialogue and story structures that may predict movie success.

 O              P     !  " # $% &  '    (+*2( 7.3.4 – Societal Contribution Several benchmark figures have been developed in order to ascertain whether a product placement has been “successful” or “unsuccessful”. According to Steortz (1987), if 20% or more of the audience is aware of a brand placed in a film, industry professionals consider this to indicate effective communication. If 30% or more of the audience is aware of it, then the placement is considered to be very successful. Since many of the recognition results in this study are higher than Steortz’s recommended benchmark figures (16/39 have recognition rates higher than 20% and 9/39 have rates higher than 30%), it would lead one to conclude that the product placements in The Island were “successful”. But what does “successful” actually mean? Do these recognition scores lead to any positive behaviour by the target audience? Is their higher awareness of the brand attributable only to the product placement which operates in a world described by Solomon and Englis (1994) where consumers are surrounded by media and advertising? Is their attitude towards the brand more favourable, or are they more likely to buy it because of the product placement?

This research is unable to answer these questions. But what it can do is suggest that the high recognition levels accrued by brands means that these placements are being consciously processed by teenage audiences, and in this way, effects of influence may be taking place – we just cannot measure them fully yet. Teenagers have high cinema attendance and are particularly vulnerable since they often use entertainment vehicles to determine how products are to be used in everyday life and to assist them in forming an identity (Bandura 1977; Moschis and Churchill 1978; King and Moulton 1996). For this reason, the complaints of opponents to product placement may be well founded since product placements are being noticed by this vulnerable teenage audience, who may not perceive these brand appearances to be advertising. Of particular interest to this issue is the fact that star use was found to enhance recognition, even without a strong liking of that star. Such a finding needs to be investigated in further research, but it does highlight the potential influence celebrities can have over their audiences, and that some form of regulation to control the practice may be necessary.

 O              P     !  " # $% &  '    (++3( 7.4 – Limitations The contribution of this research must be viewed in the context of the methodological and sample limitations which may constrain its generalisability to other groups.

7.4.1 – Limitations of the Audience Engrossment scale development Methodological limitations It may have been preferable to have the same respondent watch multiple movies when developing the Audience Engrossment scale. As it stands, one does not know which effects are individual-based and which ones are movie-based79. If multiple movies had been watched by one person, the individual effects could have been teased out, and the engrossment of each respondent for each movie better isolated. However, this approach was abandoned for several reasons. Firstly, it was logistically impossible. Using actual cinema- goers, and then encouraging them to return within the next few days to see another movie, at their expense, was unreasonable. Secondly, there were also concerns about having respondents complete the questionnaire a second time (especially so soon after the first questionnaire) as they would be watching the movie knowing what the task ahead of them was. This may affect the way they watched the movie, and may also make them over- sensitive to the items, or distract them whilst they were watching it (e.g. I just grabbed hold of my chair – I must remember to note that on the questionnaire when I fill it in). Doing this study under different circumstances, with funds to pay for participants to see multiple movies, and the gap in the process long enough for memory effects to be minimised, may be a viable avenue for future research and for further validating the scale. Once this linked design is in place and the effects of the individual removed, greater generalisation can be drawn from the scale, and possible inferences made that Movie 1 is more engrossing than Movie 2. However, it must be noted that there is no constant level of audience engrossment within the person – every movie triggers a different level within that same person.

79 However, as stated in Section 7.2.4, none of these differences were due to any general audience characteristics such as gender, age, movie-watching frequency, education or critical expertise.  O              P     !  " # $% &  '    (++*( 7.4.2 – Limitations of the Brandcast Processing Model testing Methodological limitations Whilst the use of only one film did assist in maximising the amount of observed variance between respondents relative to their engrossment with the film, it may be argued that this forced exposure is not a realistic scenario. In the real world, the audience is self-selected, with people having a choice of what movie they see. For this reason, audiences may not always include the full range of viewers from low to high levels of engrossment, since the audience has generally chosen to watch that particular film, making them more involved from the outset. This, of course, was avoided with the development and refinement of the Audience Engrossment scale as these audiences were all self-selecting, but it may be a factor in the model testing stage of this research when all participants watched The Island.

Another issue with only using one film is that whilst we know that different film genres (for example, action thriller versus romantic comedy) would demand of the audience differing levels of attention and evoke different feelings and reactions, we do not know how this would affect product placement recognition. By only using one film, this factor is unable to be investigated, and the generalisability of the results may be reduced.

One final methodological limitation of this research was that the respondents participated in a simulated theatrical experiment with a film available on DVD. Whilst prior exposure to the selected film was addressed, ideally, a first-run film currently being shown in the cinemas or not yet released would be preferable. In this case, one could be sure of no prior unconscious learning from exposure to the film. However, for logistical reasons and the need to code the movie accurately beforehand, this was unavoidable.

Sample limitations New South Wales high school students now study film as part of their compulsory English courses which may make them more aware of, and sympathetic to product placements than would be the case with a more typical audience member. Furthermore, the use of this teenage sample suggests that these results may only hold true for this particular group  O              P     !  " # $% &  '    (+++( relative to this particular movie. To make the results more generalisable, and to minimise any effects of media savvy-ness, a larger and broader sample is necessary, and more movies need to be examined.

There is also the concern that the sample was unbalanced gender wise (approximately 2/3 female; 1/3 male). This skew makes it more difficult to infer broad meaning from the results. There are indications that the female population related to the brands in the movie differently than did the males. For example, as stated in Chapter 6, it is believed that gender effects actually drove the brand familiarity, which then in turn affected recognition.

7.5 – Future Research Directions There are several opportunities to extend both the audience engrossment and product placement aspects of this research. There are also several research issues that still need to be resolved.

7.5.1 – Research issues that still need to be resolved Research with a stronger focus on star liking needs to be conducted, as to date, little has been done (except for Scott and Craig-Lees 2003). For this, a “star liking” construct needs to be developed since no existing scales measuring celebrity expertise, trustworthiness or attractiveness are relevant for this task (e.g. Ohanian 1990). Some of the star liking items from the original Audience Engrossment scale may serve as a springboard to this new measure, as they capture degrees of liking (and not liking) (see Section 4.7). Alternatively, the items used by Scott and Craig-Lees (2003) could also be used as a starting point. A further recommendation would be to note precisely which actor is present or using a given brand, and to ask for more specific star liking information regarding each actor (and not just a general question as this research asked). This allows for tests which may render particular celebrities more influential than others.

 O              P     !  " # $% &  '    (++,( This research unearthed many issues with the coding of the film which need to be considered. As one of the first studies to consider so many independent variables, these issues have most likely never arisen before, as most studies only consider one or two executional factors. But the issue found here was that many of these factors are highly inter-related, and that this is unavoidable. For example, a subtle placement is likely to be visual and have no connection to the plot (i.e. it is displayed visually in the background of a shot whilst the action takes place around it). Or alternatively, a brand is used by the star and is thus a central part of the action (e.g. the hero drives a very fast car to get away from the bad guys, with gratuitous close-ups of the car’s badge, or a verbal mention of “I love my Ferrari!”). These placements cause multi-collinearity in the data, and as seen in the latter stages of Chapter 6, this can cause co-efficients to have unexpected signs, or dilute or overplay the effects of certain variables. Indeed, this research found the behaviour of some product placement characteristics (namely depiction of the actual brand) differed when examined individually as opposed to when bundled with the other product placement characteristics.

The typology offered by d’Astous and Seguin (1999) may alleviate this problem to some extent as the first placement would be deemed implicit, and the second one integrated explicit, and could be coded in that way. However, we believe the detail of all the different possible executions outlined in this study is necessary, and it is useful for managers to know whether it is better to buy one minute of airtime and have the brand appear ten times for six seconds each, or have the brand be the subject of one entire one minute scene. Is a verbal or visual placement more effective? Does the star actually need to use the product or is it enough for it to be placed prominently on a table next to them, or be elsewhere in the movie but be zoomed in on? Unless films can be coded to this detail and then effects tested, researchers cannot offer these recommendations to managers. Furthermore, since product placements are processed as a bundle of characteristics, there needs to be an understanding of how the different executional factors inter-relate and which ones are more conducive to conscious and unconscious memory (as they may be different).

 O              P     !  " # $% &  '    (++-( One solution may be to code placements very specifically, according to a bundle of characteristics. Doing this retains the specific detail, but would minimise the collinearity amongst the characteristics. For example, instead of separate codings for prominence and plot connection (i.e. each being scored “1”), a new bundle may be created which is a single variable of high prominence + high plot connection and that same placement would be scored a single “1”. Other logical bundles may include: high prominence + high plot connection + star use; or high prominence + low plot connection + star use.

7.5.2 – Extending Audience Engrossment Research The next stage of audience engrossment research must continue to test the boundaries of generalisability of the Audience Engrossment scale. In short, more movies should be tested to ensure that the same list of items for each dimension continues to be effective. This can be done by comparing films of the same genre, and then comparing differences between genres to see if the scale works the same way between genres and to see if items continue to have similar locations and relative ordering of difficulty. If there is variance between genres or between films, this is not a shortcoming of the scale. Rather, it reflects the fact that there are qualitative differences between genres and films – and this is a valuable research finding. Perhaps it is an impossible task to develop a scale that measures all dimensions for all movies. Because of this strong focus on invariance, it is recommended that Rasch analysis be used in all subsequent testing of the Audience Engrossment scale since one of its assumptions is invariance and it is particularly sensitive to violations of this assumption. Rasch analysis will therefore identify invariance problems with the scale that a less sensitive method (e.g. factor analysis) most likely would not.

Another way to test the generalisability of the Audience Engrossment scale is to test it in other similar contexts, namely television viewing and plays. It is doubtful it would apply to wider contexts such as magazines and radio as these are processed differently and the media vehicles themselves have significantly different characteristics.

 O              P     !  " # $% &  '    (++.( Whilst focus thus far has focussed on assessing effects of audience engrossment when the brand is embedded in entertainment content, it is also worth exploring the effects that audience engrossment has on the recall and recognition of traditional television advertising to see how engrossment with the program affects conscious memory for the ads that appear in between programs and segments (i.e. those that interrupt the program and are not a seamless part of the content like product placement). Are the effects different?

Finally, there is potential for an overall Audience Engrossment score to be computed, thus taking the focus off the four individual dimensions. In that way, an individual could be given a score for a particular movie, and an individual could then be deemed to be more or less engrossed than another individual, or one movie more or less engrossing than another. The development of this score however would require considerable effort. A simple short- term solution could be to sum the raw scores of the individuals’ responses, and then say that a higher score denotes higher levels of engrossment. However, because each dimension is comprised of a different number of items, doing this would assume that some dimensions are more important than others and they would have higher weightings. The idea that the dimensions have different weightings and impacts on overall engrossment needs to be further explored.

7.5.3 – Extending Product Placement Research 7.5.3.1 - Linking Audience Engrossment and Product Placement Effectiveness In regards to using the Audience Engrossment scale to test the model, future research may consider examining the impact of individual items or re-grouping items within each dimension to explore their effects on product placement. For example, the feelings dimension has been found to have a negative impact on recognition. But perhaps the positive feelings would have a different effect to the negative feelings. This was not attempted here as Rasch analysis revealed that these items constituted one dimension and worked better this way (see Section 5.3.3). Similarly, individual items might have stronger impacts than others, so this is a much-recommended next step.

 O              P     !  " # $% &  '    (++/( 7.5.3.2 - Understanding the boundaries and working towards generalisability Further testing of the brandcast processing model is recommended in the form various replications and extensions: • Using larger and different samples • Using more or different movies as the stimuli. • Exploring other forms of entertainment media, namely television and gaming. Z Despite a large proportion of product placement research being conducted in a film context, television remains the most popular medium for paid product placement80. Research in a television context will allow for the full processing model to be tested since connectedness may become a highly relevant factor. Z Product placement in video games is a growing area81 and therefore needs to be better understood. It is a very different processing environment to the passive story- based environment of films and television, and should include the concept of flow (where challenge and skill are highly congruent) as a moderator since there is a high level of ability and active decision making and control over the story / course of events (see Section 3.7). • Since it is a model of brandcast processing, the model needs to be tested in an advertainment context (not just product placements).

7.5.3.3 – Developing an overall product placement quality score An overall quality score awarded to each product placement based on its inherent characteristics would allow different product placements to be more easily compared empirically. However, to calculate such a score, weightings would need to be given to different factors, depending on what is found to be more effective. Perhaps experiments are the recommended course of action to tease out these effects.

80 Product placement on television accounted for 71.4% of global product placement spending in 2006 (US$2.4 billion, with projected growth of 34% in 2007). Over this same period, product placement in film was worth US$885.1 million (or 26.4% of total product placement spending). However, this value of product placement in film is anticipated to grow 20.5% in 2007 due to more cross-promotional packages linking movie placements to ad spots (PQ Media 2007). 81 Although paid placement in video games and music accounts for only 2% of total spending, this is due for growth of over 30% in the next few years in a bid to target the 18-34 year old demographic (PQ Media 2007).  O              P     !  " # $% &  '    (++0( That said, there is a strong concern about taking this type of research out of the ‘real world’ and into the laboratory. Moorman (2003) found that experiments and real life settings showed opposing effects for the influence of media context on advertising effects, and that the forced exposure situation hinders realistic evaluation of the effects of medium context. Similarly, van Reijmersdal, Neijens and Smit (2007) conducted two stages of research – a questionnaire and an experiment - to look at the effects of product placement on brand image using implicit tests. In the questionnaire stage, they found that the brand image of people who watched more than two episodes of the program was more in agreement with the program image. However, in the experimentation stage, any level of exposure to the program resulted in scores on brand image becoming more in agreement with scores on program image. These results further suggest that experiments might not be the best way to test for product placement effects, as they are not reflective of the true viewing environment since participants purposely pay more attention and require fewer exposures for there to be an effect.

7.5.3.4 – Taking a longer term perspective of potential effects It remains unknown whether the effects on memory found in this research will remain for longer periods of time. For this study, in order to isolate the effects of recognition and directly link these to the product placements contained within the film, recognition was measured immediately after seeing the film (although there was a slight distracter task with respondents completing the Audience Engrossment scale first). Given that long-term memory effects are important for building brand equity, more research should be conducted with longer term measures in mind.

Once there is satisfactory evidence supporting the fact that recognition and conscious processing of product placements is taking place, research examining other potential impacts of product placement (i.e. brand preference, brand attitudes and purchase intention) is necessary to address the inconsistencies in the relatively few studies that have been conducted, and to ascertain the true effectiveness of product placement. This is particularly pertinent given the developments in technology which may make instant purchase of

 O              P     !  " # $% &  '    (++1( products seen in these programs more of a reality (see Section 1.2). Much of this work may be more suited to implicit memory tests.

As opposed to the strong focus on cognitive (memory) effects, more interpretive research examining how product placements are processed by movie audiences will also provide a more positive step towards understanding whether product placements really do have any influence over audiences. Understanding the forces that lead audiences to want to emulate characters through brand use could help describe not only whether advertisers should employ product placement as part of their promotional mix, but also, how best they could do so.

Finally, as product placement and advertainment become more prevalent, the practice may lose its novelty and become subject to the clutter that has characterised advertising over the last few decades (Hudson and Hudson 2006). Therefore research must continue to test brandcasting’s effectiveness and to see whether these effects and attitudes to the practice change over time. Content analysis studies which describe the incidence of product placement in different media should also continue to be conducted to track this growth. This is a particularly interesting area given that product placement is only really suited to certain genres. For example, it is not likely to be found in fantasy-style movies (like Lord of the Rings), period dramas or documentaries. With many productions relying increasingly on product placement to subsidise budgets, could this result in certain types of films and programs not being made? Could the convergence of advertising and entertainment see advertisers controlling what entertainment gets made in the future?

7.6 - Conclusion This research has provided some insight into the factors affecting recognition of product placements in film from both an audience and an executional perspective. This chapter aimed to draw conclusions regarding the research hypotheses, and from these, infer implications for theory and practice and outline major contributions of this research. After acknowledging the limitations of this study, areas of future research were proposed.  O              P     !  " # $% &  '    (++2( REFERENCES

Ailawadi, K.L., K. Gedenk, and S.A. Neslin (1999), "Heterogeneity and Purchase Event Feedback in Choice Models: An Empirical Analysis with Implications for Model Building," International Journal of Research in Marketing, 16 (3), 177-198.

Alba, J.W. and J.W Hutchinson (1987), "Dimensions of Consumer Expertise," Journal of Consumer Research, 13 (March), 411-454.

Anderson, D.R. and J. Burns (1991), "Paying attention to television," in Responding to the Screen: Reception and Reaction Processes, J. Bryant and D. Zillman, Ed. NJ: Erlbaum. 3- 25

Andrich, D. (1985), "An elaboration of Guttman scaling with Rasch models for measurement," in Social Methodology, N.B. Tuma, Ed. San Francisco: Jossey-Bass.

---- (1988a), "A General Form of Rasch's Extended Logistic Model for Partial Credit Scoring," Applied Measurement in Education, 1 (4), 363-378.

---- (1988b), Rasch Models for Measurement. Newberry Park, CA: Sage Publications.

Andrich, D., A. Lyne, B.S. Sheridan, and G. Luo (2003), "RUMM2020." Perth: RUMM Laboratory.

Andrich, D. and I. Styles (2004), "Final report on the psychometric analysis of the Early Development Instrument (EDI) using the Rasch Model: A technical paper commissioned for the development of the Australian Early Development Instrument (AEDI)." Perth, WA: Murdoch University.

Argan, M., M.N. Velioglu, and M.T. Argan (2007), "Audience Attitudes towards Product Placement in Movies: A case from Turkey," The Journal of American Academy of Business, Cambridge, 11 (1), 161-167.

Australian Bureau of Statistics (2007a), "2006 Australian Census."

---- (2007b), "Attendance at Selected Cultural Venues, Australia," Vol. cat no 4114.0: Canberra, Australia.

Australian Film Commission (2008), "Get the Picture" Available: http://www.afc.gov.au/gtp/wcfast.html#Rai77746, (accessed 2 January 2008)

Auty, S. and C. Lewis (2004a), "The Delicious Paradox: Preconscious Processing of Product Placements by Children," in The Psychology of Entertainment Media: Blurring the

 O              P     !  " # $% &  '    (+,3( Lines Between Entertainment and Persuasion, L.J. Shrum, Ed. New Jersey: Lawrence Erlbaum Associates. 45-62

Auty, S. and C. Lewis (2004b), "Exploring Children's Choice: The Reminder Effect of Product Placement," Psychology & Marketing, 21 (9), 699-716.

Avery, R.J. and R Ferraro (2000), "Verisimilitude or Advertising? Brand Appearances on Prime-Time Television," Journal of Consumer Affairs, 34 (2), 217-231.

Babin, L.A. and S Carder (1996), "Viewers' Recognition of Brands Placed Within Films," International Journal of Advertising, 15, 140-151.

Baddeley, A.D. (1990), Human Memory: Theory and Practice. Boston: Allyn & Bacon.

---- (2007), Essentials of Human Memory. New York: Psychology Press.

Bagozzi, R.P., M. Gopinath, and P.U. Nyer (1999), "The Role of Emotions in Marketing," Journal of the Academy of Marketing Science, 27 (2), 184-206.

Baker, M.J. and H.A. Crawford (1996), "Product Placement," in 1996 Winter Marketing Educator Conference. Chicago.

Balasubramanian, S.K (1994), "Beyond Advertising and Publicity: Hybrid Messages and Public Policy Issues," Journal of Advertising, 23 (4), 29-47.

Balasubramanian, S.K, J.A. Karrh, and H. Patwardhan (2006), "Audience response to product placements: An integrative framework and future research agenda," Journal of Advertising, 35 (3), 115-141.

Bandura, A. (1977), Social Learning Theory. Englewood Cliffs, NJ: Prentice Hall.

Baron, R.A. (2001), Psychology (5th ed.). MA: Allyn and Bacon.

Barry, T.E. (2002), "In defense of the Hierarchy of Effects: a rejoinder to Weilbacher," Journal of Advertising Research, 42 (3), 44-49.

BCMA (2006), "Branded Content Marketing Association" Available: http://www.thebcma.info/, (accessed 13 November 2006)

Bettman, J.R. (1979), "Memory factors in consumer choice: A Review," Journal of Marketing, 43 (Spring), 37-53.

 O              P     !  " # $% &  '    (+,*( Bhakta, B., A. Tennant, M. Horton, G. Lawton, and D. Andrich (2005), "Using item response theory to explore the psychometric properties of extended matching questions examination in undergraduate medical education," BMC Medical Education, 5 (9), 1-13.

Bhatnagar, N., L. Aksoy, and S.A. Malkoc (2004), "Embedding brands within media content: The impact of message, media and consumer characteristics on placement efficiency," in The Psychology of Entertainment Media, L.J. Shrum, Ed. Mahwah, NJ: Lawrence Erlbaum.

Bloxham, M (1998), "Brand affinity and television programme sponsorship," International Journal of Advertising, 17 (1), 89-98.

Blumler, J.G. (1979), "The role of theory in uses and gratifications studies," Communication Research, 6, 9-36.

Bond, T.G. and C.M. Fox (2001), Applying the Rasch Model: Fundamental Measurement in Human Sciences. Mahwah, NJ: Lawrence Erlbaum Associates, Inc.

Brennan, I. and L.A. Babin (2004), "Brand Placement Recognition: The Influence of Presentation Mode and Brand Familiarity," Journal of Promotion Management, 10 (1/2), 185-202.

Brennan, I., K.M. Dubas, and L.A. Babin (1999), "The influence of product placement type and exposure time on product placement recognition," International Journal of Advertising, 18 (3), 323-337.

Brett, S.J. (1995), "The Shrinking Screen: The Increasing Intersection of Hollywood Film and Television Programming," PhD thesis, Northwestern University.

Burns, A.C. and R.F. Bush (1998), Marketing Research (2nd ed.). New Jersey: Prentice Hall.

Calder, B.J., L.W. Phillips, and A.M. Tybout (1981), "Designing Research for Application," Journal of Consumer Research, 8 (2), 197-207.

Campbell, D.T. and J.C. Stanley (1963), Experimental and Quasi-Experimental Designs for Research. Chicago: Rand McNally and Co.

Canning, S (2003), "The subliminal sell," in The Australian. 11-17 December ed.

Celsi, R.L. and J.C. Olson (1988), "The Role of Involvement in Attention and Comprehension Processes," Journal of Consumer Research, 15 (2), 210-225.

 O              P     !  " # $% &  '    (+,+( Chang, C.H. (1996), "Finding two dimensions is MMPI-2 Depression," Structural Equation Modelling, 3 (1), 41-49.

Chartrand, T.L. (2005), "The Role of Conscious Awareness in Consumer Behaviour," Journal of Consumer Psychology, 15 (3), 203-210.

Cherry, E.C. (1953), "Some experiments on the recognition of speech with one and two ears," Journal of the Acoustical Society of America, 25, 975-979.

Churchill, G.A. (1979), "A Paradigm for Developing Better Measures of Marketing Constructs," Journal of Marketing Research, 16 (February), 64-73.

Churchill, G.A. and J.P. Peter (1984), "Research design effects on the reliability of rating scales: A meta analysis," Journal of Marketing Research, 21 (4), 360-375.

Clancey, M.J. (1994), "The Television Audience Examined," Journal of Advertising Research, 34 (4), special insert.

Cohen, G. and M. Conway (2007), Memory in the Real World. New York: Psychology Press.

Cook, T.D. and D.T. Campbell (1979), Quasi-Experimentation: Design and Analysis Issues for Field Settings. USA: Houghton Mifflin.

Coupey, E, J.R. Irwin, and J.W. Payne (1998), "Product Category Familiarity and Preference Construction," Journal of Consumer Research, 24 (2), 459-468.

Cowlett, M (2000), "Make it into the movies," in Marketing.

Craig-Lees, M., J. Scott, and R. Wong (2006), "Product Placement Practitioners: A replication for an Australian Perspective," in Proceedings of the 2006 Australia and New Zealand Marketing Academy Conference (ANZMAC). Brisbane, Australia.

Csikszentmihalyi, M. (1990), Flow: The psychology of optimal experience. New York: Harper-Perennial.

---- (1997), Finding Flow: The psychology of engagement with everyday life. New York: Basic Books.

Csikszentmihalyi, M. and J. LeFevre (1989), "Optimal Experience in Work and Leisure," Journal of Personality and Social Psychology, 56 (5), 815-822.

 O              P     !  " # $% &  '    (+,,( d'Astous, A. and F. Chartier (2000), "A Study of Factors Affecting Consumer Evaluations and Memory of Product Placements in Movies," Journal of Current Issues and Research in Advertising, 22 (2), 31-40. d'Astous, A. and N. Seguin (1999), "Consumer reactions to product placement strategies in television sponsorship," European Journal of Marketing, 33 (9/10), 896-910.

Darlin, D. (1995), "Junior Mints, I’m Gonna Make You a Star," Forbes, 6 November, 90- 94.

De Vellis, R.F. (2003), Scale Development: Theory and Applications (2nd ed.). California: Sage Publications.

Deery, J. (2004), "Reality TV as Advertainment," Popular Communication, 2 (1), 1-20.

Deighton, J., D. Romer, and J. McQueen (1989), "Using Drama to Persuade," Journal of Consumer Research, 16 (December), 335-343.

DeLorme, D.E. and L.N. Reid (1999), "Moviegoers' experiences and interpretations of brands in films revisited," Journal of Advertising, 28 (2), 71-95.

DeLorme, D.E., L.N. Reid, and M.R. Zimmer (1994), "Brands in films: Young Moviegoers' Experiences and Interpretations," in 1994 Conference of the American Academy of Advertising. Tucson, Arizona.

Diener, B.J. (1993), "The frequency and context of alcohol and tobacco cues in daytime soap opera programs: Fall 1986 and Fall 1991," Journal of Public Policy and Marketing, 12 (2), 252-261.

Donaton, S. (2004), Madison & Vine. New York: McGraw Hill.

DuRant, R.H., E.S. Rome, M. Rich, and E. Allred (1997), "Tobacco and alcohol use behaviours portrayed in music videos: A content analysis," American Journal of Public Health, 87 (7), 1131-1135.

Easterbrook, J.A. (1959), "The Effects of Emotion on Cue Utilisation and the Organisation of Behaviour," Psychological Review, 66 (May), 183-201.

Edell, J.A. and M.C. Burke (1987), "The Power of Feelings in Understanding Advertising Effects," Journal of Consumer Research, 14 (December), 421-433.

Elliot, S. (1992), "Presenting the 15 Best Ideas for Ads that work better," New York Times, 9 January, 17.

 O              P     !  " # $% &  '    (+,-( Embretson, S.E. and S.P. Reise (2000), Item Response Theory for Psychologists. Mahwah, NJ: Lawrence Erlbaum.

ERMA.org (2006), "Entertainment Resources & Marketing Association " Available: http://www.erma.org/web/index.php, (accessed 13 November 2006)

Ewing, M.T., T. Salzberger, and R.R. Sinkovics (2005), "An Alternative Approach to Assessing Cross-Cultural Measurement Equivalence in Advertising Research," Journal of Advertising, 34 (1), 17-36.

Eysenck, M. (1982), Attention and Arousal, Cognition and Performance. New York: Springer-Verlag.

Feldman, R.S. (1999), Understanding Psychology (5th ed.). MA: McGraw Hill College.

Fischer, G. (1995), "Derivations of the Rasch Model," in Rasch Models - Foundations, Recent Developments and Applications, Gerhard H. Fischer and Ivo W. Molenaar, Ed. New York: Springer. 15-38

Fischer, K., P. R. Shaver, and P. Carnochan (1990), "How emotions develop and how they organize development," Cognition and Emotion, 4 (2), 81-127.

Friestad, M. and P. Wright (1994), "The persuasion knowledge model: How people cope with persuasion attempts," Journal of Consumer Research, 21 (June), 62-74.

Gabriel, Y. and T. Lang (1995), The Unmanageable Consumer. Thousand Oaks, CA: Sage.

Galician, M. (2004), "Product Placement in the 21st Century," Journal of Promotion Management, 10 (1/2), 241-258.

Galician, M. and P.G. Bourdeau (2004), "The Evolution of Product Placements in Hollywood Cinema: Embedding High Involvement "Heroic" Brand Images," Journal of Promotion Management, 10 (1/2), 15-36.

Ganglmair, A. and R. Lawson (2003), "Measuring affective response to consumption using Rasch Modelling," Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behaviour, 16, 198-210.

Gardner, M.P. (1985), "Mood States and Consumer Behaviour: A critical review," Journal of Consumer Research, 12 (3), 281-300.

Gough, P.J. (2004), "Consumers respond favourably to product placement of brands in TV, Movies," in Media Post. 22 April ed.

 O              P     !  " # $% &  '    (+,.( Gould, S.J. and P.B. Gupta (2006), ""Come on Down": How consumers view game shows and the products placed in them," Journal of Advertising, 35 (1), 65-81.

Gould, S.J., P.B. Gupta, and S Grabner-Krauter (2000), "Product Placements in Movies: A Cross-Cultural Analysis of Austrian, French and American Consumers' Attitudes toward this emerging international promotional medium," Journal of Advertising, 29 (4), 41-67.

Green, K.E. and C.G. Frantom (2002), "Survey development and validation with the Rasch Model," in International Conference on Questionnaire Development, Evaluation and Testing. Charleston, SC.

Green, M.C. and T.C. Brock (2000), "The Role of Transportation in the Persuasiveness of Public Narratives," Journal of Personality and Social Psychology, 79 (5), 701-721.

Green, M.C., T.C. Brock, and G.F. Kaufman (2004), "Understanding Media Enjoyment: The Role of Transportation into Narrative Worlds," Communication Theory, 14 (4), 311- 327.

Greenwald, A.G. and C. Leavitt (1984), "Audience Involvement in Advertising: Four Levels," Journal of Consumer Research, 11 (1), 581-592.

Gunter, B., A. Furnham, and C. Beeson (1997), "Recall of Television Advertisements as a Function of Program Evaluation," Journal of Psychology, 131 (5), 541-553.

Gupta, P.B., S.K. Balasubramanian, and M.L. Klassen (2000), "Viewers' Evaluations of Product Placements in Movies: Public Policy Issues and Managerial Implications," Journal of Current Issues and Research in Advertising, 22 (2 ), 41-52.

Gupta, P.B. and S.J. Gould (1997), "Consumers' Perceptions of the Ethics and Acceptability of Product Placements in Movies: Product Category and Individual Differences," Journal of Current Issues and Research in Advertising, 19 (1 ), 37-50.

---- (2007), "Recall of Products Placed as Prizes versus Commercials in Game Shows," Journal of Current Issues and Research in Advertising, 29 (1), 43-53.

Gupta, P.B. and K.R. Lord (1998), "Product Placement in Movies: The Effect of Prominence and Mode on Audience Recall," Journal of Current Issues and Research in Advertising, 20 (1), 47-59.

Guttman, L. (1950), "The basis for scalogram analysis " in Measurement and Prediction, L. Guttman S. A. Stauffer, E.A. Suchman, P.F. Lazarsfeld, S.A. Star, J.A. Clausen, Ed. Vol. 4. Princeton, NJ: Princeton University Press.

 O              P     !  " # $% &  '    (+,/( Hackley, C. and R. Tiwsakul (2006), "Entertainment Marketing and Experiential Consumption," Journal of Marketing Communications, 12 (1), 63-75.

Hall, E. (2004), "Young consumers receptive to movie product placement," in Advertising Age. 29 March ed. Vol. 75.

Hirschman, E.C. and M.B. Holbrook (1982), "Hedonic Consumption: Emerging concepts, methods and propositions," Journal of Marketing, 46 (3), 92-101.

Holbrook, M.B. and R. Batra (1987), "Assessing the role of emotions as mediators of consumer responses to advertising," Journal of Consumer Research, 14 (3), 404-420.

Holbrook, M.B. and M.W. Grayson (1986), "The Semiology of Cinematic Consumption: Symbolic Consumer Behaviour in Out of Africa," Journal of Consumer Research, 13 (December), 374-381.

Holbrook, M.B. and E.C. Hirschman (1982), "The Experiential Aspects of Consumption: Consumer Fantasies, Feelings and Fun," Journal of Consumer Research, 9 (2), 132-140.

Holt, D.B. (1995), "How consumers consume: A typology of consumption practices," Journal of Consumer Research, 22 (June), 1-16.

Hudson, S. and D. Hudson (2006), "Branded Entertainment: A new advertising technique or product placement in disguise?," Journal of Marketing Management, 22, 489-504.

Isen, A.M. (1984), "The Influence of Positive Affect on Decision-Making and Cognitive Organisation," Advances in Consumer Research, 534-537.

Jacoby, J. (1978), "Consumer Research: A State of the Art Review," Journal of Marketing, 42 (April), 87-96.

Karrh, J.A. (1994), "Effects of Brand Placement in Motion Pictures," in Proceedings of the 1994 American Academy of Advertising Conference, Karen W King (Ed.). Athens, GA: American Academy of Advertising.

---- (1995), "Brand Placements in Feature Films: The Practitioners' View," in 1995 American Academy of Advertising Conference. Waco TX.

---- (1998), "Brand Placement: A Review," Journal of Current Issues and Research in Advertising, 20 (2), 31-48.

Karrh, J.A., K.T. Frith, and C. Callison (2001), "Audience attitudes towards brand (product) placement: Singapore and the United States," International Journal of Advertising, 20 (1), 3-24.

 O              P     !  " # $% &  '    (+,0(

Karrh, J.A., K.B. McKee, and C.J. Pardun (2003), "Practitioners' Evolving Views on Product Placement Effectiveness," Journal of Advertising Research (June), 138-149.

Kim, J. and A.M. Rubin (1997), "The Variable Influence of Audience Activity on Media Effects," Communication Research, 24 (2), 107-135.

King, M.F. and G.C. Bruner (2000), "Social Desirability Bias: A Neglected Aspect of Validity Testing," Psychology & Marketing, 17 (2), 79-103.

King, M.M. and K.D. Moulton (1996), "The effects of television role models on the career aspirations of African-American junior high school students," Journal of Career Development, 23 (2), 111-125.

Kolah, A. (2006), "A step closer to brandcasting," Brand Republic (7 November 2006).

Kolbe, R.H. and M.S. Burnett (1991), "Content Analysis Research: An Examination of the Applications with Directives for Improving Research Reliability and Objectivity," Journal of Consumer Research, 18 (2), 243-250.

Kretchmer, S.B. (2004), "Advertainment: The Evolution of Product Placement as a Mass Media Marketing Strategy," Journal of Promotion Management, 10 (1/2), 37-54.

Krider, R.E. (2006), "Research Opportunities at the Movies," Marketing Science, 25 (6), 662-664.

Krishnan, H.S. and D. Chakravarti (1999), "Memory Measures for Pre-testing Advertisements: An Integrative Conceptual Framework and a Diagnostic Template," Journal of Consumer Psychology, 8 (1), 1-38.

Krugman, H.E. (1967), "The Measurement of Advertising Involvement," Public Opinion Quarterly, 30.

La Ferle, C. and S.M. Edwards (2006), "Product Placement: How brands appear on television," Journal of Advertising, 35 (4), 65-86.

La Pastina, A.C. (2001), "Product placement in Brazilian prime-time television: The case of the reception of a telenovela," Journal of Broadcasting and Electronic Media, 45 (4), 541- 549.

Lavidge, R.J. and G.A. Steiner (1961), "A Model of Predictive Measurements of Advertising Effectiveness," Journal of Marketing, 25 (6), 59-62.

 O              P     !  " # $% &  '    (+,1( Law, S. and K.A. Braun-LaTour (2004), "Product Placements: How to Measure their Impact," in The Psychology of Entertainment Media: Blurring the Lines between Entertainment and Persuasion, L.J Shrum, Ed. New Jersey:: Lawrence Erlbaum Associates. 63-78

Law, S. and K. Braun (2000), "I'll have what she's having: Gauging the impact of product placements on viewers," Psychology & Marketing, 17 (12), 1059-1075.

Lee, A.Y. (2002), "Effects of implicit memory on memory-based versus stimulus-based brand choice," Journal of Marketing Research, 39 (4), 440-454.

Linacre, J.M. (1998), "Rasch first or Factor first?," Rasch Measurement Transactions, 11 (4), 603.

Livingstone, S.M. (1990), "Interpreting a Television Narrative: How Different Viewers see a Story," Journal of Communication, 40 (1), 72-85.

Lloyd, D.W. and K.J. Clancy (1991), "CPMs versus CPMIs: Implications for Media Planning," Journal of Advertising Research, August/September, 34-44.

Lockwood, P. and Z. Kunda (1997), "Superstars and me: Predicting the impact of role models on the self," Journal of Personality and Social Psychology, 73 (1), 91-103.

Lombard, M. and T. Ditton (1997), "At the Heart of it All: The Concept of Presence," Journal of Computer-Mediated Communication, 3 (2), http://www.ascusc.org/jcmc/vol3/issue2/lombard.html.

Lynch, B. and R. Bonnie (1994), "Growing up tobacco free - preventing nicotine addiction in children and youths: a report of the Institute of Medicine," National Academy Press, Washington DC.

Mackie, D.E. and A.G. Asuncion (1990), "Online and Memory-based Modification of Attitudes: Determinants of Message Recall-Attitude Change Correspondence," Journal of Personality and Social Psychology, 59 (1), 5-16.

Macklem, K (2002), "Ready for your close-up, Pepsi," in Maclean's.

Martin, C.A. and A.J. Bush (2000), "Do role models influence teenagers; purchase intentions and behaviour?," Journal of Consumer Marketing, 17 (5), 441-453.

Mathwick, C., N. Malhotra, and E. Rigdon (2001), "Experiential value: conceptualisation, measurement and application in the catalogue and Internet shopping environment," Journal of Retailing, 77, 39-56.

 O              P     !  " # $% &  '    (+,2( McCarty, J.A. (2004), "Product Placement: The Nature of the Practice and Potential Avenues of Inquiry," in The Psychology of Entertainment Media: Blurring the Lines between Entertainment and Persuasion, L.J Shrum, Ed. New Jersey: Lawrence Erlbaum Associates. 45-62

McInnes, C., P. Griffin, R. James, and H. Coates (2001), "Development of the Course Experience Questionnaire (CEQ)." Melbourne: Faculty of Education, University of Melbourne.

McIntyre, P. (2004), "Young fall for products that top guns use in films," in Sydney Morning Herald. 20 May ed. Sydney.

McKechnie, S.A. and J. Zhou (2003), "Product placement in movies: A comparison of Chinese and American consumers' attitudes," International Journal of Advertising, 22 (3), 349-374.

Mehrabian, A. and J. Russell (1974), An Approach to Environmental Psychology. Cambridge, MA: MIT Press.

Messick, S. (1995), "Validity of Psychological Assessment: Validation of Inferences From Persons' Responses and Performances as Scientific Inquiry Into Score Meaning," American Psychologist, 50 (9), 74-149.

Michell, J. (1999), Measurement in Psychology: Critical History of a Methodological Concept. Cambridge: Cambridge University Press.

Molenaar, I.W. (1995), "Some background for Item Response Theory and the Rasch Model," in Rasch Models - Foundations, Recent Developments and Applications, Gerhard H. Fischer and Ivo W. Molenaar, Ed. New York: Springer. 3-14

Moorman, M. (2003), "Context considered: The relationship between media environments and advertising effects," PhD Thesis, University of Amsterdam.

Morton, C.R. and M. Friedman (2002), ""I Saw it in the Movies": Exploring the link between product placement beliefs and reported usage behaviour," Journal of Current Issues and Research in Advertising, 24 (2), 33-40.

Moschis, G.P. (1978), "Teenagers' Responses to Retailing Stimuli," Journal of Retailing, 54 (4), 80-93.

Moschis, G.P. and G.A. Churchill (1978), "Consumer Socialisation: A Theoretical and Empirical Analysis," Journal of Marketing Research, 15 (4), 599-609.

Murdock, G (1992), "Branded Images," in Sight and Sound Vol. 2.

 O              P     !  " # $% &  '    (+-3(

Murry, J.P., J.L. Lastovicka, and S.N. Singh (1992), "Feeling and Liking Responses to Television Programs: An Examination of Two Explanations for Media-Context Effects," Journal of Consumer Research, 18 (March), 441-451.

Nachmias, D. and C. Nachmias (1976), Research Methods in the Social Sciences. London: St Martin's Press.

Nebenzahl, I.D. and E Secunda (1993), "Consumer's Attitudes Toward Product Placement in Movies," International Journal of Advertising, 12 (1), 1-11.

Nelson, M.R. (2002), "Recall of Brand Placements in Computer / Video Games," Journal of Advertising Research (March/April), 80-92.

Nelson, M.R. and N. Devanathan (2006), "Brand placements Bollywood style," Journal of Consumer Behaviour, 5 (May-June), 211-221.

Nelson, M.R. and L.E. McLeod (2005), "Adolescent brand consciousness and product placements: Awareness, liking and perceived effects on self and others," International Journal of Consumer Studies, 29 (6), 515-528.

Nelson, M.R., R.A. Yaros, and H. Keum (2006), "Examining the influence of telepresence on spectator and player processing of real and fictitious brands in a computer game," Journal of Advertising, 35 (4), 87-99.

Newell, J., C.T. Salmon, and S. Chang (2006), "The Hidden History of Product Placement," Journal of Broadcasting and Electronic Media, 50 (4), 575-594.

Nielsen Media Research (2007), "Nielsen Media Research - Place*Views" Available: http://www.nielsenmedia.com/, (accessed 17 September 2007)

Norris, C.E. and A.M. Colman (1993), "Context Effects on Memory for Television Advertisements," Social Behaviour and Personality, 21 (4), 279-296.

Nunnally, J.C. (1967), Psychometric Theory (1st ed.). New York: McGraw Hill.

---- (1978), Psychometric Theory (2nd ed.). New York: McGraw-Hill.

Obermiller, C. and E. Spangenberg (1998), "Development of a Scale to Measure Skepticism Toward Advertising," Journal of Consumer Psychology, 7 (2), 159-186.

Ohanian, R. (1990), "Construction and Validation of a Scale to Measure Celebrity Endorser's Perceived Expertise, Trustworthiness and Attractiveness," Journal of Advertising, 19 (3), 39-52.

 O              P     !  " # $% &  '    (+-*(

Ong, B.S. (2004), "A Comparison of Product Placements in Movies and Television Programs: An Online Research Study," Journal of Promotion Management, 10 (1/2), 147- 158.

Ong, B.S. and D Meri (1994), "Should Product Placement in Movies be Banned?," Journal of Promotion Management, 2 (3/4), 159-175.

Palda, K.S. (1966), "The Hypothesis of a Hierarchy of Effects: A Partial Evaluation," Journal of Marketing Research, 3 (February), 13-24.

Pallant, J.F., R.L. Miller, and A. Tennant (2006), "Evaluation of the Edinburgh Post Natal Depression Scale using Rasch Analysis," BMC Psychiatry, 6 (28), 1-10.

Panda, T.K. (2004), "Consumer response to brand placements in films: Role of brand congruity and modality of presentation in bringing attitudinal change among consumers with special reference to brand placements in Hindi films," South Asian Journal of Management, 11 (4), 7-25.

Pardun, C.J. and K.B. McKee (1996), "What Advertising Agency Media Directors Have to Say About Placing Clients' Products in Motion Pictures," in 1996 Association for Education in Journalism and Mass Communication Conference. Anaheim, CA.

---- (1999), "Product placements as public relations: An exploratory study of the role of the public relations firm," Public Relations Review, 25 (4), 481-493.

Pavelchak, M.A., J.H. Antil, and J.M. Munch (1988), "The Super Bowl: An Investigation into the Relationship Among Program Context, Emotional Experience, and Ad Recall," Journal of Consumer Research, 15 (3), 360-367.

Percy, L. (2006), "Comments: Are product placements effective," International Journal of Advertising, 25 (1), 112-114.

Perse, E.M. (1990), "Audience Selectivity and Involvement in the Newer Media Environment," Communication Research, 17, 675-697.

---- (1998), "Implications of cognitive and affective involvement for channel changing," Journal of Communication, 48 (Summer), 49-68.

Pervan, S.J. and B.A.S. Martin (2006), "Soap operas in New Zealand and the US: Product placement strategy and consumption imagery," Advances in Consumer Research, 33, 135.

Peter, J.P. (1979), "Reliability: A Review of Psychometric Basics and Recent Marketing Practices," Journal of Marketing Research, 16 (February), 6-17.

 O              P     !  " # $% &  '    (+-+(

Petty, R.E., J.T. Cacioppo, and D. Schumann (1983), "Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of Involvement," Journal of Consumer Research, 10 (September), 134-148.

Pham, M.T. (1992), "Effects of Involvement, Arousal and Pleasure in the Recognition of Sponsorship Stimuli," in Advances in Consumer Research Vol. 19. Provo, UT: Association for Consumer Research.

Poels, K. and S. Dewitte (2006), "How to Capture the Heart? Reviewing 20 Years of Emotion Measurement in Advertising," Journal of Advertising Research, 46 (March), 18- 37.

PQ Media (2006), "PQ Media Global Product Placement Forecast 2006." Stamford, Connecticut.

---- (2007), "PQ Media Global Product Placement Forecast Series 2006-2010: Country-by- country analysis." Stamford, Connecticut.

Prieto, L., H. Thorsen, and K. Juul (2005), "Development and validation of a quality of life questionnaire for patients with colostomy or ileostomy," Health and Quality of Life Outcomes, 3 (62), 1-10.

Rao, A. and K.B. Monroe (1988), "The Moderating Effect of Prior Product Knowledge on Cue Utilisation in Product Evaluations," Journal of Consumer Research, 15 (September), 253-264.

Rasch, G. (1960), Probabilistic Models for Some Intelligence and Attainment Tests. Copenhagen: Danish Institute for Educational Research.

Reding, V. (2001), "Television without frontiers: Amending the directive," Intermedia, 29 (4), 4-9.

Redondo, I. (2006), "Product Placement Planning: How is the industry placing brands in relation to moviegoer consumption?," Journal of International Consumer Marketing, 18 (4), 33-60.

Reed, J.D. (1989), "Plugging away in Hollywood," in Time. 2 January ed.

Richins, M.L. (1997), "Measuring emotions in the consumption experience," Journal of Consumer Research, 24 (2), 127-146.

Richins, M.L. and S. Dawson (1992), "Materialism as a Consumer Value: Measure Development and Validation," Journal of Consumer Research, 19 (4), 303-316.

 O              P     !  " # $% &  '    (+-,(

Ritossa, D.A. and N. Rickard (2004), "The relative utility of 'pleasantness and liking' dimensions in predicting the emotions expressed in music," Society for Education; Music and Psychology Research, 32 (1), 5-22.

Roedder, D.L. (1981), "Age Differences in Children's Responses to Television Advertising: An Information-Processing Approach," Journal of Consumer Research, 8 (September), 144-153.

Roehm, M.L., H.A. Roehm, and D.S. Boone (2004), "Plugs versus Placements: A comparison of alternatives for within-program brand exposure," Psychology & Marketing, 21 (1), 17-28.

Roskos-Ewoldsen, D.R. and R.H. Fazio (1992), "On the orienting value of attitudes: Attitude accessibility as a determinant of an object's attraction of visual attention," Journal of Personality and Social Psychology, 63 (2), 198-211.

Rossiter, J.R. (2002), "The C-OAR-SE procedure for scale development in marketing," International Journal of Research in Marketing, 19, 305-335.

Rossiter, J.R. and L. Percy (1998), Advertising and Promotions Management (2nd ed.). USA: McGraw Hill.

Rössler, P. and J. Bacher (2002), "Transcultural Effects of Product Placement in Movies," Zeitschrift für Medienpsychologie, 14 (3), 98-108.

Rubin, A.M., E.M. Perse, and D.S. Taylor (1988), "A methodological investigation of cultivation," Communication Research, 15, 107-134.

Russell, C.A. (1998), "Toward a Framework of Product Placement: Theoretical Propositions," in Advances in Consumer Research, J.W. Alba and J.W Hutchinson (Eds.) Vol. 25. Provo, UT.

---- (2002), "Investigating the effects of product placements in television shows: The role of modality and plot connection congruence on brand memory and attitude," Journal of Consumer Research, 29 (December), 306-318.

Russell, C.A. and M. Belch (2005), "A Managerial Investigation into the Product Placement Industry," Journal of Advertising Research, March, 73-92.

Russell, C.A., A.T. Norman, and S.E. Heckler (2004), "The Consumption of Television Programming: Development and Validation of the Connectedness Scale," Journal of Consumer Research, 31 (June), 150-161.

 O              P     !  " # $% &  '    (+--( Russell, C.A. and C.P. Puto (1999), "Rethinking Television Audience Measures: An Exploration into the construct of Audience Connectedness," Marketing Letters, 10 (4), 393- 407.

Russell, C.A. and B.B. Stern (2006), "Consumers, Characters and Products: A balance model of sitcom product placement effects," Journal of Advertising, 35 (1), 7-21.

Russell, J.A., A. Weiss, and G.A. Mendelson (1989), "Affect Grid: A Single-item Scale of Pleasure and Arousal," Journal of Personality and Social Psychology, 57 (3), 493-502.

Sabherwal, S., J. Pokrywczynsji, and R. Griffin (1994), "Brand Recall for Product Placements in Motion Pictures: A Memory-based Perspective," in Association for Education in Journalism and Mass Communications. Atlanta, Georgia.

Salzberger, T. (1999), "How the Rasch Model May Shift Our Perspective of Measurement in Marketing Research," in Proceedings of the 1999 Australia and New Zealand Marketing Academy Conference (ANZMAC). Sydney, Australia.

---- (2000), "An alternative way of establishing measurement in marketing research: its implications for scale development and validity," in Proceedings of the 2000 Australia and New Zealand Marketing Academy Conference (ANZMAC). Gold Coast, Australia.

---- (2004), "Reconsidering the Paradigm of Measurement in Marketing Research: Critically Reviewing Recent Contributions Challenging Churchill’s Paradigm," in Proceedings of the 2004 Australia and New Zealand Marketing Academy Conference (ANZMAC). Wellington, New Zealand.

---- (2007), "Scientific Measurement of Latent Variables in Marketing Research: An Alternative Framework," Habilitationsschrift, University of Economics and Business Administration

Salzberger, T., D. Andrich, and G. Soutar (2001), "The Measurement of Latent Constructs in Marketing Research: New Perspectives," in Proceedings of the 2001 Australia and New Zealand Marketing Academy Conference (ANZMAC). Auckland, New Zealand.

Samuel, L.R. (2004), "Advertising Disguised as Entertainment," Television Quarterly, 34 (2), 51-55.

Sanbonmatsu, D.M. and F.R. Kardes (1988), "The Effects of Physiological Arousal on Information Processing and Persuasion," Journal of Consumer Research, 15 (December), 379-385.

 O              P     !  " # $% &  '    (+-.( Sapolsky, B.S. and L. Kinney (1994), "You Oughtta Be in Pictures: Product Placements in the Top Grossing Films of 1991," in Proceedings of the 1994 American Academy of Advertising Conference. Tucson, Arizona.

Sargent, J.D., J.L. Tickle, M.L. Beach, M.A. Dalton, M.B. Ahrens, and T.F. Heatherton (2001), "Brand appearances in contemporary cinema films and contribution to global marketing of cigarettes," Lancet, 357, 29-32.

Sauer, A. (2004), "Brandsploitation: A New Genre in Film" Available: http://www.brandchannel.com/start.asp?fa_id=231, (accessed 28 September 2004)

Schacter, D.L. (1987), "Implicit Memory: History and Current Status," Journal of Experimental Psychology: Learning, Memory and Cognition, 13 (July), 501-518.

Schiffman, L., D. Bednall, J. Watson, and L. Kanuk (1997), Consumer Behaviour. Sydney: Prentice Hall.

Schmoll, N.M., J. Hafer, M. Hilt, and H. Reilly (2006), "Baby boomers' attitudes towards product placement," Journal of Current Issues and Research in Advertising, 28 (2), 33-53.

Schneider, L. and T. B. Cornwell (2005), "Cashing in on crashes via brand placement in computer games," International Journal of Advertising, 24 (3), 321-343.

Scott, J. (2002), "Product Placement: Teenage Audiences and Recall," unpublished Honours Thesis, University of New South Wales.

Scott, J. and M. Craig-Lees (2003), "Audience Characteristics and Product Placement Effects," in Proceedings of the 2003 Australia and New Zealand Marketing Academy Conference (ANZMAC). Adelaide, Australia.

---- (2004), "Optimising Success: Product Placement Quality and its Effects on Recall," in Proceedings of the 2004 Australia and New Zealand Marketing Academy Conference (ANZMAC). Wellington, New Zealand.

---- (2005a), "Measuring Media Audiences: The Need for an Audience Engrossment Scale," in Proceedings of the 2005 Australia and New Zealand Marketing Academy Conference (ANZMAC). Perth, Western Australia.

---- (2005b), "Product Placement: Developing Concepts, Constructs and Measures," in Proceedings of the 2005 North American Conference for the Association of Consumer Research. Texas, USA.

---- (2006), "Conceptualisation, Consumer and Cognition: The 3 Cs that will advance product placement research," Asia Pacific Advances in Consumer Research, 7, 364-370.

 O              P     !  " # $% &  '    (+-/(

Scott, J., J. Harris, and M. Craig-Lees (2007), "Refining Audience Engrossment: A Comparison of the Usefulness of Rasch Modelling and Factor Analysis in Scale Development," in Proceedings of the 2007 Australia and New Zealand Marketing Academy Conference (ANZMAC). Dunedin, New Zealand.

Scott, J. and T. Salzberger (2008), "Investigating the Threshold Ordering of the Audience Engrossment Scale Using the Polytomous Rasch Model," in 3rd International Rasch Conference. Perth, Australia.

Shapiro, S. and H.S. Krishnan (2001), "Memory-based measures for assessing advertising effects: A comparison of explicit and implicit memory effects," Journal of Advertising, 30 (3), 1-13.

Shernoff, D.J., M. Csikszentmihalyi, B. Schneider, and E. Steele Shernoff (2003), "Student Engagement in High School Classrooms from the Perspective of Flow Theory," School Psychology Quarterly, 18 (2), 158-176.

Singh, J. (2004), "Tackling problems with item response theory: Principles, characteristics and assessment, with an illustrative example," Journal of Business Research, 57, 184-208.

Singh, S.N. and G.A. Churchill (1987), "Arousal and Advertising Effectiveness," Journal of Advertising, 16 (1), 4-11.

Singh, S.N. and M.L. Rothschild (1983), "Recognition as a Measure of Learning from Television Commercials," Journal of Marketing Research, 20 (August), 235-248.

Singh, S.N., M.L. Rothschild, and G.A. Churchill (1988), "Recognition versus Recall as Measures of Television Commercial Forgetting," Journal of Marketing Research, 25 (February), 72-80.

Solomon, M.R. and B.G Englis (1994), "Reality Engineering: Blurring the boundaries between commercial signification and popular culture," Journal of Current Issues and Research in Advertising, 16 (2), 1-17.

Spillman, S. (1989), "Marketers race to leave their brand on films," Advertising Age, 56 (1 July), 55.

Stafford, T.F., M.R. Stafford, and L.L. Schkade (2004), "Determining Uses and Gratifications for the Internet," Decision Sciences, 35 (2), 259-288.

Steortz, E.M. (1987), "The Cost Efficiency and Communication Effects Associated with Brand Name Exposure within Motion Pictures," unpublished Masters Thesis, West Virginia University.

 O              P     !  " # $% &  '    (+-0(

Stewart, D.W., J. Morris, and A. Grover (2007), "Emotions in Advertising," in Handbook of Advertising, G.J. Tellis and T. Ambler, Ed. London, UK: Sage Publications.

Storm, C. and T. Storm (1987), "A Taxonomic Study of the Vocabulary of Emotions," Journal of Personality and Social Psychology, 53 (4), 805-816.

Stout, P.A. and J.D. Leckenby (1986), "Measuring emotional response to advertising," Journal of Advertising 15 (4), 35-42.

Strasburger, V.C. (1995), Adolescents and the Media: Medical and Psychological Impact. California: Sage.

Sweney, M. (2006), "Product placement gets European approval," in The Guardian. 14 November ed. London.

Taffel, J (2004), "Ad Nausea," in Sydney Morning Herald. 31 May ed. Sydney.

Tavassoli, N.T., C.J. Shultz, and G.J. Fitzsimons (1995), "Program Involvement: Are Moderate Levels Best for Ad Memory and Attitude Towards the Ad?," Journal of Advertising Research (Sept-Oct), 61-72.

Tennant, A., S.P. McKenna, and P. Hagell (2004), "Application of Rasch Analysis in the Development and Application of Quality of Life Instruments," Value in Health, 7 (1), S22- S26.

Tickle, J.D., J.L. Sargent, M.A. Dalton, M.L. Beach, and T.F. Heatherton (2001), "Favourite movie stars, their tobacco use in contemporary movies, and its association with adolescent smoking," Tobacco Control, 10 (1), 16-29.

Troup, M. L. (1991), "The Captive Audience: A Content Analysis of Product Placements in Motion Pictures," unpublished Masters Thesis, University of Texas, Austin.

Turcotte, S. (1995), "Gimme a Bud! The Feature Film Product Placement Industry," unpublished Masters Thesis, University of Texas. van Reijmersdal, E.A., P.C. Neijens, and E.G. Smit (2007), "Effects of television brand placement on brand image," Psychology & Marketing, 24 (5), 403-420.

Vollmers, S. and R. Miserski (1994), "A Review and Investigation into the Effectiveness of Product Placement in Films," in 1994 Conference of the American Academy of Advertising. Georgia.

 O              P     !  " # $% &  '    (+-1( Wakefield, M., B. Flay, M. Nichter, and G. Giovino (1998), "Role of the media in influencing trajectories of youth smoking," Addiction, Suppl 1, 79-103.

Wallace, W.P. (1965), "Review of the Historical, Empirical and Theoretical Status of the Von Restorff Phenomenon," Psychological Bulletin, 63 (6), 410-424.

Watson, D. and A. Tellegan (1985), "Toward a Consensual Structure of Mood," Psychological Bulletin, 98 (September), 219-235.

Waugh, R.F. (2001), "Creating a scale to measure motivation to achieve academically: Linking attitudes and behaviours using Rasch measurement," in Australian Association for Research in Education. Freemantle.

Weilbacher, W.M. (2002), "Point of View: Does advertising cause a hierarchy of effects?," Journal of Advertising Research, 41 (6), 19-26.

Wenner, L.A. (2004), "On the Ethics of Product Placement in Media Entertainment," Journal of Promotion Management, 10 (1/2), 101-132.

Winkielman, P., K.C. Berridge, and J.L. Wilbarger (2005), "Unconscious affective reactions to masked happy versus angry faces influence consumption behaviour and judgements of value," Personality and Social Psychology Bulletin 31 (1), 121-135.

Wolfe, E.W. and E.V. Smith (2007), "Instrument Development Tools and Activities for Measure Validation: Part II - Validation Activities," Journal of Applied Measurement, 8 (2), 204-234.

Worth, L.T. and D.M. Mackie (1987), "Cognitive Mediation of Positive Affect in Persuasion," Social Cognition, 5 (1), 76-94.

Wright, B.D. (1996), "Comparing Rasch Measurement and Factor Analysis," Structural Equation Modelling, 3 (1), 3-24.

Wright, B.D. and G. Masters (1981), "The Measurement of Knowledge and Attitude," in Research Memorandum (number 30). Chicago: Statistical Laboratory, Department of Education, University of Chicago.

Zaichkowsky, J.L. (1985), "Measuring the Involvement Construct," Journal of Consumer Research, 12 (December), 341-352.

---- (1994), "The Personal Involvement Inventory: Reduction, Revision and Application to Advertising," Journal of Advertising, 15 (1), 38-46.

Zaltmann, G. (2003), How Customers Think. Boston, MA: Harvard Business School Press.

 O              P     !  " # $% &  '    (+-2( LIST OF APPENDICES

APPENDIX 2.1: EXTENDED PRODUCT PLACEMENT LITERATURE REVIEW...... 250 APPENDIX 4.1: SECTIONS 5 AND 6 OF AUDIENCE ENGROSSMENT SURVEY ...... 269 APPENDIX 4.2: OVERVIEW OF KEY STATISTICS AND INDICATORS...... 271 APPENDIX 5.1: STAGE 2 - ALL FEELING ITEMS ...... 279 APPENDIX 5.2: STAGE 2 – ALL FEELING ITEMS WITH COLLAPSED THRESHOLDS...... 282 APPENDIX 5.3: STAGE 2 - ALL AROUSAL ITEMS...... 285 APPENDIX 5.4: STAGE 2 - ALL AROUSAL ITEMS WITH COLLAPSED THRESHOLDS...... 288 APPENDIX 5.5: STAGE 2 – ALL APPRAISAL ITEMS...... 291 APPENDIX 5.6: STAGE 2 – APPRAISAL ITEMS WITHOUT ACTOR OR GENRE ITEMS...... 294 APPENDIX 5.7: STAGE 2 – APPRAISAL ITEMS WITHOUT ACTOR AND GENRE ITEMS AND MIND3...... 297 APPENDIX 5.8: STAGE 2 – ALL COGNITIVE EFFORT ITEMS ...... 300 APPENDIX 5.9: STAGE 2 – TWO COGNITIVE EFFORT DIMENSIONS...... 303 APPENDIX 5.10: STAGE 3 – ALL FEELING ITEMS...... 308 APPENDIX 5.11: STAGE 3 – FEELINGS DIMENSION WITHOUT APPALLED AND CROSS...... 313 APPENDIX 5.12: STAGE 3 - ALL AROUSAL ITEMS...... 318 APPENDIX 5.13: STAGE 3 – ALL AROUSAL ITEMS (RECODED)...... 323 APPENDIX 5.14: STAGE 3 – ALL APPRAISAL ITEMS...... 328 APPENDIX 5.15: STAGE 3 - ALL APPRAISAL ITEMS WITH COLLAPSED THRESHOLDS...... 333 APPENDIX 5.16: STAGE 3 – ALL COGNITIVE EFFORT ITEMS ...... 338 APPENDIX 5.17: STAGE 3 - ALL COGNITIVE EFFORT ITEMS WITH COLLAPSED THRESHOLDS ...... 343 APPENDIX 5.18: STAGE 3 – “DIFFICULT” COGNITIVE EFFORT ITEMS ONLY ...... 348 APPENDIX 5.19: FINAL FEELINGS DIMENSION – ALL MOVIES...... 353 APPENDIX 5.20: FINAL AROUSAL DIMENSION – ALL MOVIES ...... 375 APPENDIX 5.21: FINAL APPRAISAL DIMENSION – ALL MOVIES ...... 397 APPENDIX 5.22: STAGE 4 – ALL COGNITIVE EFFORT ITEMS ...... 419 APPENDIX 5.23: FINAL COGNITIVE EFFORT DIMENSION (THE ISLAND ONLY) ...... 421 APPENDIX 5.24: HOLISTIC PLOTS ...... 425 APPENDIX 5.25: BEST CASE SCENARIO PLOTS ...... 428 APPENDIX 5.26: WORST CASE SCENARIO PLOTS ...... 430 APPENDIX 6.1: LETTERS TO SCHOOLS, PARENTS AND PARTICIPANTS...... 432 APPENDIX 6.2: SURVEY TO MEASURE FILM AND STAR PREFERENCES ...... 438 APPENDIX 6.3: DETAILED CONTENT ANALYSIS PROCEDURE ...... 442

 O              P     !  " # $% &  '  ()( APPENDIX 6.4: INSTRUCTIONS FOR CONTENT ANALYSIS AND FILM CODING FORM ...... 446 APPENDIX 6.5: RESULTS FROM CONTENT ANALYSIS...... 451 APPENDIX 6.6: CONTENT ANALYSIS OF ‘THE ISLAND’...... 457 APPENDIX 6.7: BRAND FAMILIARITY QUESTIONNAIRE ...... 461 APPENDIX 6.8: POST-FILM TESTING – AUDIENCE ENGROSSMENT, PRODUCT PLACEMENT RECOGNITION AND PROGRAM LIKING ...... 467 APPENDIX 6.9: PEARSON CORRELATIONS BETWEEN PRODUCT PLACEMENT CHARACTERISTICS...... 477

 O              P     !  " # $% &  '  ()( APPENDIX 2.1: EXTENDED PRODUCT PLACEMENT LITERATURE REVIEW

Past Research into Product Placement As evidenced, product placement is an important emerging area of marketing communications and is becoming increasingly important in practice. Indeed, as early as 1992, a report from a group of major advertisers included the recommendation to recognise product placement in movies as a new medium (Elliot 1992). However, whilst research on product placement has begun to appear in recent years, it has not kept pace with the growth of the practice. Similarly, the tools that have been used to evaluate its effectiveness and impact on audiences are still relatively unsophisticated, so there remains little knowledge about whether product placements are effective and how best to measure their impact.

A thorough review of the literature dealing with product placement as led to the conclusion that to date, there have been seven major dimensions of product placement research – • content analysis of films, television programs and music videos; • investigation into practitioner attitudes and practices regarding product placement; • research dealing with audience attitudes towards product placement (often with a focus on ethical issues); • analysis of the impact of product placement on brand awareness via recall and recognition tests (i.e. memory-based research); • examination of behavioural impacts such as purchase intention and changes in brand image and brand attitudes (both explicitly and implicitly); • cross-cultural studies often comparing how different cultures respond to one or more of these aforementioned dimensions; • broader exploratory qualitative research

Determining the prevalence of product placement Content analysis studies by researchers such as Diener (1993), Sapolsky and Kinney (1994), DuRant, Rome, Rich and Allred (1997) and Galician and Bourdeau (2004) seek to identify

 O              P     !  " # $% &  '  (*+,( the broad product categories and specific brands that have featured in entertainment vehicles such as films, television shows and music videos. A specific focus of many of these studies has been to investigate the frequency of the appearance of ethically-charged products such as cigarettes, alcohol and drugs. Such content analyses have highlighted that these products do appear frequently in entertainment programs and that their incidence is increasing (Sapolsky and Kinney 1994; DuRant et al. 1997; Tickle, Sargent, Dalton, Beach and Heatherton 2001). Indeed, Sargent et al (2001) found that of the top 25 US box office films from 1988 to 1997, 32% of those rated for adolescent audiences contained tobacco brand names.

In 1991, through examining the 25 top-grossing movies of 1989, Troup found that comedies contained the most product placement, with low involvement goods accounting for 68% of product placement in movies. Three years later, using a similar methodology, Sapolsky and Kinney (1994) found that the average number of brands in each movie had dropped from 18 in Troup’s (1991) study to 14 in theirs, and that low involvement brands now constituted 70% of all placements. A further 18% of placements involved automobiles (a high involvement product category).

Galician and Bourdeau (2004) took a more longitudinal approach to their content analysis, studying the fifteen top-grossing movies of 1977, 1987 and 1997 in order to track changes to the product placement practice. In doing so, they found that both the number of product placements appearing in films and the proportion of film time featuring product placements was fairly stable over this period. By breaking down specific product categories further than previous researchers, Galician and Bourdeau (2004) could identify that automobiles were consistently the most dominant type of product placement, followed by beer and then soft drink. They also found an increase in the range of product categories now appearing in films, indicating a growing interest among marketers to add product placement to their promotional mix. Correspondingly, marketers have become savvier about how they want their products to appear. Placed products are now more central to the plot than ever before,

 O              P     !  " # $% &  '  (*+-( are idealised in visual presentation, and are often endorsed by the stars that portray the major characters.

In a bid to determine whether product placement on television is less prevalent and more subtle than that in film, Avery and Ferraro (2000) performed a content analysis on 112 hours of prime-time television airing on four American networks during one week in 1997 (so as to capture both television programs and movie re-runs). A total of 2,945 brand appearances were identified, averaging fifteen brand appearances per half hour of programming. Breaking this data down further, 882 brand appearances occurred in scripted made-for-television programs and 274 appeared in commercial movie re-runs. However, a higher proportion of brand appearances in movie re-runs (34.3%) had the characteristic of visual prominence (i.e. extended portrayal, in the foreground, with close-up camera view) compared to those appearing in made-for-television programs (16.1%). Therefore, they found that the degree to which films had become peppered with prominent commercial brands still exceeded that found in television programs.

La Ferle and Edwards (2006) conducted the most recent large scale content analysis, in 2002, and examined placement techniques in a range of different television programs during one week of prime time American television on five major networks. In doing so, they also attempted to determine whether product placement was increasing, and how prevalent it was as opposed to ‘plugs’ (i.e. an on-camera verbal discussion of a brand delivered by a personality). They found that on average, a brand appeared within every three minutes of programming, with game shows showing the most brands, and storied programs showing the fewest and doing so less prominently. Of those brands shown, the highest number was for consumer products, but those for services appeared more prominently. In regards to the specific execution, brands tended to appear either visually or aurally, but not both, and were featured for less than five seconds. So whilst the frequency of brand exposures may be high, the quality of the placements is actually poor in that they are not featured for long or via a dual modality. In this way, the execution of this practice does not seem to have evolved from those results found by Avery and Ferraro (2000) and suggests that practitioners need to

 O              P     !  " # $% &  '  (*+*( take greater control of the process and have their brands appear for longer, be both spoken and seen, interact with the star, and/or become better integrated into the content.

It is worthy to note that with the exception of content analyses examining the prevalence of tobacco in films, no further content analyses have been conducted across a broad range of program genres more recently than 2002. In light of the recent increased spend on product placement, it would be an important contribution to determine whether any changes in frequency of product placement have occurred in the past decade. Is it still used mainly for low involvement products? What types of movies or television shows tend to have more product placements? How many product placements on average appear in films and television programs today? From a local perspective, it could also be useful to conduct some Australian-based content analysis studies, examining both Australian films and television programs to assess the level of product placement currently contained within them. With all content analysis studies imported (mainly from the United States), there is currently no true empirical study to describe the state of product placement activity in this country. The closest is a comparison of product placement in American and New Zealand soap operas, in which Pervan and Martin (2006) found that American soap operas tended to display more product placement related to leisure and appearance-related items, whereas New Zealand soaps included more products for transportation and food. Interestingly, there were higher levels of product placement in New Zealand soaps (70% of episodes) compared to US soaps (26% of all episodes).

Investigating practitioner attitudes to and practices of product placement An important step in the study of product placement is the examination of what practitioners believe to be the goals of product placement. Why is it being used as a method of communication in the first place? What do practitioners hope to gain from it?

Karrh (1995) conducted the first published study of practitioner beliefs about product placement using Entertainment Resources and Marketing Association (ERMA) members, and found that these practitioners believed unaided recall to be the best measure of product

 O              P     !  " # $% &  '  (*+.( placement success and that portraying the product or service in a favourable light was the most important executional factor. However, considering that most of the respondents were placement agents, whose goal is to get the product on screen (and not necessarily worry about whether it affects sales or brand attitudes), this might explain their emphasis on positive exposure.

Pardun and McKee (1996) surveyed 89 advertising agency media directors, 70% of whom considered themselves knowledgeable or extremely knowledgeable about product placement. Asked to list the factors most important in making a product placement decision for feature films, respondents listed national viewing potential, price of the placement and theme of the movie. Coming from an agency management perspective, it is not surprising that there was such a focus on cost-per-thousand (i.e. balancing reach and cost).

In 1999, Pardun and McKee surveyed 106 public relations firms and 156 advertising agencies to explore their level of involvement with product placement as part of their strategies. Significantly, even in 1999, more than one third of the PR respondents had personally been involved in placing a client’s product in a film as part of a public relations strategy, and many already believed that viewers were aware of placed products. Another third were open to using product placements in the future. This survey revealed that more public relations firms had made product placements (34%) than had the advertising agencies (26%). Less than one third of the public relations practitioners said they would never use product placements for their clients. What the researchers did not explore was why this was the case. Was this because their client’s products were not suitable, or because the practitioners themselves had an attitudinal problem with product placement? It would be interesting to replicate this study now, and see how the results might have changed. Are there still so many practitioners refusing to use product placement?

As with their 1996 study, Pardun and McKee (1999) asked respondents to rank the ten most important reasons a public relations firm might give for placing a client’s product in a particular film. Interestingly, the public relations practitioners provided the same answers as

 O              P     !  " # $% &  '  (*+/( the advertising directors did three years prior. The national viewing potential that a successful film could provide was the most important criterion, followed by the theme of the movie and the price of the placement. They also found that practitioners believed that product placements positively affect the image of a product, and it is this benefit that is sought, not an increase in sales. They also felt that clients would ask them to include product placements more in the future, that product placement was here to stay, and that more marketing dollars would be allocated to product placements. Interestingly, public relations practitioners felt more strongly on these issues than did their 156 advertising agency counterparts who were also surveyed.

With the stakes rising and more evaluative tools available to marketers, Karrh, McKee and Pardun (2003) collaborated to explore whether practitioners had changed their views concerning product placement usage and effectiveness (c.f. Karrh, 1995; Pardun and McKee, 1996, 1999), and did this by surveying 28 members of the Entertainment Resources and Marketing Association. Whilst no factors had decreased in importance, this study revealed that practitioners were now thinking more broadly and considered more factors to be pivotal to a product placement’s success. They wanted to show the brand being used and prevent competing brands from also appearing. They also believed that gaining publicity for the placement itself was important for success. This has been reflected in the rise of cross- promotional campaigns (e.g. BMW and James Bond). However, there still remained a high degree of reliance upon subjective criteria for decision making across these studies, with the results of the Karrh, McKee and Pardun (2003) study indicating only a mild move towards more quantitative measures of product placement effectiveness. Quantitatively, unaided recall and recognition remained the two most popular means of assessing placements, although the tracking of subsequent related sales or measurements of press coverage of the placement were methods that were said to be growing in use.

In 2005, Russell and Belch conducted interviews with 56 American product placement practitioners in order to gain a greater understanding of the objectives they sought and how product placement efforts were measured. They found that the role of product placement

 O              P     !  " # $% &  '  (*++( among clients is generally inconsistent and that there is no one department that takes ownership of the practice and depending on the company, it could be someone from the public relations department, corporate communications, the director of advertising, or a brand manager. They also found that in many organisations, product placement was not part of the integrated marketing plan, but rather an additional activity, tacked on the side, often with no real objectives of its own.

Russell and Belch (2005) note that industry practitioners have only recently started to develop tools to systematically assess placements’ effectiveness. Broadly speaking, two types of tools are being developed in an attempt to establish an industry standard - some focussing on the monetary value of the product placement, others focussing on outcomes such as recall and association – but there is still little consensus as to which one is most suitable. Interestingly, Russell and Belch (2005) also suggest that product placement agencies appear not to advocate any form of measurement, perhaps to avoid being held accountable for their performances, and that many clients appear satisfied by simply seeing the brand included in the entertainment program. Other practitioners believed that the limited financial investment did not necessarily warrant much attention being paid to the returns. However, other practitioners were wanting reliable and valid measures to be developed in order to understand how and when product placement works and how best to integrate it into the overall marketing plan, as well as allowing its effectiveness to be known.

Closer to home, Craig-Lees, Scott and Wong (2006) explored congruence between Australian and US product placement practitioners in terms of decision-making and planning by replicating the Karrh, McKee and Pardun (2003) study in Australia. Australian practitioners viewed the brand having a “natural congruency / relatedness to the program”, “omitting competing brands”, and “showing the product in a favourable light” as the most important executional factors for placement success. They also showed a high level of agreement with the notions that “brands with a very recognisable package or design” and

 O              P     !  " # $% &  '  (*+0( “brands that have their own unique ‘personality’" are more likely to have an impact on an audience.

Practitioners believed that the most effective measurement tool was “sales of the brand after the program is aired/released” which suggests that the measurement of product placement effects are associated with calculating a return on investment and are similar to the methods used to calculate media advertising effects. With post-sales figures and a change in purchase intentions seeming to be the most important measurement tools viewed by Australian practitioners, the question arises as to whether or not Australian practitioners really understand why they should be using product placement, and whether their objectives are realistic and measurable. This emphasis placed on sales performance is in contrast to pre- existing literature by Karrh (1998) who stipulated that product placement can have the effect of reminding consumers about the brand and be integrated with other forms of advertising.

Broadly, Craig-Lees, Scott and Wong (2006) found that the attitudes and beliefs held by Australian practitioners were more similar to those held by US practitioners in 1995 than in 2003, and that like their US counterparts, Australian practitioners were disinclined to implement academic research findings into their decision making. However, in terms of measurement options and beliefs about the practice, Australian practitioners were more aligned to the 2003 results (although these did not differ too much from those in 1995), with no significant statistical differences apparent. These results show support for the statement that Australia is 10-15 years behind that of the US (Canning 2003). Despite this however, there is reasonable evidence to suggest that these views are in transition towards the views held by the practitioners in the 2003 study, and that the differences in views could be a result of time and usage rather than cultural influence. Indeed, the practitioners surveyed concurred that the local industry was still in its infancy and less developed than their counterparts in offshore markets - ‘many markets have a lot more volume and have trialled far deeper product integration into scripts outside the reality genre’.

 O              P     !  " # $% &  '  (*+1( Understanding audience attitudes and perceptions about product placement From the previous discussion, the evidence that product placement is viewed favourably by practitioners is clear. But what do audiences think of product placement? When exploring these attitudes, an evolutionary view to discern whether audiences are becoming more or less approving of product placement is appropriate.

Nebenzahl and Secunda (1993) found that 70.1% of the responses from their student sample considering product placement were positive, and that these attitudes towards product placement were more positive than were attitudes towards commercials. Furthermore, 77.9% agreed that they would allow product placement with varying degrees of encouragement or restrictions. Morton and Friedman’s (2002) findings echoed those of Nebenzahl and Secunda nearly ten years earlier, confirming that consumers do not want product placement prohibited, and that they are not willing to pay more for a movie that contains no product placements. Any objections to product placement have been made on ethical grounds (Gupta and Gould 1997; Gould, Gupta and Grabner-Krauter 2000).

Picking up on this sentiment, Gupta and Gould (1997) focused on the practice’s acceptability and ethical concerns. Here, they found that subjects who watched more movies found product placement more acceptable. However, all respondents (especially women) found guns, cigarettes and alcohol less acceptable than other products. A strength of Gupta and Gould’s (1997) research compared to that of Nebenzahl and Secunda (1993) is that as an introduction to their research, they defined what product placement meant to them, thus providing respondents with a similar context with which to answer the questions. A concern regarding the Nebenzahl and Secunda (1993) study is that such a definition was not reported, and considering that this study was conducted fifteen years ago, there is a high chance that respondents may not have fully understood the concept of product placement.

Gould and Gupta extended this study in 2000, collaborating with Austrian researcher Grabner-Krauter, to examine how different cultures (i.e. American, French, Austrian) may hold similar or different attitudes to product placement. Whilst they confirmed their

 O              P     !  " # $% &  '  (*+2( previous finding that males and frequent moviegoers were more accepting of ethically charged products, they found that there was no difference between the groups in accepting non-ethically-charged products. The Americans found product placement the most acceptable, with the Austrians finding it the least acceptable. However, the three samples were not completely comparable given that the American results had been gathered three years earlier than the European results - the American results might have changed over this time had they been re-surveyed.

Another study considering the consistency of attitudes of product placement across nations and cultures is that of Karrh, Frith and Callison (2001). This research contrasted American and Singaporean students, and found that there was no difference between the amount of attention paid to product placement by either group. However, American audiences were savvier, as they were more likely to believe that brand appearances in programs were the result of paid advertising efforts. Singaporean audiences had more ethical and regulatory concerns about the practice. In a different study, McKechnie and Zhou (2003) found that Chinese consumers were less accepting of product placement than American consumers. Yet unlike American consumers, male and female Chinese consumers differed little in terms of their attitudes towards ethically-charged products being placed in movies.

Panda (2004) examined Indian consumers’ evaluations of product placement as a strategy for providing brand communication and the reactions to the level of ethicality involved in product placement within a Hindi film context. Viewers were generally very positive about product placement describing it as acceptable, frank, amusing and pleasant. Similarly, Argan, Velioglu and Argan (2007) found that amongst 277 actual Turkish moviegoers, there was a favourable attitude towards product placement, but extensive commercial activity in movies was perceived as less ethically acceptable. They also found that higher movie-going frequency and higher levels of movie enjoyment positively impacted the amount of attention paid to product placements.

 O              P     !  " # $% &  '  (*+3( In a global study of 11,300 people by Mediaedge:cia (including Australians), approximately 50% of consumers say they have noticed product placements, with 60% of these consumers willing to try the brands featuring in a film. Of the Australian sample, one third said that they would really consider trying brands that they had seen in a film (Gough 2004; McIntyre 2004). This placed Australian consumers on par with Hong Kong consumers, and ahead of US and European consumers in regards to their acceptability of the practice (Hall 2004).

In regards to age, Mediaedge:cia found that in all countries, 16-34 year olds were the most likely to notice product placements and to consider trying the products they see in films, with these figures declining as respondents got older (Hall 2004). 62% of people aged 16-34 felt it makes sense to see brands in films, compared to 44% of people aged over 55. Similarly, 32% of people aged 16-34 are inclined to buy brands after seeing them in a movie they like, compared to 22% of adults over 55. Furthermore, Nelson and McLeod (2005) found that those adolescents who were more attuned to brands and were more brand conscious were more aware of product placements and were more favourable towards the practice. Adolescents also considered other people to be more influenced by product placements than themselves, especially those peers that they were not as close to (i.e. other classmates as opposed to their own friends).

Looking specifically at baby boomers, Schmoll, Hafer, Hill and Reilly (2006) found they too generally had a positive attitude, regardless of their level of movie-going consumption. Any concerns regarded ethically charged products or activity directed towards children. d’Astous and Seguin (1999) found that consumer evaluations of product placements differed depending on the type of program in which they occurred, with evaluations most negative when they featured in a mini-series / drama. In contrast, high sponsor-program congruity led to better evaluations and better ethical judgements about product placement. They also found that implicit placements (i.e. creative, subtle, background placements) were perceived as less ethical than more explicit placements. Like Russell (2002), Panda (2004)

 O              P     !  " # $% &  '  (*0,( found that a high level of congruence between the brand and the storyline was received more favourably by audiences and that when the brand looked out of context or was forced upon the plot, it created irritation among the audience. He also found, like d’Astous and Seguin (1999) that the consumer’s evaluation of an explicit integrated product placement was more positive than that of an implicit product placement. Furthermore, implicit placements were judged to be significantly less ethical than explicit placements.

Gould and Gupta (2006) investigated the product placement environment of game shows as a site of consumer meaning and cultural production, using interpretive approaches to see how the roles of the brands differed in this context to that of film and television drama. They highlighted a need to understand the meanings consumers derive from product placement, as opposed to only looking at the effects, especially since these effects are “intertextually embedded in meaning, discourse and reflexivity: only through these can attitudinal and behavioural change take place” (Gould and Gupta 2006, p78). They found that whilst in movies, product placement appropriateness is constructed in terms of realism and interferences with the experience, in game shows, products are a major part of the action, with appropriateness determined in terms of fitting the product to the show and its viewers. In this way, ‘appropriateness’ (akin to the established marketing communications tool of targeting) becomes the most important aspect of product placement meaning, and is related to the realism, acceptability and plot connection that other researchers talk about.

Criticism of product placement as a deceptive practice is based on the premise that the appearance of placements in movies has a causal relationship to purchase behaviour, with moviegoers unaware of the persuasive intent behind placements and naïve about the practice in general. However, this review convincingly demonstrates that moviegoers are more sophisticated in their understanding of the practice of product placement than critics would have public officials believe. Moviegoers are active interpreters, not passive receptors of encountered brands. They are not uniformly influenced, and they are very aware of the persuasive intent of product placement, a condition that leads to skepticism and resistance of persuasive attempts. Moviegoers may allow themselves certain buying indulgences in some

 O              P     !  " # $% &  '  (*0-( buying situations, and for varying reasons, but they are not deceived into rushing out to buy everything they see in a movie (DeLorme and Reid 1999). Indeed, from a public policy standpoint, it is evident that product placement is not a stealth marketing tactic that works at consumers' subconscious levels. Rather, the practice is widely known and understood by the public (Ong 2004).

Measuring product placement effectiveness Considering its use in practice, there has been little research regarding product placement effectiveness. Most research into product placement effectiveness has focused on memory for the placement. So in this context, effectiveness = memory. But is there more to it than this?

Such research has generally employed aided or unaided same-day or day-after recall or recognition, and has simply aimed to determine how many products could be remembered (e.g. Babin and Carder 1996; Gupta and Lord 1998; Brennan, Dubas and Babin 1999). These studies have found relatively mild and mixed effects on memory from product placement, suggesting that not all product placements are equally effective. Thus, there is a paucity of research relating not only to how product placement works, but also, if it works.

Basic models of advertising effectiveness (all of which are based on the Hierarchy of Effects model), assume that unless awareness (i.e. conscious attention or processing) takes place (and is understood by researchers and practitioners), any attitudes or intentions that may follow are irrelevant. In this way, the first stage of the hierarchy, awareness, must take place for the other effects to happen (Barry 2002; Weilbacher 2002). So unless moviegoers are aware of, remember, and store product placement information, there is no way that these product placements can be thought to reliably correlate with any level of preference or intention formation. Furthermore, if the assumption is that the key benefits of product placement are the positive associations that the placement provides (thereby enhancing social learning), then it is useful to know if the audience is at least aware of the products being placed. It is also useful to explore whether these audiences are potentially able to

 O              P     !  " # $% &  '  (*0*( transfer this recognised association into memory. Thus, a continued focus on memory and recall is justified, particularly if aiming to gain a clearer understanding of the effects of product placements on memory in order for any potential influence of product placement to be proposed.

Early memory-based product placement studies were simplistic and accrued recall results that were fairly diverse. For example, Stoertz (1987) found that when given product category cues, 38% of respondents could correctly recall brands in films they had seen the day before. Ong and Meri (1994) found no improvement in memory for some product placements but remarkably large improvements in memory for others. Babin and Carder (1996) obtained inconsistent results from two similar movies and two similar samples. 40% of brands in Rocky III were recognised, compared to 24.1% of brands in Rocky IV. In addition, there was a high incidence of “false positives”.

The mixed and weak results of these early studies on effectiveness are in part because they generally failed to recognise the multi-dimensional nature of product placement. They aimed to determine how many brands could be remembered and made no attempt to explain why this may be. In doing so, they tended to define product placements as similar, regardless of their modality, prominence, or level of plot connection, and considered all audience members to process placements in the same way. Recently however, there have been some studies that have attempted to consider these complexities.

Gupta and Lord (1998) were the first researchers to extend the product placement literature, by considering the effects of prominence and mode of product placement on recall. In doing so, they demonstrated that some placements were better recalled than others. For example, prominent placements accrued higher recall than subtle placements. Brennan, Dubas and Babin (1999) also concluded that prominent placements were better recognised than subtle placements. Furthermore, they determined that prominence accounted for a greater percentage of variation in viewer recognition than that which was explained by placement

 O              P     !  " # $% &  '  (*0.( exposure time. d’Astous and Chartier (2000) also found that prominence enhanced recognition (but had a negative impact on recall).

However, research addressing the effect of aural versus visual versus audio-visual delivery on placement recall and recognition has produced conflicting results. Sabherwal, Pokrywczynski & Griffin (1994) found that audio-visual placements accrue the highest recall. Russell (2002) and Gupta and Lord (1998) found that audio placements lead to higher recall than visual placements. Law and Braun (2000) found that audio-visual placements have the highest recall, followed by visual and then aural placements (although audio placements are better recognised than visual placements).

The next significant variable that was considered empirically was that of integration or plot connection. Such research seemed essential since the success of product placement is supposedly grounded in the notion of seamless integration between product and plot. A brand with a higher plot connection contributes much to the story, providing a major thematic element due to its high integration to the story (Russell 1998). d’Astous and Chartier (2000) found that integration of the placement with the plot had a positive impact on consumer liking and level of acceptance of the placement, but a negative impact on recall. This finding was reinforced by Russell (2002) who found that incongruent placements (i.e. those that were visual and had high plot connection, or aural and had low plot connection) were better recalled than congruent placements. This was particularly the case for visual placements – plot connection did not have significant effects on recall of audio placements. However, like d’Astous and Chartier (2000), Russell (2002) also found that congruent placements led to greater attitudinal / persuasive changes than incongruent placements.

The effect of connectedness on recall has also been studied. Connectedness refers to the intensity of the relationships viewers develop with TV programs and the characters in those programs (Russell, Norman and Heckler 2004). The authors suggest that the processing and storage of program-specific information will differ between high and low levels of the

 O              P     !  " # $% &  '  (*0/( construct, with highly connected viewers finding the information in the program to be more essential to their lives than less connected viewers. Highly connected viewers consider the content more important and relevant to their world and may even mould characteristics of their lives to match those depicted in the program, and form relationships with the characters, who then become a source of influence, especially in relation to their product consumption. Therefore Russell, Norman and Heckler (2004) hypothesised that as connectedness increased, memory for product placements would improve, and found that highly connected viewers recalled significantly more brands than low-connected viewers. Extending this research, Russell and Stern (2006) looked to answer the question of how characters’ relations to placed products and consumers’ relations to those characters may affect consumers’ attitudes to the product. They found that consumers align their attitudes towards products with the inside-program characters’ attitudes to the product, with this alignment process driven by the consumers’ extra-program attachment to the characters. This demonstrates the high importance of a character’s effect on viewers’ attitudes towards product placements, especially when the character’s attitude towards the product is positive.

Arguably, the broadest research to this point was that done by Scott (2002) who, in the one study, aimed to identify which audience characteristics and which product placement characteristics led to improvements in recognition by applying theory such as memory organisation and storage, program involvement, spokesperson attractiveness and endorsement theory. Through a multi-stage research strategy, culminating in the screening of the movie Legally Blonde, and measuring the characteristics of the product placements, the involvement that the participants (males and females aged 12-16) had with the movie, the degree to which they liked the star Reese Witherspoon, and their level of familiarity with the product categories and brands in the movie, this research revealed felt pleasure, cognitive effort and star liking to have positive relationships with recognition. Relative to the characteristics of the placement itself, product placements which were visual in nature, had a high amount of screen time, and depicted products that were familiar to the audience, were found to have positive relationships with recognition.

 O              P     !  " # $% &  '  (*0+( Prior to this research, no single study had investigated so many variables that could impact recognition. It did this by considering these factors from both an audience perspective and a product perspective. Both perspectives are equally worth investigating because if effective product placement decisions are to be made, placement managers need to have some idea of both types of factors that can impact and what their relative impact may be. Of particular interest and significance was the concept of movie involvement, which was considered in product placement research for the first time. In this context, movie involvement was comprised of both emotional and cognitive aspects.

Since all prior research has utilised explicit memory tests, this in part can explain why, to date, no known research has found effects on brand attitudes from product placement or purchase intention. Vollmers and Mizerski (1994) found no differences in either attitude toward the brand or attitude toward a paired actor between experimental and control groups. Similarly, Karrh (1994) found no changes in the evaluation of placed brands. Ong and Meri (1994) found that even among those respondents who could remember the brands in a movie, there was no increase in purchase intention, brand awareness, favourable brand perception or brand confidence. The only study to show any potential signs of influence must be embraced with caution. Baker and Crawford (1996) found that after watching a movie, purchase intention for brands that appeared in that movie was 16% higher than the purchase intention of those brands that respondents had previously named as their favourite brands. However, the small sample size (n=43) and the inability for these results to be replicated brings the generalisability and validity of these results under much doubt.

As can be seen, prior product placement studies have relied almost exclusively on explicit memory retrieval (i.e. recall, recognition), whereby a consumer consciously thinks back to a prior exposure episode (in this case, a product placement), and intentionally attempts to access the information that was presented (Shapiro and Krishnan 2001). However, the lesser studied form of memory retrieval in marketing is implicit memory, whereby an individual changes their performance on a task due to a prior exposure episode, but does not deliberately attempt to recollect this previously encoded information (Schacter 1987). Law

 O              P     !  " # $% &  '  (*00( and Braun (2000) were the first researchers to consider using implicit measures to explore product placement effects. In doing so, they found that although placed products were chosen more frequently than products that were not placed, this choice was not found to reliably correlate with recognition or recall. Auty and Lewis (2004a) also found that there was no difference in the product children chose following exposure to a film, regardless of whether they could recall the product placement or not.

Law and Braun-LaTour (2004) contend that research into product placement effects has been held back by the research methods employed. Specifically, they argue that recall and recognition are not capable of detecting the more subtle effects of product placement, and thus advocate the use of implicit measures. Indeed, implicit memories have been shown to lead to a response bias in which there is a greater likelihood of using the previously seen information to complete a task without the awareness of doing so (Shapiro and Krishnan 2001; Lee 2002). So, failure to remember exposure to a product placement does not preclude the possibility that it has affected consumer behaviour processes such as brand consideration (Auty and Lewis 2004b). Consequently, these researchers argue that implicit tests should be considered more and utilised in product placement research.

Despite using explicit measures herself, Russell (2002) recognised the limitations of pure explicit memory-based research, stating that the reliance on brand recall and recognition measures presumes that the effects for memory are similar to the effects for attitude, and pointed to the absence of correlations between memory and attitude measures often found in the persuasion literature (e.g. Petty, Cacioppo and Schumann 1983) which suggest that the memory-attitude relationship is not necessarily linear. Since recall may be a poor predictor of persuasion (Mackie and Asuncion 1990), she suggested that research on the effectiveness of product placements should investigate both memory and attitude effects (Russell 2002). She was right. Her study showed that conditions that maximised memory did not necessarily maximise persuasion. Whilst incongruency between modality and plot connection improved memory, congruency enhanced persuasion (Russell 2002).

 O              P     !  " # $% &  '  (*01( van Reijmersdal et al (2007) also used implicit tests to look at the effects of product placement on brand image – something that had previously remained unstudied, which is surprising given that brand image change is often mentioned as one of the benefits of using product placement (Karrh 1998; DeLorme and Reid 1999). They conducted two stages of research – a survey and an experiment. In both stages, brand image changed in the direction of the program. In the survey stage, van Reijmersdal et al (2007) found that the brand image of people who watched more than two episodes of the program was more in agreement with the program image. This effect held when controlled for age, education, gender, attitude towards product placement and brand use. However, there was no effect of brand memory on brand image. Therefore, without consciously remembering having seen the product placement, exposure to the product placement did affect brand image. However, in the experimentation stage, any level of exposure to the program resulted in scores on brand image becoming more in agreement with scores on program image (again, memory did not affect brand image or brand attitude). These memory results are in line with Law and Braun (2000) and Auty and Lewis (2004b) who showed that product placement effects on brand choice were unrelated to memory. They also indicate that brand image is influenced implicitly, which means that image is influenced without explicit memory of the exposure. In sum, these results support the idea that brand image and brand memory are processed differently, and are in agreement with the evolving view that different measures are needed to estimate effects of product placement (Law and Braun 2000). The implication for managers is that since program image places such an important role in the changes in brand image, advertisers should make a well-defined choice for a program that conveys the desired image before they place their product in it (van Reijmersdal et al. 2007). For researchers, these results highlight the importance of research context, and suggest that experiments might not be the best way to test for product placement effects, as they are not reflective of the true viewing environment since participants purposely pay more attention. In this research, the higher attention levels in the experimentation group may explain why fewer exposures were needed to change brand image as respondents were asked to watch the episodes attentively.

 O              P     !  " # $% &  '  (*02( APPENDIX 4.1: SECTIONS 5 AND 6 OF AUDIENCE ENGROSSMENT SURVEY

PART 5

In this section we would like to gain a more general understanding of your movie-going experience.

For each statement, please circle the number that best Agree describes your thoughts. Again, these questions ask Disagree Slightly Agree Strongly Agree Strongly you to agree or disagree with each statement. Slightly Disagree Strongly Disagree Strongly

There were some external circumstances that prevented me from paying full attention to the movie 1 2 3 4 5 6 (e.g. there was a baby crying, outside noise, people talking) I was comfortable in my seat 1 2 3 4 5 6 The air temperature was pleasant 1 2 3 4 5 6 I had a clear view of the screen 1 2 3 4 5 6 The sound quality was good 1 2 3 4 5 6 The image was good quality 1 2 3 4 5 6 I was in a good mood when I came to see this movie 1 2 3 4 5 6

 O              P     !  " # $% &  '  (*03(

ABOUT YOU: Research shows that there are some individual characteristics that affect how people interact with and respond to entertainment programs. We would now like to ask you about these.

Are you Male % (1) Female % (2)

Which option represents the age bracket you belong to? 14 or under % (1) 35-44 % (4) 65+ % (7) 15–24 % (2) 45-54 % (5) 25-34 % (3) 55-64 % (6)

On average, how many movies would you watch each month (this includes those on television, on DVD/video, at the cinema)? 0-5 % 1) 6-10 % (2) 11-20 % (3) 21+ % (4)

What is the highest level of education you have completed or are currently completing? High school % (1) Undergraduate degree % (3) Technical qualification % (2) Postgraduate degree % (4)

In regards to movie watching, I consider myself a: Formally qualified critical viewer % (1) Self-taught critical viewer % (2) Non-critical viewer % (3)

 O              P     !  " # $% &  '  (*1,( APPENDIX 4.2:

OVERVIEW OF KEY STATISTICS AND INDICATORS

Thresholds Thresholds are the point between two adjacent response categories where either response is equally probable. As a straightforward generalisation of the classical Rasch model for dichotomous responses (Rasch 1960), the polytomous model (Andrich 1988) provides additional threshold parameters which characterise the transition points between two adjacent response categories. A polytomous item with m categories requires m-1 threshold parameters. The average of the threshold parameters can be regarded as an overall location of the item.

When applying the polytomous Rasch model, one expects the item scores to be ordered categorical. This means that the order of the categories (from easy to hard) should correspond to the order of respondents (from less able to more able). This would be indicated by an ordered set of thresholds within each of the items (Pallant, Miller and Tennant 2006).

However, since the thresholds between two adjacent response categories are estimated independently of other thresholds and no constraints are imposed on the order of the thresholds by the model, the threshold estimates can theoretically take any order. Consequently, the order of the thresholds reflects characteristics of the response process which became manifest in the data. Given that fit need not be affected by threshold disordering, there has been much debate as to whether disordered (or reversed) thresholds matter. In particular, if reversed thresholds are located close to each other, the reversal might actually be accidental rather than an indication of substantial problems (Scott and Salzberger 2008).

If one endorses the view that properly ordered thresholds are a critical requirement, albeit being external to the model, then the investigation of the threshold ordering precedes fit assessment. In any case, correctly ordered thresholds are a desirable

 O              P     !  " # $% &  '  (*1-( property. In many cases reversed thresholds hint at problems underlying the process of responding to the item leading to misfit. This can occur, for instance, when there are too many response options, or when the labelling is confusing – thus suggesting a lack of common understanding of use of the rating scale between the researcher and the participants (Ewing, Salzberger and Sinkovics 2005; Pallant et al. 2006). However, these problems can be resolved (post hoc) by combining categories and reanalysing the data to reassess the optimal number of categories for that data. A more thorough approach however, is the collection of new data based on a revised response scale. This approach was taken for this research, with the threshold structure investigated carefully during each stage of scale development and the response scale format amended when necessary.

Person Separation Index An estimate of the internal consistency of the items is based on the Person Separation Index where the estimates on the logit scale for each person are used to calculate reliability (Bhakta, Tennant, Horton, Lawton and Andrich 2005). The Person Separation Index describes how well the items discriminate between people and provides an indication of replicability for person placement across other items measuring the same construct. In this way, it has a very important role in understanding the fit statistics in the Rasch model as it impacts the power of the test of fit. Whilst the formulas are different, it is conceptually equivalent to Cronbach’s alpha (Green and Frantom 2002). Similar to Cronbach’s alpha, perfect reliability would be 1.0 and random data would generate a relationship of 0. The value of the PSI depends in part on the actual variance of the persons as well as error variance.

Fit Statistics A statistic known as “fit” provides an internal mechanism for identifying inappropriate responses to items, thus allowing for exclusion or re-assessment of persons whose responses make no sense or of items that do not perform as expected (i.e. do not fit according to our understanding of the construct). This provides us with a sense of the usefulness of the measure (Green and Frantom 2002). In checking how well the data fit the model, it is important to be able to diagnose very quickly where the misfit is worst,  O              P     !  " # $% &  '  (*1*( and then try to understand this misfit in terms of the construction of the item and the understanding of the variable in terms of its theoretical development.

Person fit to the Rasch model is an index of whether individuals are responding to items in a consistent manner, or if responses are idiosyncratic or erratic. Responses may fail to be consistent when people are bored and inattentive to the task, when they are confused, or when an item evokes an unusually salient response from an individual (Green and Frantom 2002). Item fit (the more commonly reported statistic of the two) is an index of whether items function logically and provide a useful continuum for respondents. An item may misfit because it is too complex, confusing, or because it actually measures a different construct (Green and Frantom 2002).

To determine how well each question fits the model, and so contributes to a single trait, a set of “fit” statistics are used which test how far the observed data match those expected by the model (Bhakta et al. 2005). Items with an acceptable fit index (i.e. that fit better) are more useful in measuring a trait than items that fit poorly (Green and Frantom 2002). A significant chi-square value (as per the Bonferroni adjustment) indicates a high degree of misfit and that the relative location of the question difficulty is not constant across the trait. Item fit statistics are also examined as residuals (a summation of deviations of individual responses from the expected response for the question). Both people and items can overfit or underfit. Overfit (represented by a large negative residual <-2.5 in RUMM) is interpreted as too little variation (over- discrimination) in the response pattern, perhaps indicating the presence of redundant items. Although it provides a guide to refining an instrument, it is otherwise of little concern. In contrast, underfit (noise) is indicated by a high positive residual (>2.5) and suggests unusual or inappropriate response patterns, with the item under-discriminating. These indices can be used to identify and sometimes correct a measurement disturbance (Green and Frantom 2002)

Misfit of a question indicates a lack of expected probabilistic relationship between that item and other items measuring that dimension, thus indicating that the item does not

 O              P     !  " # $% &  '  (*1.( contribute to the trait under consideration (Bhakta et al. 2005). Alternatively, the item may suffer from Differential Item Functioning (DIF) whereby it operates differently for different groups, this violating a key assumption of Rasch modelling – invariance. From a practical perspective, where the data is shown to misfit, it is a signal that the item data needs to be examined and in most cases, the item is excluded from the scale. Relative to this research, the rationale for the omission is based on the assumption that the relationship between audience engrossment and the likelihood of a specific response pattern should be consistent across items. Where this relationship breaks down, it is assumed that the item is measuring a separate variable. In this way, fit provides an index of the degree to which responses conform to a logical pattern as well as an indication of the measure’s validity for a specific individual. Similarly, fit permits assessment of the validity of the overall measure by providing a means to identify poorly functioning and/or biased items.

Item Characteristic Curves Given an item and its difficulty, the non-linear curve depicting the relations of person parameter and probability of agreement is called the item characteristic curve (ICC). ICCs provide a visual impression of the probability of success on the item for each possible location of a person on the continuum. The dots on the graph represent the observed class interval averages and whether they are in accord with the model predictions. Ideally the observed proportions are close to the theoretical curve. Items with good fit will show each of the group plots lying on the curve.

 O              P     !  " # $% &  '  (*1/( A good fitting Item Characteristic Curve

Poorly fitting items can easily be identified as the class intervals plotted on the ICC for that item show that success on the item does not increase consistently with an increase in ability. Moreover, ICCs can reflect how well the item discriminates. Those with plots that are flatter than the curve and fit residuals >2.5 are said to be under- discriminating. Those with plots that are steeper than the curve and with fit residuals <-2.5 are considered to be over-discriminating and there is a greater dependence among responses in one form or another.

Poorly fitting Item Characteristic Curves

Item Maps Person-item location maps provide a simultaneous positioning of items and person responses on a common scale to illustrate where responses place each person with respect to those items. They provide visual insights into the data which can be used to determine the extent to which item positions match person positions (Green and Frantom 2002). If the positions do not line up, the items are likely to be inappropriate  O              P     !  " # $% &  '  (*1+( for the people (i.e. too easy or too hard). Similarly, gaps in the measure can be detected, suggesting where items might be added. Item maps also allow item order to be reviewed to assess the validity of the measure and whether item difficulty as perceived by the scale developer is the same as that of the respondent. This assessment of the construct validity requires that the items designed to provoke a correct response from the most able people should only appear towards the most difficult part of the continuum and vice versa.

Differential Item Functioning In promoting generalisability, it is necessary to test whether the items work in the same way for different groups of people who might be compared. In this study, these groups included people of different genders, age groups, movie watching frequency, educational background and movie critiquing expertise, as well as those that had seen the movie previously compared to those who were seeing it for the first time.

Differential Item Functioning (DIF) refers to the occurrence of different ICC patterns for different person groups. This can be a problem because it indicates that the variable may operate differently across different groups and that the scale is not invariant across these groups. This situation is not consistent with the Rasch model since the basic premise is that for the same location of a person (as best as can be estimated from the data), the expected value on an item is the same irrespective of what group the person might belong to. The main advantage of the Rasch model in the study of invariance is that it has this property built into its own structure and it can check not only whether there is DIF, but if so, where it is (Andrich and Styles 2004).

In RUMM2020 the presence of DIF can be detected both statistically and graphically. In checking for DIF across groups, it is possible to estimate the parameters in the groups, and check if they are invariant, or to estimate the ICCs in the different groups and compare them. By dividing the sample into class intervals within subgroups, it is possible to generate a plot for each group separately. Essentially, the ICC for any

 O              P     !  " # $% &  '  (*10( subgroup should not differ from the curve of the whole group. Because the ICC is derived from the parameters, these two approaches must reflect each other.

A good fitting item with no DIF

An item showing DIF between genders

Statistically, DIF is identified by a two-way ANOVA of the residuals, with the person factor as one factor (whether the person had seen the movie before, gender, age, movie watching frequency, education level and a self-assessment of their critical movie- watching abilities), and the class interval as the other. Where there are more than two

 O              P     !  " # $% &  '  (*11( levels of a factor, Tukey’s post hoc test is used to indicate which groups are contributing to the significant differences (Bhakta et al. 2005). If an item shows DIF, this means that people with the same total score have different probabilities of responding at a particular level of that item. This is also referred to as bias, since people in one group have a greater probability of receiving a particular score than members of another group who have achieved the same total score (Andrich and Styles 2004). This is significant because any item bias can affect model fit.

Two types of DIF may be identified. 1. Uniform DIF is where the group shows a consistent systematic difference in their responses to an item, across the whole range of the attribute being measured. Uniform DIF is indicated by a significant main effect for the person factor (e.g. gender). When detected, the problem can be remedied by splitting the file by group (for that item only) and separately calibrating the item for each group. 2. Non-uniform DIF is when there is non-uniformity in the differences between the group (i.e. it varies across levels of the attribute). The presence of non-uniform DIF is indicated by a significant interaction effect (person factor x class interval). There is little that can be done to correct the problem, and it is often necessary to remove the item from the scale.

 O              P     !  " # $% &  '  (*12( APPENDIX 5.1: STAGE 2 - ALL FEELING ITEMS

 O              P     !  " # $% &  '  (*13(  O              P     !  " # $% &  '  (*2,(  O              P     !  " # $% &  '  (*2-( APPENDIX 5.2: STAGE 2 – ALL FEELING ITEMS WITH COLLAPSED THRESHOLDS

 O              P     !  " # $% &  '  (*2*(  O              P     !  " # $% &  '  (*2.(  O              P     !  " # $% &  '  (*2/( APPENDIX 5.3: STAGE 2 - ALL AROUSAL ITEMS

 O              P     !  " # $% &  '  (*2+(  O              P     !  " # $% &  '  (*20(  O              P     !  " # $% &  '  (*21( APPENDIX 5.4: STAGE 2 - ALL AROUSAL ITEMS WITH COLLAPSED THRESHOLDS

 O              P     !  " # $% &  '  (*22(  O              P     !  " # $% &  '  (*23(  O              P     !  " # $% &  '  (*3,( APPENDIX 5.5: STAGE 2 – ALL APPRAISAL ITEMS

 O              P     !  " # $% &  '  (*3-(  O              P     !  " # $% &  '  (*3*(  O              P     !  " # $% &  '  (*3.( APPENDIX 5.6: STAGE 2 – APPRAISAL ITEMS WITHOUT ACTOR OR GENRE ITEMS

 O              P     !  " # $% &  '  (*3/(  O              P     !  " # $% &  '  (*3+(  O              P     !  " # $% &  '  (*30( APPENDIX 5.7: STAGE 2 – APPRAISAL ITEMS WITHOUT ACTOR AND GENRE ITEMS AND MIND3

 O              P     !  " # $% &  '  (*31(  O              P     !  " # $% &  '  (*32(  O              P     !  " # $% &  '  (*33( APPENDIX 5.8: STAGE 2 – ALL COGNITIVE EFFORT ITEMS

 O              P     !  " # $% &  '  (.,,(  O              P     !  " # $% &  '  (.,-(  O              P     !  " # $% &  '  (.,*( APPENDIX 5.9: STAGE 2 – TWO COGNITIVE EFFORT DIMENSIONS

 O              P     !  " # $% &  '  (.,.(  O              P     !  " # $% &  '  (.,/(  O              P     !  " # $% &  '  (.,+(  O              P     !  " # $% &  '  (.,0(  O              P     !  " # $% &  '  (.,1( APPENDIX 5.10: STAGE 3 – ALL FEELING ITEMS

 O              P     !  " # $% &  '  (.,2(  O              P     !  " # $% &  '  (.,3(  O              P     !  " # $% &  '  (.-,(  O              P     !  " # $% &  '  (.--(  O              P     !  " # $% &  '  (.-*( APPENDIX 5.11: STAGE 3 – FEELINGS DIMENSION WITHOUT APPALLED AND CROSS

 O              P     !  " # $% &  '  (.-.(  O              P     !  " # $% &  '  (.-/(  O              P     !  " # $% &  '  (.-+(  O              P     !  " # $% &  '  (.-0(  O              P     !  " # $% &  '  (.-1( APPENDIX 5.12: STAGE 3 - ALL AROUSAL ITEMS

 O              P     !  " # $% &  '  (.-2(  O              P     !  " # $% &  '  (.-3(  O              P     !  " # $% &  '  (.*,(  O              P     !  " # $% &  '  (.*-(  O              P     !  " # $% &  '  (.**( APPENDIX 5.13: STAGE 3 – ALL AROUSAL ITEMS (RECODED)

 O              P     !  " # $% &  '  (.*.(  O              P     !  " # $% &  '  (.*/(  O              P     !  " # $% &  '  (.*+(  O              P     !  " # $% &  '  (.*0(  O              P     !  " # $% &  '  (.*1( APPENDIX 5.14: STAGE 3 – ALL APPRAISAL ITEMS

 O              P     !  " # $% &  '  (.*2(  O              P     !  " # $% &  '  (.*3(  O              P     !  " # $% &  '  (..,(  O              P     !  " # $% &  '  (..-(  O              P     !  " # $% &  '  (..*( APPENDIX 5.15: STAGE 3 - ALL APPRAISAL ITEMS WITH COLLAPSED THRESHOLDS

 O              P     !  " # $% &  '  (...(  O              P     !  " # $% &  '  (../(  O              P     !  " # $% &  '  (..+(  O              P     !  " # $% &  '  (..0(  O              P     !  " # $% &  '  (..1( APPENDIX 5.16: STAGE 3 – ALL COGNITIVE EFFORT ITEMS

 O              P     !  " # $% &  '  (..2(  O              P     !  " # $% &  '  (..3(  O              P     !  " # $% &  '  (./,(  O              P     !  " # $% &  '  (./-(  O              P     !  " # $% &  '  (./*( APPENDIX 5.17: STAGE 3 - ALL COGNITIVE EFFORT ITEMS WITH COLLAPSED THRESHOLDS

 O              P     !  " # $% &  '  (./.(  O              P     !  " # $% &  '  (.//(  O              P     !  " # $% &  '  (./+(  O              P     !  " # $% &  '  (./0(  O              P     !  " # $% &  '  (./1( APPENDIX 5.18: STAGE 3 – “DIFFICULT” COGNITIVE EFFORT ITEMS ONLY

 O              P     !  " # $% &  '  (./2(  O              P     !  " # $% &  '  (./3(  O              P     !  " # $% &  '  (.+,(  O              P     !  " # $% &  '  (.+-(  O              P     !  " # $% &  '  (.+*( APPENDIX 5.19: FINAL FEELINGS DIMENSION – ALL MOVIES

 O              P     !  " # $% &  '  (.+.(  O              P     !  " # $% &  '  (.+/( Threshold Map – Shooter

RUMM2020 Project: READING Analysis: XFEELING99 Title: X - W/O APPALLED, CROSS. SCARED, SAD G/S Date: 16 Aug 2007 05:13:55 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------3.0 | | | | miser.2 | 2.0 | sadXF.2 | | dpres.2 dstrs.2 horif.2 heart.2 distr.2 dsmay.2 scaXF.2 | sadXM.2 elate.2 | raged.2 angry.2 1.0 | | terif.2 | uplif.2 cncer.2 happy.2 scaXM.2 o | miser.1 dpres.1 oo | comft.2 0.0 o | oooo | ooo | distr.1 scaXM.1 edge.2 ooooo | good.2 oooooooo | terif.1 sadXM.1 -1.0 oooo | elate.1 oooooooooooo | dstrs.1 raged.1 comft.1 horif.1 oooooooooo | uplif.1 heart.1 sadXF.1 ooooooooooo | dsmay.1 oooooooooooo | angry.1 scaXF.1 -2.0 ooooooooooo | happy.1 oooooooooooooo | ooooooooo | oooo | cncer.1 good.1 oooooooooo | -3.0 o | edge.1 oooooo | | oooo | | -4.0 | oooooo | | | | -5.0 ooooo | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.++( Item Map – Shooter

RUMM2020 Project: READING Analysis: XFEELING99 Title: X - W/O APPALLED, CROSS. SCARED, SAD G/S Date: 16 Aug 2007 05:14:05 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------2.0 | | | | miser | 1.0 | dpres | | distr o | oo | dstrs scaXM horif elate sadXM sadXF 0.0 o | raged scaXF terif dsmay heart oooo | angry ooo | uplif ooooo | comft oooooooo | happy -1.0 oooo | cncer oooooooooooo | oooooooooo | ooooooooooo | edge good oooooooooooo | -2.0 ooooooooooo | oooooooooooooo | ooooooooo | oooo | oooooooooo | -3.0 o | oooooo | | oooo | | -4.0 | oooooo | | | | -5.0 ooooo | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.+0(  O              P     !  " # $% &  '  (.+1( Threshold Map – Spiderman

RUMM2020 Project: READING Analysis: SFEELING99 Title: S - W/O CROSS, APPALLED. SCARED, SAD G/S Date: 16 Aug 2007 05:15:05 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------4.0 | | | | | 3.0 | scaSM.2 | | sadSM.2 horif.2 | | terif.2 2.0 | | distr.2 scaSF.2 | dstrs.2 dsmay.2 cncer.2 angry.2 | miser.2 dpres.2 | heart.2 sadSF.2 1.0 | | miser.1 raged.2 o | uplif.2 dpres.1 elate.2 o | oo | raged.1 0.0 oo | distr.1 horif.1 comft.2 ooooo | dsmay.1 edge.2 oooo | dstrs.1 terif.1 happy.2 oooooooo | ooooo | scaSM.1 good.2 angry.1 -1.0 oooooo | ooooooooooo | oooo | ooooooooooo | cncer.1 ooooooooooo | elate.1 -2.0 oooo | sadSM.1 ooooooooooooo | ooooo | comft.1 scaSF.1 oooooooo | heart.1 edge.1 ooooooo | -3.0 oooo | sadSF.1 uplif.1 ooo | ooo | ooooo | oooo | -4.0 | good.1 oo | oo | happy.1 | oo | -5.0 o | | | | oo | -6.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.+2( Item Map – Spiderman

RUMM2020 Project: READING Analysis: SFEELING99 Title: S - W/O CROSS, APPALLED. SCARED, SAD G/S Date: 16 Aug 2007 05:15:13 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------2.0 | | | | horif | scaSM 1.0 | distr dpres miser | terif o | raged dstrs dsmay o | angry oo | sadSM 0.0 oo | cncer ooooo | scaSF oooo | oooooooo | elate ooooo | heart -1.0 oooooo | sadSF ooooooooooo | comft uplif oooo | edge ooooooooooo | ooooooooooo | -2.0 oooo | ooooooooooooo | ooooo | good happy oooooooo | ooooooo | -3.0 oooo | ooo | ooo | ooooo | oooo | -4.0 | oo | oo | | oo | -5.0 o | | | | oo | -6.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.+3(  O              P     !  " # $% &  '  (.0,( Threshold Map – Perfect Stranger

RUMM2020 Project: READING Analysis: PFEELING99 Title: P - W/O APPALLED, CROSS. G/S SCARED, SAD Date: 16 Aug 2007 ------LOCATION PERSONS ITEMS [uncentralised thresholds] ------6.0 | | | scaPM.2 | | 5.0 | | | | | 4.0 | | | | | 3.0 | | | | raged.2 | sadPF.2 2.0 | miser.2 angry.2 | o | dsmay.2 | dpres.2 | distr.2 happy.2 terif.2 1.0 | dstrs.2 sadPM.2 horif.2 | good.2 heart.2 | elate.2 comft.2 o | uplif.2 cncer.2 | 0.0 oo | scaPF.2 edge.2 oooo | oo | elate.1 dpres.1 miser.1 uplif.1 ooo | comft.1 oooooooooo | raged.1 -1.0 oooooooooo | happy.1 angry.1 oooooo | heart.1 ooooooooo | distr.1 ooooooooo | dsmay.1 ooooooo | good.1 sadPF.1 scaPM.1 terif.1 -2.0 oooooo | sadPM.1 dstrs.1 oooooooooooooooooooo | horif.1 oooooooo | oooooooooo | | cncer.1 -3.0 oooooooooo | scaPF.1 oo | oooooooooo | | ooooooo | edge.1 -4.0 | | ooooooooooo | | | -5.0 | oooooo | ooooo | | | -6.0 |

 O              P     !  " # $% &  '  (.0-( Item Map – Perfect Stranger

RUMM2020 Project: READING Analysis: PFEELING99 Title: P - W/O APPALLED, CROSS. G/S SCARED, SAD Date: 16 Aug 2007 05:15:59 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------2.0 | | scaPM o | | | 1.0 | | raged miser | angry dpres o | | happy sadPF 0.0 oo | uplif comft dsmay elate oooo | terif heart distr oo | sadPM ooo | good horif dstrs oooooooooo | -1.0 oooooooooo | oooooo | cncer ooooooooo | scaPF ooooooooo | ooooooo | edge -2.0 oooooo | oooooooooooooooooooo | oooooooo | oooooooooo | | -3.0 oooooooooo | oo | oooooooooo | | ooooooo | -4.0 | | ooooooooooo | | | -5.0 | oooooo | ooooo | | | -6.0 | ------o = 1 Person ------

 O              P     !  " # $% &  '  (.0*(  O              P     !  " # $% &  '  (.0.( Threshold Map – Mr Bean’s Holiday

RUMM2020 Project: READING Analysis: BFEEL99 Title: B - W/O CROSS, APPALLED. G/S SCARED, SAD Date: 16 Aug 2007 ------LOCATION PERSONS ITEMS [uncentralised thresholds] ------7.0 | | | scaBF.2 | | 6.0 | | | | | 5.0 | | | | | 4.0 | | | o | terif.2 | sadBF.2 horif.2 3.0 | | | | | 2.0 | cncer.2 | | dsmay.2 | miser.1 | distr.1 dpres.1 1.0 | raged.2 scaBF.1 scaBM.2 | angry.2 raged.1 angry.1 horif.1 o | distr.2 | dstrs.1 terif.1 | edge.2 0.0 | scaBM.1 sadBM.1 sadBM.2 oo | | heart.2 dstrs.2 sadBF.1 o | dpres.2 o | cncer.1 miser.2 elate.2 dsmay.1 -1.0 oo | comft.2 oooo | edge.1 oo | uplif.2 oooooooooooo | | -2.0 oooooooooo | ooooooo | heart.1 | good.2 ooooooo | oooooooooo | -3.0 | elate.1 oooooooo | uplif.1 happy.2 o | ooooooo | comft.1 o | -4.0 oooooo | | happy.1 o | ooo | o | good.1 -5.0 | o | | |  O              P     !  " # $% &  '  (.0/( Item Map – Mr Bean’s Holiday

RUMM2020 Project: READING Analysis: BFEEL99 Title: B - W/O CROSS, APPALLED. G/S SCARED, SAD Date: 16 Aug 2007 05:16:47 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------4.0 | | scaBF | o | | 3.0 | | | | | 2.0 | horif | terif | | sadBF | 1.0 | | angry raged distr o | scaBM cncer | dpres dsmay | miser 0.0 | dstrs sadBM oo | | edge o | o | -1.0 oo | oooo | oo | heart oooooooooooo | | elate -2.0 oooooooooo | ooooooo | comft | uplif ooooooo | oooooooooo | -3.0 | oooooooo | o | ooooooo | good o | happy -4.0 oooooo | | o | ooo | o | -5.0 | o | | | | -6.0 | o | | | | -7.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.0+(  O              P     !  " # $% &  '  (.00( Threshold Map – The Island

RUMM2020 Project: ALLMOVIEFA Analysis: IFEEL7 Title: W/O APPALLED, CROSS. GSPLIT SAD, SCARED Date: 28 Aug 2007 04:59:37 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------5.0 | | IMsca.2 | | IMsad.2 | 4.0 | | | | | 3.0 | | miser.2 | | | 2.0 | | dsmay.2 | horif.2 | depre.2 elate.2 | dstra.2 heart.2 1.0 | dstrs.2 comft.2 raged.2 IFsad.2 IFsca.2 uplif.2 | angry.2 | good.2 terif.2 o | happy.2 o | 0.0 o | miser.1 oooooo | cncer.2 o | depre.1 IMsca.1 dstra.1 ooooo | edge.2 oooooo | terif.1 -1.0 oooooo | dsmay.1 raged.1 elate.1 IMsad.1 oooo | horif.1 dstrs.1 ooooooo | oooooooooo | comft.1 ooooooooooo | -2.0 oooooooooo | IFsca.1 uplif.1 oooooooooooooo | ooooooooo | IFsad.1 heart.1 angry.1 ooooo | ooooooooooo | cncer.1 good.1 -3.0 ooooo | edge.1 happy.1 ooooo | o | ooo | oooo | -4.0 oo | | oooo | o | | -5.0 | oooooo | ooooo | | | -6.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.01( Item Map – The Island

RUMM2020 Project: ALLMOVIEFA Analysis: IFEEL7 Title: W/O APPALLED, CROSS. GSPLIT SAD, SCARED Date: 28 Aug 2007 04:59:44 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------3.0 | | | | | IMsca 2.0 | | IMsad | | miser | 1.0 | | | o | dsmay dstra depre o | horif elate 0.0 o | terif raged oooooo | comft dstrs o | uplif ooooo | IFsad heart IFsca oooooo | angry -1.0 oooooo | good oooo | happy ooooooo | oooooooooo | cncer ooooooooooo | edge -2.0 oooooooooo | oooooooooooooo | ooooooooo | ooooo | ooooooooooo | -3.0 ooooo | ooooo | o | ooo | oooo | -4.0 oo | | oooo | o | | -5.0 | oooooo | ooooo | | | -6.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.02(  O              P     !  " # $% &  '  (.03( Threshold Map – Babel ------LOCATION PERSONS ITEMS [uncentralised thresholds] ------6.0 | | | | Bscar.2 angry.2 | Gscar.2 depre.2 raged.2 horif.2 heart.2 5.0 | | | comft.2 | happy.2 uplif.2 | 4.0 | | | | | 3.0 | | | | | terif.2 2.0 | | | miser.2 | | dsmay.2 1.0 | | dstrs.2 appre.2 elate.2 | dstra.2 | good.2 cncer.2 o | Gsad.2 Bsad.2 0.0 | o | ooo | | elate.1 oo | -1.0 ooooooo | oo | ooo | ooooo | uplif.1 oooooo | terif.1 -2.0 oooooo | comft.1 Bscar.1 happy.1 oo | oooooo | good.1 ooooo | Gscar.1 raged.1 miser.1 ooo | heart.1 horif.1 depre.1 dstra.1 -3.0 oo | angry.1 ooo | oooooo | o | oo | -4.0 o | appre.1 | | dsmay.1 ooo | dstrs.1 | Gsad.1 -5.0 ooo | | | cncer.1 | | Bsad.1 -6.0 | | o | | ------o = 1 Person ------

 O              P     !  " # $% &  '  (.1,( Item Map – Babel RUMM2020 Project: MANLY Analysis: BFEEL98GS Title: BABEL FINAL FEELINGS GENDER SPLIT SAD SCARED Date: 10 Jan 2008 02:29:31 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------2.0 | | | Bscar | raged uplif | angry depre heart horif happy comft Gscar 1.0 | | | | o | terif elate 0.0 | o | ooo | miser | oo | -1.0 ooooooo | dstra good oo | ooo | ooooo | dsmay appre oooooo | -2.0 oooooo | dstrs oo | Gsad oooooo | cncer ooooo | ooo | Bsad -3.0 oo | ooo | oooooo | o | oo | -4.0 o | | | ooo | | -5.0 ooo | | | | | -6.0 | | o | | | -7.0 o | ------o = 1 Person ------

 O              P     !  " # $% &  '  (.1-(  O              P     !  " # $% &  '  (.1*( Threshold Map – Wild Hogs ------LOCATION PERSONS ITEMS [uncentralised thresholds] ------6.0 | | | | Mscar.2 angry.2 | Fscar.2 depre.2 raged.2 horif.2 heart.2 5.0 | | | comft.2 | happy.2 uplif.2 | 4.0 | | | | | 3.0 | | | | | terif.2 2.0 | | | miser.2 | | dsmay.2 1.0 | | dstrs.2 appre.2 elate.2 | dstra.2 | good.2 cncer.2 o | Fsad.2 Msad.2 0.0 | o | ooo | | elate.1 oo | -1.0 ooooooo | oo | ooo | ooooo | uplif.1 oooooo | terif.1 -2.0 oooooo | comft.1 Mscar.1 happy.1 oo | oooooo | good.1 ooooo | Fscar.1 raged.1 miser.1 ooo | heart.1 horif.1 depre.1 dstra.1 -3.0 oo | angry.1 ooo | oooooo | o | oo | -4.0 o | appre.1 | | dsmay.1 ooo | dstrs.1 | Fsad.1 -5.0 ooo | | | cncer.1 | | Msad.1 -6.0 | | o | | ------o = 1 Person ------

 O              P     !  " # $% &  '  (.1.( Item Map – Wild Hogs RUMM2020 Project: MANLY Analysis: WFEEL98GS Title: FINAL WH SOLUTION, SPLIT FOR GENDER SAD SCARED Date: 10 Jan 2008 02:22:51 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------2.0 | | | Mscar | raged uplif | angry depre heart horif happy comft Fscar 1.0 | | | | o | terif elate 0.0 | o | ooo | miser | oo | -1.0 ooooooo | dstra good oo | ooo | ooooo | dsmay appre oooooo | -2.0 oooooo | dstrs oo | Fsad oooooo | cncer ooooo | ooo | Msad -3.0 oo | ooo | oooooo | o | oo | -4.0 o | | | ooo | | -5.0 ooo | | | | | -6.0 | | o | | | -7.0 o | ------o = 1 Person ------

 O              P     !  " # $% &  '  (.1/( APPENDIX 5.20: FINAL AROUSAL DIMENSION – ALL MOVIES

 O              P     !  " # $% &  '  (.1+(  O              P     !  " # $% &  '  (.10( Threshold Map – Shooter

RUMM2020 Project: ALLMOVIEFA Analysis: SHOOTER44 Title: W/O SOBBED Date: 8 Nov 2007 11:41:27 AM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------6.0 | | | | tears.2 | 5.0 | | | | | 4.0 | | | | | 3.0 | | cried.1 | | | 2.0 | | | | unezy.2 | tears.1 jumpy.2 1.0 oo | o | | cried.2 anx.2 tense.2 ooo | grab.2 | 0.0 ooo | ooooooo | | calm.2 ooooooooooo | grab.1 ease.2 ooooooooooo | unezy.1 -1.0 | relax.2 ooooooooooooooo | ooooooooooooooo | | ooooooooooooooooo | jumpy.1 -2.0 oooooooooooooooooo | ease.1 | calm.1 anx.1 tense.1 relax.1 ooooooooo | | | -3.0 ooooooooo | | | ooooooooo | | -4.0 | | oooooooooo | | | -5.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.11( Item Map – Shooter

RUMM2020 Project: ALLMOVIEFA Analysis: SHOOTER44 Title: W/O SOBBED Date: 8 Nov 2007 11:42:21 AM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------4.0 | | | | | tears 3.0 | | | | | 2.0 | | | cried | | 1.0 oo | o | | ooo | unezy | 0.0 ooo | ooooooo | grab | jumpy ooooooooooo | ooooooooooo | anx tense -1.0 | ooooooooooooooo | ooooooooooooooo | calm ease | relax ooooooooooooooooo | -2.0 oooooooooooooooooo | | ooooooooo | | | -3.0 ooooooooo | | | ooooooooo | | -4.0 | | oooooooooo | | | -5.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.12(  O              P     !  " # $% &  '  (.13( Threshold Map – Spiderman

RUMM2020 Project: ALLMOVIEFA Analysis: SPIDER44 Title: W/O SOBBED Date: 8 Nov 2007 11:39:30 AM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------4.0 | | | | | 3.0 o | | | o | unezy.2 | 2.0 o | | jumpy.2 o | | tears.2 cried.2 o | tense.2 1.0 oo | | anx.2 ooo | grab.2 oo | cried.1 ooo | 0.0 | relax.2 calm.2 unezy.1 ooooooooo | ease.2 ooooooo | ooooooooooo | grab.1 oooooooooooooo | -1.0 oo | ooooooooooo | tears.1 ooooooooooooooo | tense.1 jumpy.1 anx.1 o | ease.1 oooooooooooooo | relax.1 -2.0 | calm.1 ooooooooo | o | oooooooooooo | | -3.0 | oooooooo | | | | -4.0 ooooooooo | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.2,( Item Map – Spiderman

RUMM2020 Project: ALLMOVIEFA Analysis: SPIDER44 Title: W/O SOBBED Date: 8 Nov 2007 11:40:22 AM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------4.0 | | | | | 3.0 o | | | o | | 2.0 o | | o | | o | unezy 1.0 oo | | cried ooo | oo | ooo | jumpy 0.0 | grab tears ooooooooo | anx tense ooooooo | ooooooooooo | oooooooooooooo | relax -1.0 oo | calm ease ooooooooooo | ooooooooooooooo | o | oooooooooooooo | -2.0 | ooooooooo | o | oooooooooooo | | -3.0 | oooooooo | | | | -4.0 ooooooooo | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.2-(  O              P     !  " # $% &  '  (.2*( Threshold Map – Perfect Stranger

RUMM2020 Project: ALLMOVIEFA Analysis: PERFECT44 Title: W/O SOBBED Date: 8 Nov 2007 11:36:44 AM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------6.0 | | | | | 5.0 | cried.2 | | | | 4.0 | | | | | 3.0 | | | | cried.1 | 2.0 | | jumpy.2 | | | tears.1 1.0 | tears.2 unezy.2 | anx.2 tense.2 o | grab.2 o | | 0.0 oo | oooo | | oooo | calm.2 oooooooooooo | relax.2 -1.0 | ease.2 unezy.1 grab.1 oooooooo | ooooooooo | | jumpy.1 ooooooooo | calm.1 -2.0 o | oooooo | relax.1 ease.1 anx.1 ooooooo | tense.1 | | -3.0 ooooooooo | | | ooo | | -4.0 | | oooooo | | | -5.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.2.( Item Map – Perfect Stranger

RUMM2020 Project: ALLMOVIEFA Analysis: PERFECT44 Title: W/O SOBBED Date: 8 Nov 2007 11:37:05 AM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------4.0 | | cried | | | 3.0 | | | | | 2.0 | | | | | tears 1.0 | | o | o | | 0.0 oo | jumpy unezy oooo | grab | oooo | anx oooooooooooo | tense -1.0 | oooooooo | calm ooooooooo | | ease relax ooooooooo | -2.0 o | oooooo | ooooooo | | | -3.0 ooooooooo | | | ooo | | -4.0 | | oooooo | | | -5.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.2/(  O              P     !  " # $% &  '  (.2+( Threshold Map – Mr Bean’s Holiday

RUMM2020 Project: ALLMOVIEFA Analysis: BEAN44 Title: W/O SOBBED Date: 8 Nov 2007 11:38:35 AM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------2.0 | | | | cried.1 | unezy.2 jumpy.2 1.0 | anx.2 | tense.2 | tears.2 unezy.1 o | grab.2 | tense.1 cried.2 0.0 ooo | ooo | relax.2 tears.1 oo | ooooo | ease.2 jumpy.1 grab.1 | calm.2 -1.0 ooooo | anx.1 ooooooo | o | oooooooo | relax.1 ooooooooooooo | ease.1 -2.0 | oooooooooooooo | calm.1 | | o | -3.0 oooooooooo | | | | oooooooooooo | -4.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.20( Item Map – Mr Bean’s Holiday

RUMM2020 Project: ALLMOVIEFA Analysis: BEAN44 Title: W/O SOBBED Date: 8 Nov 2007 11:38:53 AM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------1.0 | | cried unezy | tense o | | tears jumpy 0.0 ooo | grab anx ooo | oo | ooooo | | -1.0 ooooo | relax ooooooo | ease o | calm oooooooo | ooooooooooooo | -2.0 | oooooooooooooo | | | o | -3.0 oooooooooo | | | | oooooooooooo | -4.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.21(

 O              P     !  " # $% &  '  (.22( Threshold Map – The Island

RUMM2020 Project: ALLMOVIEFA Analysis: ISLAND44 Title: W/O SOBBED Date: 8 Nov 2007 11:43:00 AM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------3.0 | | cried.1 | | | 2.0 | o | | tears.1 unezy.2 oo | tears.2 o | cried.2 1.0 | jumpy.2 tense.2 oo | calm.2 oo | ooooo | anx.2 | grab.2 0.0 oooooo | ease.2 oooooooo | grab.1 | relax.2 oooooooooooooo | oooooooooo | unezy.1 -1.0 | ooooooooooooo | tense.1 | ooooooooooooo | jumpy.1 | -2.0 oooooooooooooo | | calm.1 anx.1 oooooooooooooo | relax.1 o | ease.1 oooooooooooooo | -3.0 | o | | oooooooooooo | | -4.0 | | ooooooo | | | -5.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.23( Item Map – The Island

RUMM2020 Project: ALLMOVIEFA Analysis: ISLAND44 Title: W/O SOBBED Date: 8 Nov 2007 11:43:11 AM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------3.0 | | | | | 2.0 | cried o | | oo | tears o | 1.0 | oo | oo | ooooo | unezy | 0.0 oooooo | grab oooooooo | tense | jumpy oooooooooooooo | oooooooooo | anx calm -1.0 | ooooooooooooo | | relax ease ooooooooooooo | | -2.0 oooooooooooooo | | oooooooooooooo | o | oooooooooooooo | -3.0 | o | | oooooooooooo | | -4.0 | | ooooooo | | | -5.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (.3,(  O              P     !  " # $% &  '  (.3-( Threshold Map – Babel RUMM2020 Project: MANLY Analysis: BAROUS9 Title: BABEL - AROUSAL - COLLAPSED, 9 ITEMS Date: 9 Jan 2008 04:35:04 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------7.0 | | | | | grab.2 6.0 | | | | | 5.0 | cried.2 | | | | 4.0 | | | | jumpy.2 | 3.0 | | | | | tears.2 2.0 | | o | | | tense.2 uneas.2 1.0 ooo | anx.2 | oo | | | cried.1 0.0 oooooo | | oooooooooo | ease.2 relax.2 o | grab.1 ooooooooooooooo | -1.0 | o | oooooooo | | ooooooo | jumpy.1 -2.0 | tears.1 | ooooo | o | oooo | -3.0 o | anx.1 | ease.1 ooo | uneas.1 tense.1 | | relax.1 -4.0 oooo | | | oo | | -5.0 | ------o = 1 Person

 O              P     !  " # $% &  '  (.3*( Item Map – Babel

RUMM2020 Project: MANLY Analysis: BAROUS9 Title: BABEL - AROUSAL - COLLAPSED, 9 ITEMS Date: 9 Jan 2008 04:35:15 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------3.0 | | grab | cried | | 2.0 | | o | | | 1.0 ooo | | jumpy oo | | | tears 0.0 oooooo | | oooooooooo | o | ooooooooooooooo | -1.0 | anx o | uneas tense oooooooo | | ooooooo | ease -2.0 | relax | ooooo | o | oooo | -3.0 o | | ooo | | | -4.0 oooo | | | oo | | -5.0 | ------o = 1 Person ------

 O              P     !  " # $% &  '  (.3.(  O              P     !  " # $% &  '  (.3/( Threshold Map – Wild Hogs

RUMM2020 Project: MANLY Analysis: WAROUS9 Title: WILD HOGS - AROUSAL - COLLAPSED, 9 ITEMS Date: 9 Jan 2008 04:35:37 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------5.0 | | | jumpy.2 | uneas.2 | 4.0 | | cried.2 | | | 3.0 | tears.2 | | | | 2.0 | | | | | 1.0 | tense.2 | | | anx.2 | grab.2 0.0 | o | cried.1 o | o | | -1.0 | ease.2 o | relax.2 | jumpy.1 tears.1 oo | tense.1 uneas.1 | grab.1 -2.0 oooooo | anx.1 o | oooo | | oooooooooooooooooo | -3.0 | | ooooooooooooooooooo | ease.1 o | | relax.1 -4.0 ooooooooooooooooo | | | | | -5.0 ooooooooooooo | ------o = 1 Person ------

 O              P     !  " # $% &  '  (.3+( Item Map – Wild Hogs

RUMM2020 Project: MANLY Analysis: WAROUS9 Title: WILD HOGS - AROUSAL - COLLAPSED, 9 ITEMS Date: 9 Jan 2008 04:36:15 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------2.0 | | cried | jumpy | uneas | 1.0 | | tears | | | 0.0 | o | tense o | o | | anx grab -1.0 | o | | oo | | -2.0 oooooo | o | ease oooo | relax | oooooooooooooooooo | -3.0 | | ooooooooooooooooooo | o | | -4.0 ooooooooooooooooo | | | | | -5.0 ooooooooooooo | ------o = 1 Person ------

 O              P     !  " # $% &  '  (.30( APPENDIX 5.21: FINAL APPRAISAL DIMENSION – ALL MOVIES

 O              P     !  " # $% &  '  (.31(  O              P     !  " # $% &  '  (.32( Threshold Map – Shooter

RUMM2020 Project: READING Analysis: XLIK6 Title: SHOOTER - LIKE W/O XLIK2, MLIK1 Date: 24 Aug 2007 02:08:48 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------6.0 | | ooo | | | 5.0 | | | ooo | | 4.0 | Mlik3.3 Mlik2.3 | ooo | | | 3.0 ooo | | Xlik1.3 o | inter.3 ooooooo | immer.3 | 2.0 | | wait.3 oooooooo | | | 1.0 oooooooooooooo | | | Mlik3.2 ooooooooooooooooooo | | 0.0 | oooooooooo | | Mlik2.2 oooooo | | immer.2 -1.0 oooooo | | Mlik3.1 o | | Xlik1.2 oo | -2.0 | wait.2 | immer.1 Mlik2.1 wait.1 | inter.2 Xlik1.1 | inter.1 o | -3.0 | ------o = 3 Persons ------

 O              P     !  " # $% &  '  (.33( Item Map – Shooter

RUMM2020 Project: READING Analysis: XLIK6 Title: SHOOTER - LIKE W/O XLIK2, MLIK1 Date: 24 Aug 2007 02:09:18 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------6.0 | | ooo | | | 5.0 | | | ooo | | 4.0 | | ooo | | | 3.0 ooo | | o | ooooooo | | 2.0 | | oooooooo | | | Mlik3 1.0 oooooooooooooo | | | ooooooooooooooooooo | Mlik2 | 0.0 | oooooooooo | immer | Xlik1 oooooo | | inter wait -1.0 oooooo | | o | | oo | -2.0 | | | | o | -3.0 | ------o = 3 Persons ------

 O              P     !  " # $% &  '  (/,,(  O              P     !  " # $% &  '  (/,-( Threshold Map – Spiderman

RUMM2020 Project: READING Analysis: SLIKE6 Title: SPIDERMAN - LIKE W/O XLIK2, MLIK1 Date: 24 Aug 2007 ------LOCATION PERSONS ITEMS [uncentralised thresholds] ------7.0 | | o | | | 6.0 | | oooo | | | Mlik3.3 Mlik2.3 5.0 | ooo | | | | 4.0 ooooo | | immer.3 | inter.3 ooooooo | Xlik1.3 | 3.0 | | oooooooo | | wait.3 | 2.0 | oooooooooooooo | | | | Mlik3.2 1.0 | ooooooooooo | | o | | Mlik2.2 0.0 ooooooooo | | | oooooo | immer.2 | -1.0 | ooo | Xlik1.2 | | ooooo | wait.2 -2.0 | Mlik3.1 inter.2 ooo | | | o | Mlik2.1 -3.0 | oo | | | | Xlik1.1 -4.0 | inter.1 | immer.1 | | wait.1 | -5.0 |

------o = 3 Persons

 O              P     !  " # $% &  '  (/,*( Item Map – Spiderman

RUMM2020 Project: READING Analysis: SLIKE6 Title: SPIDERMAN - LIKE W/O XLIK2, MLIK1 Date: 24 Aug 2007 ------LOCATION PERSONS ITEMS [locations] ------7.0 | | o | | | 6.0 | | oooo | | | 5.0 | ooo | | | | 4.0 ooooo | | | ooooooo | | 3.0 | | oooooooo | | | 2.0 | oooooooooooooo | | | Mlik3 | 1.0 | ooooooooooo | Mlik2 | o | | 0.0 ooooooooo | | immer | Xlik1 oooooo | | inter -1.0 | ooo | | wait | ooooo | -2.0 | ooo | | | o | -3.0 | oo | | | | -4.0 | |

------o = 3 Persons ------

 O              P     !  " # $% &  '  (/,.(  O              P     !  " # $% &  '  (/,/( Threshold Map – Perfect Stranger

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------7.0 | o | | | | 6.0 | | Mlik3.3 | | oo | 5.0 | | o | | oo | Mlik2.3 4.0 | | | | ooo | 3.0 | inter.3 | ooo | immer.3 Xlik1.3 | | 2.0 ooooo | wait.3 | | ooooooooo | | Mlik3.2 1.0 o | oooooooo | | Mlik2.2 | o | 0.0 ooooooooooooooooo | | | ooooooooooooo | | Mlik3.1 -1.0 oooooo | | Xlik1.2 | immer.2 oooo | | -2.0 oo | inter.2 wait.2 | oo | Mlik2.1 | oo | Xlik1.1 -3.0 | immer.1 o | | o | o | -4.0 | inter.1 wait.1 | | o | | -5.0 | | o | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (/,+( Item Map – Perfect Stranger ------LOCATION PERSONS ITEMS [locations] ------7.0 | o | | | | 6.0 | | | | oo | 5.0 | | o | | oo | 4.0 | | | | ooo | 3.0 | | ooo | | | 2.0 ooooo | Mlik3 | | ooooooooo | | 1.0 o | oooooooo | Mlik2 | | o | 0.0 ooooooooooooooooo | | | Xlik1 ooooooooooooo | immer | -1.0 oooooo | inter | | wait oooo | | -2.0 oo | | oo | | oo | -3.0 | o | | o | o | -4.0 | | | o | | -5.0 | | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (/,0(  O              P     !  " # $% &  '  (/,1( Threshold Map – Mr Bean’s Holiday

RUMM2020 Project: READING Analysis: BLIKE6 Title: BEAN - LIKE W/O XLIK2, MLIK1 Date: 24 Aug 2007 02:11: ------LOCATION PERSONS ITEMS [uncentralised thresholds] ------6.0 | | o | | | 5.0 | | | | | 4.0 | Mlik2.3 inter.3 oo | | | oooo | Mlik3.3 3.0 | | Xlik1.3 immer.3 ooooooo | | | 2.0 ooo | wait.3 | | | ooooooooooooo | 1.0 | | ooooooooooooooo | | Mlik2.2 | Mlik3.2 0.0 ooooooooo | o | oooooooooo | immer.2 | ooooooo | -1.0 | | wait.2 inter.2 ooo | | Mlik3.1 oooo | Xlik1.2 -2.0 | ooo | Mlik2.1 | o | | inter.1 -3.0 o | | wait.1 immer.1 | o | Xlik1.1 | -4.0 | | | | | -5.0 | o | | | | -6.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (/,2( Item Map – Mr Bean’s Holiday RUMM2020 Project: READING Analysis: BLIKE6 Title: BEAN - LIKE W/O XLIK2, MLIK1 Date: 24 Aug 2007 ------LOCATION PERSONS ITEMS [locations] ------6.0 | | o | | | 5.0 | | | | | 4.0 | oo | | | oooo | 3.0 | | ooooooo | | | 2.0 ooo | | | | ooooooooooooo | 1.0 | | Mlik2 ooooooooooooooo | Mlik3 | | 0.0 ooooooooo | inter o | immer oooooooooo | | ooooooo | Xlik1 wait -1.0 | | ooo | | oooo | -2.0 | ooo | | o | | -3.0 o | | | o | | -4.0 | | | | | -5.0 | o | | | | -6.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (/,3(  O              P     !  " # $% &  '  (/-,( Threshold Map – The Island

RUMM2020 Project: ISLANDPC Analysis: LIKING6 Title: LIKING W/O XLIK2, MLIK1 Date: 23 Aug 2007 10:51:34 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------6.0 | | | oo | | 5.0 | | | | oo | 4.0 | Mlik3.3 | Mlik2.3 o | oooo | | 3.0 | | ooooooooooo | imers.3 | ooooooooooo | Xlik1.3 2.0 oo | | ooooooooooooo | | inter.3 oooooooooooooooo | Mlik3.2 1.0 o | wait.3 Mlik2.2 | oooooooooooooo | | oooooooooooooooooo | 0.0 | imers.2 oooooooooooo | | | ooooooo | -1.0 | wait.2 ooooooo | Xlik1.2 Mlik3.1 | oooooooo | Mlik2.1 | -2.0 ooo | | inter.2 oooo | | wait.1 imers.1 oo | Xlik1.1 -3.0 | | oo | | inter.1 | -4.0 oo | | | | ooo | -5.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (/--( Item Map – The Island

RUMM2020 Project: ISLANDPC Analysis: LIKING6 Title: LIKING W/O XLIK2, MLIK1 Date: 23 Aug 2007 10:51:43 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------6.0 | | | oo | | 5.0 | | | | oo | 4.0 | | o | oooo | | 3.0 | | ooooooooooo | | ooooooooooo | 2.0 oo | | ooooooooooooo | | Mlik3 oooooooooooooooo | Mlik2 1.0 o | | oooooooooooooo | | oooooooooooooooooo | 0.0 | imers oooooooooooo | | | Xlik1 ooooooo | wait -1.0 | ooooooo | | inter oooooooo | | -2.0 ooo | | oooo | | oo | -3.0 | | oo | | | -4.0 oo | | | | ooo | -5.0 | ------o = 2 Persons ------

 O              P     !  " # $% &  '  (/-*(  O              P     !  " # $% &  '  (/-.( Threshold Map – Babel

RUMM2020 Project: MANLY Analysis: BPLEAS98 Title: BABEL - OPTIMAL PLEASURE Date: 9 Jan 2008 04:36:45 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------4.0 | | | | Mlik3.4 Mlik2.4 o | 3.0 | oo | Xlik2.4 wait.4 | | ooo | 2.0 | oooooo | | Mlik3.2 oooooo | Mlik3.3 | immer.4 1.0 oooo | inter.4 | Mlik2.3 ooooooooo | ooooooooo | | 0.0 oooooooo | ooooooo | | Mlik2.2 oooooo | Xlik2.2 wait.3 immer.2 ooooo | Mlik3.1 -1.0 ooo | Xlik2.3 wait.1 oo | o | inter.3 | inter.1 Mlik2.1 wait.2 o | immer.3 -2.0 | o | | Xlik2.1 inter.2 | immer.1 o | -3.0 | ------o = 1 Person ------

 O              P     !  " # $% &  '  (/-/( Item Map – Babel

RUMM2020 Project: MANLY Analysis: BPLEAS98 Title: BABEL - OPTIMAL PLEASURE Date: 9 Jan 2008 04:36:58 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------4.0 | | | | o | 3.0 | oo | | | ooo | 2.0 | oooooo | | oooooo | Mlik3 | 1.0 oooo | | ooooooooo | Mlik2 ooooooooo | | 0.0 oooooooo | wait ooooooo | | Xlik2 oooooo | ooooo | -1.0 ooo | immer oo | inter o | | o | -2.0 | o | | | o | -3.0 | ------o = 1 Person ------

 O              P     !  " # $% &  '  (/-+(  O              P     !  " # $% &  '  (/-0( Threshold Map – Wild Hogs

RUMM2020 Project: MANLY Analysis: WPLEAS98 Title: WH - OPTIMAL PLEASURE Date: 9 Jan 2008 04:37:22 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [uncentralised thresholds] ------5.0 | | | | oo | 4.0 | | o | | Mlik3.4 Mlik2.4 | wait.4 3.0 o | inter.4 | o | immer.4 | | 2.0 oooo | | ooooooo | Xlik2.4 | Mlik2.3 | 1.0 oooooooo | ooo | | Mlik3.2 ooooooooo | immer.3 oooooo | wait.3 0.0 | ooo | Mlik3.3 oooo | ooooooooo | | Xlik2.3 -1.0 ooooooo | oooo | inter.3 Mlik2.2 ooooo | Mlik2.1 wait.2 Mlik3.1 immer.1 immer.2 | inter.2 ooo | Xlik2.2 -2.0 oo | wait.1 inter.1 o | | | | -3.0 oo | | | | o | -4.0 | Xlik2.1 | | | oo | -5.0 | ------o = 1 Person ------

 O              P     !  " # $% &  '  (/-1( Item Map – Wild Hogs

RUMM2020 Project: MANLY Analysis: WPLEAS98 Title: WH - OPTIMAL PLEASURE Date: 9 Jan 2008 04:37:38 PM Display: ITEM MAP

------LOCATION PERSONS ITEMS [locations] ------5.0 | | | | oo | 4.0 | | o | | | 3.0 o | | o | | | 2.0 oooo | | ooooooo | | | 1.0 oooooooo | ooo | | Mlik2 Mlik3 ooooooooo | oooooo | 0.0 | wait immer ooo | oooo | ooooooooo | inter | -1.0 ooooooo | oooo | Xlik2 ooooo | | ooo | -2.0 oo | o | | | | -3.0 oo | | | | o | -4.0 | | | | oo | -5.0 | ------o = 1 Person ------

 O              P     !  " # $% &  '  (/-2( APPENDIX 5.22: STAGE 4 – ALL COGNITIVE EFFORT ITEMS

 O              P     !  " # $% &  '  (/-3(  O              P     !  " # $% &  '  (/*,( APPENDIX 5.23: FINAL COGNITIVE EFFORT DIMENSION (THE ISLAND ONLY)

 O              P     !  " # $% &  '  (/*-(  O              P     !  " # $% &  '  (/**( Threshold Map – The Island

RUMM2020 Project: ISLANDPC Analysis: COGN2 Title: ALL COG W/O DIFF3 Date: 23 Aug 2007 ------LOCATION PERSONS ITEMS [uncentralised thresholds] ------5.0 | | | | | 4.0 | | | | | diff1.3 3.0 | attn2.3 | easy2.3 | attn3.3 | diff2.3 o | easy1.3 2.0 | ooo | diff2.2 | attn1.3 ooooo | | 1.0 ooooooo | | easy2.2 ooooooo | o | ooooooo | easy1.2 0.0 | attn3.2 ooooo | | diff1.2 oooooooooooooo | | attn2.2 -1.0 | ooooooooooo | | ooooo | | attn1.2 easy2.1 -2.0 oooooo | diff2.1 | oooooo | attn3.1 | easy1.1 oooo | diff1.1 -3.0 | oo | | | attn2.1 ooo | -4.0 | | attn1.1 | oo | | -5.0 | | o | | | -6.0 | ------o = 3 Persons ------

 O              P     !  " # $% &  '  (/*.( Item Map – The Island

RUMM2020 Project: ISLANDPC Analysis: COGN2 Title: ALL COG W/O DIFF3 Date: 23 Aug 2007

------LOCATION PERSONS ITEMS [locations] ------5.0 | | | | | 4.0 | | | | | 3.0 | | | | o | 2.0 | ooo | | ooooo | | 1.0 ooooooo | | diff2 ooooooo | easy2 o | ooooooo | 0.0 | easy1 diff1 attn3 ooooo | | attn2 oooooooooooooo | | -1.0 | ooooooooooo | | attn1 ooooo | | -2.0 oooooo | | oooooo | | oooo | -3.0 | oo | | | ooo | -4.0 | | | oo | | -5.0 | | o | | | -6.0 | ------o = 3 Persons ------

 O              P     !  " # $% &  '  (/*/( APPENDIX 5.24: HOLISTIC PLOTS

Location for all Feeling items, per movie

4 IFsca

3.5

3

2.5 IMsca horif 2 terif IMsca IMs ad

1.5 horif mis er IFsad IMsca mis er depre dstra depre 1 raged dstra angry raged terif mis er dsmay angry raged cncer dstrs dstra depre IMsca angry dsmay depre 0.5 dsmay dstra depre IMs ad IFsad elate horif mis er happy horif dstrs IFsad IMsca elate dsmay comft heart uplif raged cncer dstrs terif IMs ad IFsca 0 dstrs dstra heart angry terif comft IFsca edge uplif dstrs uplif IMs ad IFsca -0.5 good elate horif comft heart IFsad happy angry heart IFsad cncer -1 good cncer comft uplif happy edge heart IFsca good cncer -1.5 edge edge elate edge -2 happy comft good uplif -2.5

-3

-3.5 good happy

-4

SHOOTER SPIDERMAN PERFECT MR BEAN ISLAND

 O              P     !  " # $% &  '  (/*+( Location for all Arousal items, per movie

4 cried

3.5 tears

3

2.5

2 cried

cried

tears 1.5 unezy tears

1 cried unezy cried

tense 0.5 unezy unezy jumpy tears tears jumpy unezy anx grab 0 tense grab anx jumpy

-0.5 anx calm tense tense anx relax anx ease -1 calm ease calm ease calm calm relax relax -1.5 ease relax

-2

SHOOTER SPIDERMAN PERFECT MR BEAN IS LA ND

 O              P     !  " # $% &  '  (/*0( Location for all Appraisal items, per movie

2

Mlik3 Mlik3

1.5 Mlik3

Mlik2 Mlik3

Mlik2 1 Mlik3 Mlik2

Mlik2 Mlik2

0.5

inter imers 0

imers Xlik1 Xlik1 imers Xlik1 -0.5 Xlik1

inter wait inter wait inter -1 wait wait

wait Xlik1 inter

-1.5

SHOOTER SPIDERMAN PERFECT MR BEAN ISLAND

 O              P     !  " # $% &  '  (/*1( APPENDIX 5.25: BEST CASE SCENARIO PLOTS

Item locations for Shooter and The Island were plotted against each other, as denoted by the dots.

These positions were examined to see whether they fell within 95% or 99% confidence intervals.

The Island x Shooter - feelings

2.5

2

1.5

1

0.5

0 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 -0.5

-1

-1.5

-2

-2.5

 O              P     !  " # $% &  '  (/*2( The Island x Shooter - arousal

4

3.5

3

2.5

2

1.5

1

0.5

0 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 -0.5

-1

-1.5

-2

-2.5

The Island x Shooter - appraisal

2

1.5

1

0.5

0 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2

-0.5

-1

-1.5

 O              P     !  " # $% &  '  (/*3( APPENDIX 5.26: WORST CASE SCENARIO PLOTS

Item locations for Perfect Stranger and Mr Bean’s Holiday were plotted against each other, as denoted by the dots.

These positions were examined to see whether they fell within 95% or 99% confidence intervals.

Perfect Stranger x Mr Bean's Holiday - feelings

3

2

1

0 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2

-1

-2

-3

-4

 O              P     !  " # $% &  '  (/.,( Perfect Stranger x Mr Bean's Holiday - arousal

3

2.5

2

1.5

1

0.5

0 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 -0.5

-1

-1.5

-2

Perfect Stranger x Mr Bean's Holiday - appraisal

2

1.5

1

0.5

0 -1.5-1-0.500.5 11.5 22.5

-0.5

-1

-1.5

 O              P     !  " # $% &  '  (/.-( APPENDIX 6.1: LETTERS TO SCHOOLS, PARENTS AND PARTICIPANTS

THE UNIVERSITY OF NEW SOUTH WALES 

 Dr Jennifer Harris Senior Lecturer School of Marketing Phone: 02.9385.1823 Fax: 02.9663.1985 Email: [email protected]

XXX Principal, XXXXXXXXXXXX XXXXXXXXXXXX

Dear XXXX

Thank you for your expression of interest in our study and a willingness to learn more about your school’s possible involvement.

As stated in our conversation on XXXXXX, the study, endorsed and supported by the School of Marketing at the University of New South Wales is an investigation of the impact of product placement on vulnerable audiences. Product placement is the practice of using brands and/or products as part of the story line in entertainment programs. It is now a multi-billion dollar industry and the practice has increased based on the assumption that, unlike advertisements, the exposure by-passes our critical evaluation and as such potentially increases not only the perceived value of products such as cigarettes, alcohol and illicit drugs, but also the brand/product displayed in the entertainment program. Within marketing, ethical marketers are particularly concerned about this practice. However, regulation is conditional on establishing that there is a link between product placement and the use of these brands/products.

Our research seeks to determine what characteristics of an audience member may result in them noticing and remembering a given product placement. We are particularly interested in the individual’s adopted learning style, their level of engagement with the entertainment program, and their present level of product and brand familiarity, and investigating how these factors impact how teenage audiences process the information and storylines contained within films.

 O              P     !  " # $% &  '  (/.*( To proceed with this study we now need the involvement of some schools, so as to access these young adults. The Catholic Education Office, Sydney, has approved this study and our request to contact your school for possible involvement (approval letter attached). Participation in this study by your school is purely voluntary, as is the participation of your students, should you accept. Ideally, we are seeking students in years 10 and/or 11.

We are requesting your involvement in two stages:

Firstly, students will be asked to complete a short questionnaire indicating their familiarity with a number of brands. This should only take 10-15 minutes. Ideally we would like this to take place in early June.

The second stage would be conducted 2-3 weeks later (i.e., mid-late June). At this time students will be asked to watch an appropriately rated movie selected by the researcher and approved by the School. After watching the movie, they will be asked to fill in a questionnaire which will capture how they felt about the movie. Did they enjoy it? Were they paying full attention? Were they bored and distracted? Which specific brands did they remember seeing in the film they just watched? It is anticipated that this stage will take approximately two hours (including the screening of the film).

If it is possible to factor this research into a suitable learning environment (for example, English or General Studies classes), we would be more than willing to liaise with your teachers in order to maximise the potential learning outcomes for your students. Alternatively, the screening of the film could be offered as an end-of-term or end-of- exam treat.

We are willing to abide by any regulations in regard to participating in this research, in particular, the avenues used in order to gain consent from the student and/or parent. We are also willing to bear the expense of any correspondence to gain this consent.

We would appreciate your confirmation of participation in this study at your earliest convenience. Should you have any further questions concerning this study, or wish to receive a summary of the research findings, please do not hesitate to contact Jennifer Harris (details below)

Yours sincerely,

Jennifer Harris Margaret Craig-Lees Jane Scott

Dr Jennifer Harris Dr Margaret Craig-Lees PhD Candidate Senior Lecturer Associate Professor School of Marketing School of Marketing School of Marketing & Advertising University of New South Wales University of New South Wales Auckland University of Technology Sydney 2052 Sydney 2052 Auckland 1020 AUSTRALIA AUSTRALIA NEW ZEALAND Ph: 9385 1823

 O              P     !  " # $% &  '  (/..( Approval No (when available)

[Use letterhead paper for page 1]

THE UNIVERSITY OF NEW SOUTH WALES

PARENTAL (OR GUARDIAN) INFORMATION STATEMENT

Information Processing in Films: A Teenage Perspective

Invitation and purpose of study You are invited to permit your child to participate in a study of how teenage audiences process the information and storylines contained within films. We hope to learn about the possible influence that films may have over their audiences. Your child was selected as a possible participant in this study because we are trying to engender the support of a cross-section of Sydney teenagers aged 12-16.

Description of study and risks If you decide to permit your child to participate, we will show your child a movie and then ask them to answer some questions about whether you enjoyed it or not and how it made them feel. This should take approximately 2 hours.

Confidentiality and disclosure of information Any information that is obtained in connection with this study and that can be identified with you will remain confidential and will be disclosed only with your permission, except as required by law. If you give us your permission by signing this document, we plan to publish the results in an appropriate marketing, sociological or communications-related journal or conference. In any publication, information will be provided in such a way that you cannot be identified.

Recompense to Parent/Guardian of Participants Complaints may be directed to the Ethics Secretariat, The University of New South Wales, SYDNEY 2052 AUSTRALIA (phone 9385 4234, fax 9385 6648, email [email protected]). Any complaint you make will be treated in confidence and investigated, and you will be informed of the outcome.

Your consent Your decision whether to not to permit your child to participate will not prejudice you or your child’s future relations with The University of New South Wales. If you decide to permit your child to participate, you are free to withdraw your consent and to discontinue your child’s participation at any time without prejudice.

If you have any questions, please feel free to ask us. If you have any additional questions later, Dr Jennifer Harris, phone 9385 1823 or Miss Jane Scott, phone 0413 802 104 will be happy to answer them.

You will be given a copy of this form to keep.

 O              P     !  " # $% &  '  (/./( THE UNIVERSITY OF NEW SOUTH WALES

PARENTAL (OR GUARDIAN) INFORMATION STATEMENT (continued)

Information Processing in Films: A Teenage Perspective

You are making a decision whether or not to permit your child to participate. Your signature indicates that, having read the attached Parental (or Guardian) Information Statement, you have decided to permit your child to take part in the study.

…………………………………………………… .……………………………………………………. Signature of Parent/Guardian Signature of Witness

…………………………………………………… .……………………………………………………. Please PRINT name Please PRINT name

…………………………………………………… .……………………………………………………. Date Nature of Witness

…………………………………………………… .……………………………………………………. Signature(s) of Investigator(s)

…………………………………………………… Please PRINT Name

REVOCATION OF CONSENT BY PARENT (OR GUARDIAN) Information Processing in Films: A Teenage Perspective

I hereby wish to WITHDRAW my consent for my child/ward to participate in the research proposal described above and understand that such withdrawal WILL NOT jeopardise any treatment, or my child/ward’s relationship, with The University of New South Wales

…………………………………………………… .……………………………………………………. Signature Date

…………………………………………………… Please PRINT Name

The section for Revocation of Consent by the parent/guardian should be forwarded to Miss Jane Scott, School of Marketing, Level 3 John Goodsell Building, UNSW 2052.

 O              P     !  " # $% &  '  (/.+( [Use letterhead paper for page 1]

Approval No (when available)

THE UNIVERSITY OF NEW SOUTH WALES

PARTICIPANT INFORMATION STATEMENT AND CONSENT FORM

Research Project: Information Processing in Films: A Teenage Perspective

Invitation and purpose of study You are invited to participate in a study of film preferences. We hope to learn more about what films you do and do not like and better understand the different ways you process information in films. You were selected as a possible participant in this study because we are particularly interested in the film preferences of teenagers.

Description of study and risks If you decide to participate, we will show you a movie and then ask you to answer some questions about whether you enjoyed it or not and how it made you feel. This should take approximately 2 hours.

Confidentiality and disclosure of information Any information that is obtained in connection with this study and that can be identified with you will remain confidential and will be disclosed only with your permission, except as required by law. If you give us your permission by signing this document, we plan to publish the results in an appropriate marketing, sociological or communications-related journal or conference. In any publication, information will be provided in such a way that you cannot be identified.

Complaints may be directed to the Ethics Secretariat, The University of New South Wales, SYDNEY 2052 AUSTRALIA (phone 9385 4234, fax 9385 6648, email [email protected]). Any complaint you make will be treated in confidence and investigated, and you will be informed of the outcome.

Your consent Your decision whether or not to participate will not prejudice your future relations with The University of New South Wales. If you decide to participate, you are free to withdraw your consent and to discontinue participation at any time without prejudice.

If you have any questions, please feel free to ask us. If you have any additional questions later, (Dr Jennifer Harris, phone 9385 1823 or Miss Jane Scott, phone 0413 802 104) will be happy to answer them.

You will be given a copy of this form to keep.

 O              P     !  " # $% &  '  (/.0( THE UNIVERSITY OF NEW SOUTH WALES

PARTICIPANT INFORMATION STATEMENT AND CONSENT FORM (continued) Information Processing in Films: A Teenage Perspective

You are making a decision whether or not to participate. Your signature indicates that, having read the Participant Information Statement, you have decided to take part in the study.

…………………………………………………… .……………………………………………………. Signature of Research Participant Signature of Witness

…………………………………………………… .……………………………………………………. (Please PRINT name) (Please PRINT name)

…………………………………………………… .……………………………………………………. Date Nature of Witness

…………………………………………………… Signature(s) of Investigator(s)

.……………………………………………………. Please PRINT Name

REVOCATION OF CONSENT Information Processing in Films: A Teenage Perspective

I hereby wish to WITHDRAW my consent to participate in the research proposal described above and understand that such withdrawal WILL NOT jeopardise any treatment or my relationship with The University of New South Wales.

…………………………………………………… .……………………………………………………. Signature Date

…………………………………………………… Please PRINT Name

The section for Revocation of Consent should be forwarded to Miss Jane Scott, School of Marketing, Level 3 John Goodsell Building, UNSW 2052.

 O              P     !  " # $% &  '  (/.1( APPENDIX 6.2: SURVEY TO MEASURE FILM AND STAR PREFERENCES



About You

How old are you? ______

Are you Male OR Female

Movie preferences

Please list your five favourite movies (in any order): i) ______ii) ______iii) ______iv) ______v) ______

What 3 types of movie do you most enjoy watching from the following list - comedy, action, thriller, historical, romantic comedy, horror, foreign, drama, romance, animated, classic, documentary, musical, teen, war, science fiction/fantasy, western, sporting, other. Please circle your 3 choices.

What 3 types of movie bore you most from the following list - comedy, action, thriller, historical, romantic comedy, horror, foreign, drama, romance, animated, classic, documentary, musical, teen, war, science fiction/fantasy, western, sporting, other.

Please circle your 3 choices.

 O              P     !  " # $% &  '  (/.2( Now can you please list your 3 favourite movie stars (in any order): i) ______ii) ______iii) ______

When you are watching a movie or TV show that you really like, what kinds of words explain the way that you feel? How do you behave? What do you do? ______

______

What about when you go to the movies and you just can’t get into the movie you are watching, or you are at a friend’s house and they are watching something you are not interested in. How does this make you feel? How do you behave then? What words describe these feelings and behaviours? ______

______

If you were watching a movie that scared you or made you feel anxious, what kinds of words would describe how you were feeling? What words would describe your behaviour? ______

______

What about if you were watching a movie with a really complicated plot? Where you had to watch every second of the movie otherwise you might lose track of the story? Maybe the story has lots of detail and information to process or there are lots of characters? How would you feel then? How would you behave? ______

______

 O              P     !  " # $% &  '  (/.3( You watch a movie and give it 10/10. What words describe how you felt watching the movie? How did you behave when you were watching it? ______

______

______

You watch a movie and think it is the worst you have ever seen. What words describe how you felt watching the movie? How did you behave when you were watching it? ______

______

______

You are watching a movie that is really exciting – maybe a thriller – and you don’t know what is going to happen next. What words describe how you are feeling? How do you behave? ______

______

______

You watch a movie that gives you warm fuzzies. The story is really uplifting, maybe even inspirational. It may have even changed the way you think about certain things. What words describe how you would feel when watching this movie? How would you behave? ______

______

______

 O              P     !  " # $% &  '  (//,( Word Association Test Engrossment:

• to occupy exclusively; to be absorbed • being “really into” the story and concentrating completely • to be completely consumed by the movie or program that you are watching

After reading these definitions of the word engrossment, what other words come to mind?

______

______

______

Can you think of 2 movies that you found “engrossing”? i) ______ii) ______

 O              P     !  " # $% &  '  (//-( APPENDIX 6.3: DETAILED CONTENT ANALYSIS PROCEDURE

Construction of the Film Coding Form Central to this movie evaluation was the construction of a film coding form (see Appendix 6.4). This was designed by the researcher and was filled in by the coders whilst watching a film in order to assist them in identifying the necessary aspects of the product placement as related to the research hypotheses and to maximise the objectivity, standardisation and depth of the analysis. Coding instructions were also given to provide the coders with a structured guide to assist them in their coding (see Appendix 6.4). Guidelines defining the specificity of the response were made clear to the coders both during training, and on the coding instruction sheet.

Kaid and Wadsworth (1989) suggest that no step in the content analysis is more crucial than the formulation of categories and their units of analysis. Bowers (1970, in Kaid and Wadsworth, 1989) suggests that definitions of categories be as explicit as possible so that coders know exactly how to use each category. By clearly specifying what is to be included and excluded, the researcher may reduce disagreements among coders from inadequate definition of categories. The categories formulated by the researcher were therefore carefully devised to represent the concepts embodied in the research hypotheses.

For every category or set of categories, a unit of analysis was also selected. Following Krippendorf’s (1980) methods of defining a unit, this content analysis sought to describe thematic units (recurring elements). Closely related to the unit of analysis is the unit of enumeration. In this case, this was simply a frequency count of the classifications of a category.

This coding form underwent three major changes resulting from actual use and feedback from the coders. The end result was a highly usable form which effectively captured all the necessary information.

 O              P     !  " # $% &  '  (//*( The most important piece of information to be captured by the content analysis was of the actual brands that appeared in the film. Then, specifically to each of these brands, the total number of exposures it received in a given scene was recorded, as was the length of each of these exposures.

The next part of the coding form underwent the most revisions, and was by far the most subjective measure on the form. This was to do with deciding on the prominence of the placement within a scene. Originally, this section was constructed using the types of product placements identified by DeLorme and Reid (1999) – namely, that a product placement may occur as a logo, as an advertisement in the background of a scene, or by having the actual product appear in a scene. However, a better distinction between types of product placements was later sourced by Murdock (1992) who defined product placements as either on-set (when a product is displayed more prominently in a given shot) or creative (when a product appears in the background of a shot). DeLorme and Reid’s (1999) definitions of product placements were collapsed into two measures, “logo / ad / signage” and “actual brand shown/spoken) as these were still deemed important types of placements to be captured and analysed separately.

Each product placement was assessed as to its connection to the plot, in regards to this connection being high or low. High plot connection was awarded to a product placement that was highly integrated with the plot, and without which, the scene could not have existed.

Addressing a limitation from research conducted by Brennan, Dubas and Babin (1999), placements were further analysed to see whether the product was consumed or mentioned by one of the lead actors.

Finally, each product placement was to be categorised as either audio only, visual only, or audio-visual.

 O              P     !  " # $% &  '  (//.( Training of the Coders Following the recommendation of Kaid and Wadsworth (1989), a structured training session was provided for the coders whereby the procedures for the content analysis were outlined, and the coding instrument explained. The objectives of the study were also detailed as was the importance of the content analysis to the overall study. Indeed, the accuracy and dependability of the coding process in content analysis is greatly influenced by the selection and training of the coders (Kaid and Wadsworth 1989). Training coders is particularly important to objectivity because it increases their familiarity with the coding scheme and operational definitions, thereby improving inter- coder and intra-coder reliability (Kolbe and Burnett 1991)1.

The final step of the training process was to have coders individually code a representative sample of the content to be analysed. Several goals were accomplished by this step. Firstly, the researcher could assess the inter-coder reliability from this sample. Secondly, it provided the researcher with a further assessment of the ability of the coders to implement the analysis and to use the coding instrument, thus providing both parties with immediate feedback. Finally, it allows the researcher to make any last minute modifications to the coding instrument, definitions of categories and to the instructions for the coders, in order to accommodate the experiences of the coders who are using them (Krippendorf 1980; Kaid and Wadsworth 1989).

In order to minimise variance in the coding procedure, only four coders were recruited. In their first training session, they practice coded You’ve Got Mail, a movie selected by the researcher which is known to contain several product placements. The results of the coding were discussed and compared, and any questions the coders had were addressed by the researcher. Following this, the coders independently viewed and coded Kate and

1 Objectivity refers to the process by which analytical categories are developed and used. Precise operational definitions and detailed rules and procedures for coding are needed to facilitate an accurate and reliable coding process. Detailed rules and procedures reduce coders’ subjective biases and allow replication by others (Kolbe and Burnett, 1991). Furthermore, objectivity is also maximised through training coders, pre-testing measures, having independent coders, and having these coders work independently of each other (Kolbe and Burnett, 1991). Indeed, independent coders are important because as Krippendorf (1980, p74) states, “the worst practice in content analysis is when the investigator develops his recording instructions and applies them all himself or with the help of a few close colleagues, and thus prevents independent reliability checks”.  O              P     !  " # $% &  '  (///( Leopold, another movie with known product placements. The group then met again in a second training session to review their results, ask further questions of the researcher, and resolve any discrepancies.

 O              P     !  " # $% &  '  (//+( APPENDIX 6.4: INSTRUCTIONS FOR CONTENT ANALYSIS AND FILM CODING FORM

     M   

Definitions

Product Placement: the inclusion of products – branded and/or unbranded – to support entertainment story content. Put more simply, product placements consist of real brand-name products (as opposed to fictitious or generic products) which are included as props in a range of entertainment media including movies and television shows.

Scene: A continuous block of storytelling either set in a single location or following a particular character. The end of a scene is typically marked by a change in location, style, or time.

Frame: In film, video, or TV, the particular area of action that is captured by the camera and forms the rectangular image that appears on the screen.

1. Preliminary Information

1.1 Movie Name – please list the full title of the film as it appears on the video / DVD cover.

1.2 Year of Release - please state the year that this film was first released. This information should easily be found on the video / DVD cover or at the end of the closing credits of the film.

1.3 Rating – please list what rating the film has been given by the Australian Office of Film and Literature Classification. This can be found on the front of the video / DVD.

1.4 Film Length - please list the length of the film, in minutes. Again, this information should be found on the cover of the video / DVD.

1.5 Major Stars - please list the full names of the major stars of the film. These may be found on the video / DVD cover.

1.6 Coder Name – please list your full name. If any discrepancies arise during the coding process, you may be contacted and may be required to recode the movie and/or discuss how you arrived at your findings.

1.7 Date of Analysis – please note the date that this content analysis was conducted.

 O              P     !  " # $% &  '  (//0( 2. The Content Analysis

2.1 Product Category / Brand Name 2.1.1 Brand Name – please state the brand name of any branded product you see featured in a scene, or which you hear spoken of. 2.1.2 Product Category – please list any generic product category which you deem significant to a particular scene or character.

2.2 Duration of Product Placement within the Scene 2.2.1 Exposure # (in this scene) – it is likely that a given product will appear visually or be spoken of, multiple times in any one scene. Each of these “exposures” needs to be accounted for. For this reason, please state how many times this product is featured in a particular scene and deal with each “exposure” separately. For example, in this scene, exposure #1 lasted 3 secs, exposure #2 lasted 1 second. For the next scene, the exposure number starts from #1 again. 2.2.2 Length of Exposure (in seconds) – relative to the itemized list of exposures per scene, please time the precise length of each of these, using the number of seconds as the unit of measure. That is, how long does each product appear on screen before there is a cutaway to another image? If a product is the subject of a conversation, please time the length of the conversation (aural communication) of which the product is the subject.

2.3 Form of Product Placement within the Scene 2.3.1 Creative Placement – this occurs when a product / brand appears in the background of a shot. Please tick.

2.3.2 On-set Placement – this occurs when a product is displayed more prominently in a given shot. This prominence may also be reinforced by the fact that the products are consumed or mentioned by one of the. Aural placements should be coded as on-set since they are part of the dialogue and therefore harder to avoid and need to be processed in order to follow the story. Please tick.

2.3.3 Star presence – was one of the major stars in the same frame as the featured brand? Please tick.

2.3.4 Use by Star – in conjunction with star presence, the product is consumed or mentioned by one of the actors. If this is the case, please tick all appropriate boxes (i.e. star presence + use by star).

2.3.5 Relevance of brand to the scene / plot connection – Could the scene have existed without the brand or product featuring? That is, how essential to the scene was it? Was there a high level of integration between the brand and the scene it which it featured? Please select High / Low.

2.3.6 Actual brand shown / spoken – did the actual brand appear in a scene? Or was it spoken of? Please tick.

2.3.7 Logo / Ad / Signage – in contrast to an actual product appearing visually in a scene, a branded logo, advertisement, signage, shopping bag, packaging etc (i.e. non-product form of brand identifier) appears in the scene. Please tick.

 O              P     !  " # $% &  '  (//1( 2.3.8 Visual – please tick this box if the product placement exposure has been visual in nature. If the product placement exposure has been both visual and aural, please tick both boxes.

2.3.9 Audio - please tick this box if the product placement exposure has been aural in nature (e.g. mentioned in conversation, an advertisement is heard in the background on a television or radio). If the product placement exposure has been both visual and aural, please tick both boxes.

3. Additional Comments / Observations

Please note any further observations or comments you have about any of the product placements that have occurred in this film. You may also want to give your opinion as to the level of product placement activity.

For example: • Did the amount of product placement irritate you? • Were the brands used congruent with the story? • Was there too much product placement? • Did any of the product placements seem out of place? • Do you think any of the brands would be offended by the way they appeared?

 O              P     !  " # $% &  '  (//2( FILM CODING FORM (CONTENT ANALYSIS)

Movie Name ______Year of Release ______Rating ______Film Length ______Major Stars ______Coder Name ______Date of Analysis ______

Instructions: Please watch the selected movie at least twice in order to capture all the brand appearances that take place and the length of each exposure. Please detail brand appearances within different scenes separately. Please do not collaborate with anyone else during this content analysis, or discuss your answers with any other coder. If you have any questions about the coding process, please contact Jane Scott (0413 802 104).

Duration of product placement Form of Product Placement within the scene within scene Brand A Number Total Creative On-set Star Use Relevance Actual Logo / V U of length of Placement Placement present by of brand to brand Ad / I D exposures exposure (ie. low (ie. high in Star scene / plot shown Signage S I (secs) prominence) prominence) scene connection / O spoken

 O              P     !  " # $% &  '    (//3(

Duration of product placement Form of Product Placement within the scene within scene Brand A Number Total Creative On-set Star Use Relevance Actual Logo / V U of length of Placement Placement present by of brand to brand Ad / I D exposures exposure (ie. low (ie. high in Star scene / plot shown Signage S I (secs) prominence) prominence) scene connection / O spoken

Additional comments / observations about product usage or product placement within this film:

 O              P     !  " # $% &  '    (/+,( APPENDIX 6.5: RESULTS FROM CONTENT ANALYSIS

Movie Name Year of Rating Length Major Stars Products / Brands Release 10 Things I 1999 PG 97 mins Heath Ledger, Fender guitar, Apple computer, McDonalds, Sketchers, Hate About Julia Stiles Prada, Diet Coke, Guiness, Busch, Bottled Water, You Sportscar, Toyota, Budweiser, Volvo, spray-on hair Beautiful 2000 PG 112 mins Minnie Driver, None Kathleen Turner Big Momma’s 2006 PG 99 mins Martin Lawrence BMW, Brillo, Burberry, HP, Coca Cola, Dairy Queen, House 2 Lucky Charms, Mace, Mercedes, Oklahoma State University, Playboy, Porsche, Princeton University, Ray Ban, YMCA, Victoria’s Secret, Apple Bridget 2001 M 97 mins Renee Absolut, Apple, Budweiser, Coca-Cola, Evian, Ford, Fuji, Jones’ Diary Zellwegger, Hugh Häagen-Dazs, Heinz, Hello!, Herbal Essences, Hitachi, Grant, Colin Firth Lancôme, Lemsip, McDonald's, Mercedes, Microsoft, MINI, Monopoly, Montblanc, MSN, Nescafé, New York Yankees, Nokia, Panerai, Pemberley Press, Polo, Polo Ralph Lauren, Samsung, Sanyo, Silk Cut, Sprite, TDK, Tesco, Tiffany & Co., Twister Charlie’s 2001 M 98 mins Drew Barrymore, Nokia, UPS, Mercedes, CNN, Goodyear, Coke, Shake ‘n Angels Cameron Diaz, Bake, Scrabble Lucy Lui, Bill Murray Clueless 1995 M 93 mins Alicia Jeep, Cranberries, Nine Inch Nails, Mobile phone, Silverstone, Paul Amnesty International, Greenpeace, Diet Coke, Snickers, Rudd, Brittany Cliffs Notes, Starbucks, McDonalds, Minute Maid, Murphy, Breckin Special K, M&Ms, Pro Kennex, Sfida, Coke, Mentos, Meyer Hertz, Calvin Klein, Lays, BMW, Naya, Fred Segal, Tiffany & Co, Christian Dior, Spalding Confetti 2006 M 100 mins Martin Freeman Marriott, Prince (tennis racquets)  O              P     !  " # $% &  '    (/+-( Coyote Ugly 2000 PG 97 mins Piper Perabo, Sharp, Pepsi, Heinz, Kahlua, Tequila, Beer, Jim Beam, Adam Garcia, Jack Daniels, Johnny Walker, Absolut, Canadian Club, John Goodman KFC, Spiderman, Casio, Roland, McDonalds, ER, Old Spice, Apple, Playboy, Yamaha Die Another 2002 M 133 mins Pierce Brosnan Aston Martin, Bollinger, British Airways, Jaguar, Oxford Day University, Harvard University, Ferrari, Lamborghini, Porsche, Range Rover, Sony Ericsson, Ford, Finlandia, Heineken, Omega Drive Me 1999 PG 91 mins Melissa Joan None Crazy Hart, Adrian Grenier EdTV 1999 PG 123 mins Matthew Evian, Puma, UPS, FedEx, KFC, Mountain Dew, Pepsi, McConaughey, Pop Tarts, Bud Light, Nike, Diet Pepsi, Coke, Motorola, Jenna Elfman, Yahoo, Energiser, Lipton Woody Harrelson, Elizabeth Hurley Elf 2003 G 97 mins Will Ferrell Aldo, Bob the Builder, Bert and Ernie, Chanel, Coca Cola, Clinique, Estee Lauder, Etch-a-Sketch, Gimbels, Greyhound, Kangol, Kodak, Lego, Mr Potato Head, Nike, NY1, Pop Tarts, Sesame Street Elizabethtown 2005 M 123 mins Orlando Bloom, DCS shoes, American Airlines, Maker Mark Kirsten Dunst Failure to 2006 M 97 mins Sarah Jessica Powerbar, Coca Cola Launch Parker, Matthew McConaughey Happy 1996 PG 92 mins Adam Sandler Pepsi Gilmore Hitch 2005 M 118 mins Will Smith, Eva Benadryl, Canon, Coca Cola, Corona, Google, Grey Mendes, Kevin Goose, Beatles, New York Standard, Ellis Island, Krispy James Kreme, Lacoste, Mini Cooper, New York Jets, New York Knicks, San Pellegrino, Sheraton, Sony, Starbucks,

 O              P     !  " # $% &  '    (/+*( Subway, Sony Ericsson, adidas, Motorola How to Lose 2003 PG 116 mins Matthew Advil, Sports Illustrated, NBA, Calvin Klein Eternity, A Guy in 10 McConaughey, Webster’s Dictionary, Celine Dion, Apple, BCBG, Days Kate Hudson Blimpie, Botox, Budweiser, Burberry, Carefree, Coca Cola, Dreyfus, Duke University, Harry Winston, Helmut Lang, HJC, Madison Square Garden, New York Knicks, Pedigree, People, Playtex, Poppa Corn, Vagisil, Russell Athletics, Shoei, Starbucks, Verizon Jerry 1996 M 139 mins Tom Cruise, NFL, Moet and Chandon, ESPN, Jack Daniels, New York Maguire Renee Zellwegger Yankees, Boston Red Sox, Hertz, Les Miserables, JVC, Reebok, Chevrolet, Pepsi, Energiser, Nike, Coca Cola, United Airlines, Budweiser, Visa, Fuji, ABC, Marriott, Toshiba, Gatorade Josie and the 2001 PG 98 mins Rachael Leigh British Airways, Panasonic, Target, Coke, Seventeen, Pussycats Cook, Tara Reid, Puma, Diesel, Benetton, TJ Maxx, Ray Ban, Starbucks, Rosario Dawson Sobe, Pizza Hut, Gatorade, McDonalds, AMEX, Steve Madden, Vogue sunglasses, Bloomingdales, Reebok, MTV, mobile phone, Motorola, Evian, The Beatles, The Rolling Stones, Backstreet Boys, Big Mac, Footlocker, Snapple, Billboard, bebe, Rolling Stone magazine, Revlon, Pringles, Steinlager, Barneys, Red Bull, Pepsi, AOL, Sega, Ford, Advil, Snickers, Kodak Kate and 2001 PG 121 mins Meg Ryan, Hugh Palm Pilot, Apple, Gillette, Colgate, Starbucks Leopold Jackman, Breckin Meyer Legally 2001 PG 100 mins Reese Cosmopolitan, InStyle, Jane, Seventeen, Fox and Hound, Blonde Witherspoon, Upscale, Clairol / Herbal Essences, Clinique, Red Bull, Luke Wilson, FedEx, Vogue, Evian, Fred Segal, Harry Winston, Selma Blair Porsche, IBM, Wilson, Apple, UPS, Calvin Klein, Prada, Malibu Barbie, Taco Bell, Mercedes, Perkiset, Bekins, New Balance, Chanel, Harvard, Yale

 O              P     !  " # $% &  '    (/+.( Minority 2002 M 145 mins Tom Cruise American Express, Aquafina, Ben & Jerry’s, Bvlgari, Report Burger King, Century 21, Fox, Gap, Guinness, Lexus, New York Mets, Nokia, Pepsi, Reebok, Revo, USA Today Napoleon 2005 PG 82 mins Jon Heder Lemsip, Chapstick, Loch Ness, Sledgehammer (bicycle) Dynamite Never Been 1999 PG 103 mins Drew Barrymore, Chicago Sun Times, Sprite, Buick, Coke, Apple Kissed David Arquette Computer, Nike Reality Bites 1994 M 94 mins Winona Ryder, BMW, Ford, The Gap, Coke, Domino's Pizza, Evian, Ben Stiller, Ethan Pringles, Camel cigarettes, Nutrasweet, Diet Coke, Pizza Hawke, Janeane Hut, Beer Garofalo Rocky III 1982 PG 99 mins Sylvester Stallone Wheaties, AMEX, GQ, Nike, People magazine, Muppet Show, Maserati, Newsweek, DeLorean, Gatorade, Radio City Music Hall, Nikon, Marines, Power Crunch Candy, TWA, Coke, Exlax, Easter Seals, Press Herald, Marantz, Wurlitzer, London Examiner, Woolworths, Vogue, Bazaar, Lowenbrau, Tennis Magazine, Sugar Babies, Madison Square Garden, Harley Davidson, Caeser's Palace, World Boxing, Ring Magazine, Budweiser, Tuf- Wear, Hamm's Beer Serendipity 2001 PG 91 mins John Cusack, Bloomingdales, woollen gloves, David Duncan antiques, Kate Beckinsale, Serendipity Restaurant / General store, Thomas Travel, Jeremy Piven The Waldolf Hotel, cigarettes, Hermes, Prada, mobile phone, American Airlines, Louis Vuitton Talladega 2006 M 108 mins Will Ferrell, 3M, Bosch, Budweiser, Cadillac, Caterpillar, Chevrolet, Nights Sasha Baron Coca Cola, Coors, Dickies, Dominos Pizza, EA Sports, Cohen, John C ESPN, FedEx, Ford, Fox, Gillette, Goodyear, Honda, Reilly Hummer, Jenga, Jim Beam, KFC, Nextel, Old Spice, Pepsi, Perrier, Playgirl, Powerade, Puma, Taco Bell, Tide Thank You 2005 M 92 mins Aaron Eckhart, None for Smoking William H Macy,

 O              P     !  " # $% &  '    (/+/( Katie Holmes, Adam Brody The Break-Up 2006 M 105 mins Jennifer Aniston, Chicago Cubs, Clorox, Club Med, EA Sports, Panasonic, Vince Vaughn Ruffles, Starbucks, Mastercard, Pepsi, Smith and Wollensky, Staples The Holiday 2006 M 138 mins Cameron Diaz, Sony, Porsche, Daily Telegraph, Blackberry, Google, Kate Winslet, Kiehles, Mini Cooper, Killers CD, Audi, Scrabble, British Jude Law, Jack Airways, Lexus, Western Union, Fed Ex, Barbie, Black weather.com, Hugo Boss, Mercedes, Blockbuster, Universal, MGM, Random House The Island 2005 M 136 mins Ewan McGregor, Puma, Aquafina, Speedo, Apple, Xbox, Reebok, Scarlett Budweiser, Bacardi, Jack Daniels, Popular Mechanics, Johansson, Steve Sprint, Samsung, Michelob, Chevrolet, Bentley, AmTrak, Buscemi Cosmopolitan, NFL, Johnny Rockets, Land Rover, Calvin Klein, Advanced Armor, msn, Cisco Systems, Chrysler, Lexus, Pontiac, BMW, Nissan, Mack, Hummer, KPMG, Maxim, Esquire, NBC, Nokia, Cadillac, Volvo, GMC, Ben & Jerry’s, Tag Heuer The Princess 2001 G 114 mins Julie Andrews, None Diaries Anne Hathaway, Mandy Moore The Pursuit of 2006 M 117 mins Will Smith Coca Cola, Life Savers, Mercedes, Rubik’s Cube, San Happyness Francisco 49ers, Sesame Street, Utz, American Express, Mastercard, Budweiser, Converse, Ferrari, Pepsi The Truman 1998 PG 103 mins Jim Carrey, Laura She magazine, Ford Show Linney, Ed Harris What Women 2000 PG 122 mins Mel Gibson, Advil, Cigarettes, Anti-wrinkle cream, Mascara, Want Helen Hunt, Moisturising Lipstick, At home wax kit, Bath beads, Marisa Tomei Quick dry nail polish, Amore wonder bra, Vidal Sassoon Hair Volumiser, Pore cleansing strips, EPT home pregnancy kit, Control Pantyhose, Visa, Red Wine, Flat

 O              P     !  " # $% &  '    (/++( screen TV, Record player, Alanis Morrissette CD, Herbals, Hairdryer, Fed EX, Fosters, Estee Lauder, Chanel, Nike, Club Med, Apple, Saks Fifth Avenue Whatever it 2000 M 90 mins James Franco, make-up, condoms, beer, computer, Furby, perfume Takes Shane West, Marla Sokoloff, Jodi Lyn O’Keefe You’ve Got 1998 PG 119 mins Meg Ryan, Tom America Online (AOL), Starbucks, Bloomingdales, Pride Mail Hanks and Prejudice, Apple, IBM, Visa, Stolichnaya, The Godfather

 O              P     !  " # $% &  '    (/+0( APPENDIX 6.6: CONTENT ANALYSIS OF ‘THE ISLAND’

Movie Name ____THE ISLAND______Year of Release ____2005______Rating ______M (15+)______Film Length ______136 MINS______Major Stars EWAN McGREGOR, SCARLETT JOHANSSON, STEVE BUSCEMI

Duration of product placement Form of Product Placement within the scene within scene Brand A Number Total Creative On-set Star Use Relevance Actual Logo / V U of length of Placement Placement present by of brand to brand Ad / I D exposures exposure (ie. low (ie. high in scene Star scene / plot shown Signage S I Recognition (secs) prominence) prominence) connection / O spoken msn 93.2% 4 16.7     High    Puma 77.0% 3 11.32     High  

Calvin Klein 69.9% 2 7.73    High   Nokia 63.5% 3 8.14    High   NBC 63.0% 2 2    High   Xbox 61.3% 21 47.13     High   

Mack 52.1% 3 7.22   High  

Maxim 37.6% 2 2    Low  

 O              P     !  " # $% &  '    (/+1( Duration of product placement Form of Product Placement within the scene Brand within scene Actual A Number Total Creative On-set Star Use Relevance brand Logo / Ad V U of length of Placement Placement present by of brand to shown / Signage I D

Recognition exposures exposure (ie. low (ie. high in Star scene / plot / S I

(secs) prominence) prominence) scene connection spoken O NFL 32.4% 1 2.64  Low  

Hummer 29.4% 3 7.1   High   Bentley 25.5% 1 1    High   Chrysler 24.5% 47 113.17  High   Apple 22.3% 4 25.3    Low   Ben & Jerry’s 21.8% 3 5.77    Low   Advanced 21.0% 6 11.4 High     Armor Popular 20.5% 1 3.63    Low   Mechanics BMW 18.9% 3 3.57  High  

Esquire 18.8% 1 1    Low   Cadillac 18.0% 21 84.72     High    Chevrolet 17.6% 4 14.69    High   Aquafina 16.8% 3 7.16   Low   Land Rover 13.2% 2 4.7  High   Budweiser 11.2% 12 24.42     Low    

 O              P     !  " # $% &  '    (/+2( Duration of product placement Form of Product Placement within the scene Brand within scene A Number Total Creative On-set Star Use Relevance Actual Logo / Ad V U of length of Placement Placement present by of brand to brand / Signage I D exposures exposure (ie. low (ie. high in Star scene / plot shown S I Recognition (secs) prominence) prominence) scene connection / O

spoken Lexus 11.2% 2 5.16  High  

AmTrak 11.1% 3 15.35     Low   GMC 8.0% 1 1.53  High   Volvo 7.4% 1 0.7  High   Tag Heuer 6.9% 1 1.45    Low   Jack Daniels 5.3% 2 1.35    Low    Nissan 5.3% 2 4.6  High   Reebok 4.2% 2 3.64     Low     Samsung 3.2% 1 3.63    Low   Johnny 2.6% 1 2.45 Low    Rockets Pontiac 2.6% 1 0.6  High   Cisco 2.1% 4 8.19     Low    Systems KPMG 2.1% 1 2.27  Low   Cosmopolitan 2.1% 1 0.72  Low  

 O              P     !  " # $% &  '    (/+3( Duration of product placement Form of Product Placement within the scene Brand within scene A Number Total Creative On-set Star Use Relevance Actual Logo / Ad V U of length of Placement Placement present by of brand to brand / Signage I D exposures exposure (ie. low (ie. high in Star scene / plot shown S I Recognition (secs) prominence) prominence) scene connection / O spoken Bacardi 1.6% 1 3.88   Low   Michelob 1.1% 5 18.22    Low   Speedo n/a 2 4.19  Low  

 O              P     !  " # $% &  '    (/0,( APPENDIX 6.7: BRAND FAMILIARITY QUESTIONNAIRE

In order to ensure your anonymity, but for us to be able to identify all your surveys as yours, we ask you to give us two 4-digit numbers that are easy for you to remember.

We suggest that you use the last four digits of your telephone number as your first code, and then the day and month of your birthday for your second code. Please remember the numbers that you give us, as you will have to use them again later.

For example: If your phone number is 9123 5490, your code number = 5490 If your birthday is 3rd May, your code number = 0305

Unique Code Number #1 ______(The last 4 digits of your telephone number)

Unique Code Number #2 ______(The day and month of your birthday)

First Name Initial ______(Please do not write your full name)

Firstly, we would like you to tell us IF YOU HAVE EVER HEARD OF the following brands, whether you would RECOGNISE THEM (OR THEIR LOGOS) IF YOU SAW THEM, and whether you have ever PURCHASED OR USED them.

Please tick the boxes of the brands that you have heard of, could recognise if you saw them, or have purchased or used. This may mean that you tick several boxes per brand. If you have not heard of the brand (and therefore have not used it or would not recognise it), please tick the “never heard of” box only.

For example, I have I have heard I would I have Product Brand NEVER of this brand recognise purchased or Category heard of this this brand or used this brand its logo if I brand saw it Yoghurt Yoplait  Ski   Fruche  Yogo   

Now it’s your turn……  O              P     !  " # $% &  '    (/0-(

I have I have I would I have Product Brand NEVER heard of recognise this purchased or Category heard of this brand or its used this this brand brand logo if I saw it brand Sporting Puma

apparel Nike

adidas

Reebok

Bottled water Mount Franklin

Evian

Aquafina

Perrier

Computers IBM

Dell

Apple

Sony / VAIO

Game consoles Nintendo Wii

Sony Playstation

Microsoft Xbox

Beer VB

Crown Lager

Budweiser

Heineken

Michelob

Technology Popular Mechanics magazines Wired

PC User

 O              P     !  " # $% &  '    (/0*(

I have I have I would I have Product Brand NEVER heard of recognise this purchased Category heard of this brand or its or used this this brand brand logo if I saw it brand Alcoholic Bacardi

Spirits Jim Beam

Jack Daniels

Smirnoff

Absolut

Men’s Ralph

Magazines FHM

Maxim

Esquire

GQ

Women’s Cosmopolitan magazines marie claire

Vogue

Girlfriend

Dolly

American Amtrak transportation Greyhound companies Southwest Airlines

United Airlines

Restaurant McDonalds

Chains Hard Rock Café

Johnny Rockets

TGI Fridays

 O              P     !  " # $% &  '    (/0.( I have I have I would I have Product Brand NEVER heard of recognise this purchased Category heard of this brand or its or used this this brand brand logo if I saw it brand Cars BMW

Lexus

Bentley

Porsche

Ferrari

Cadillac

Hummer

Jeep

Land Rover

GMC

Chrysler

Chevrolet

Citroën

Nissan

Volvo

Pontiac

Trucks Mack

Isuzu

Mercedes Benz

Advanced Armor

Internet search msn engines Yahoo

Google

AltaVista

 O              P     !  " # $% &  '    (/0/( I have I have I would I have Product Brand NEVER heard of recognise this purchased Category heard of this brand or its or used this this brand brand logo if I saw it brand Luxury clothing / Polo Ralph Lauren perfume brands Calvin Klein

Gucci

Armani

Burberry

Ice cream Gelatissimo

Ben & Jerry’s

Baskin-Robbins

New Zealand Natural

Computer Nortel Network Juniper Hardware Cisco Systems

Avaya

Mobile phones Nokia

Samsung

Motorola

Sony Ericsson

American ABC television NBC networks Fox

ESPN

Soft drinks Coca-Cola

Sprite

Pepsi

Fanta

 O              P     !  " # $% &  '    (/0+(

I have I have I would I have Product Brand NEVER heard of recognise this purchased Category heard of this brand or its or used this this brand brand logo if I saw it brand Accounting PwC

firms KPMG

Deloitte

Sporting NFL Associations NBL

NRL

Watches Rolex

Omega

Tag Heuer

 O              P     !  " # $% &  '    (/00( APPENDIX 6.8: POST-FILM TESTING – AUDIENCE ENGROSSMENT, PRODUCT PLACEMENT RECOGNITION AND PROGRAM LIKING

A few weeks ago when we came to ask you questions about your familiarity of different brands, we asked you to provide us with two code numbers that we could use to identify all your surveys as yours.

We ask you to provide us with these same codes today so we can match the responses you gave us last time with the ones you give us today. Most likely, the codes you gave us were the last four digits of your telephone number as your first code, and then the day and month of your birthday for your second code.

For example: If your phone number is 9123 5490, your code number = 5490 If your birthday is 3rd May, your code number = 0305

Unique Code Number #1 ______(The last 4 digits of your telephone number)

Unique Code Number #2 ______(The day and month of your birthday)

First Name Initial ______(Please do not write your full name)

Had you seen this movie before? Yes % (1) No % (0)

Are you Male % (1) Female % (2)

On average, how many movies would you watch each month (this includes those on television, on DVD/video, at the cinema)? 0-5 % (1) 6-10 % (2) 11-20 % (3) 21+ % (4)

 O              P     !  " # $% &  '    (/01(

Firstly, we are interested in understanding your feelings and reactions to the movie you just saw (The Island).

Please answer EVERY QUESTION in the way that it is specified

Work through these questions at your own pace.

There are no right or wrong answers.

PART 1

Firstly, we would like you to tell us how you reacted DURING the movie.

Never For EACH STATEMENT, please circle the number that best describes your movie reactions. At times Throughout the the Throughout

I laughed so hard that I cried 1 2 3 I found myself looking around the room 1 2 3 I was anxious 1 2 3 I had trouble keeping my eyes open 1 2 3 I was calm 1 2 3 I laughed out loud 1 2 3 I fell asleep 1 2 3 I was uneasy 1 2 3 I smiled or chuckled to myself 1 2 3 I was close to tears (e.g. eyes welled up with tears / there was a lump in my throat) 1 2 3 I needed to grab hold of something (e.g. the person next to me, the armrest) 1 2 3 I couldn’t sit still 1 2 3 I cried 1 2 3 I was at ease 1 2 3 I felt jumpy 1 2 3 I found myself fidgeting 1 2 3 I yawned 1 2 3 My stomach felt like it was tied in knots 1 2 3 I was totally relaxed 1 2 3 I sobbed 1 2 3 I was tense 1 2 3

 O              P     !  " # $% &  '    (/02(

PART 2

In this section, we would like you to tell us how you felt DURING the movie.

Never movie

For EACH STATEMENT, please circle the number that best describes At times

your feelings. the Throughout

The movie made me feel appalled 1 2 3 The movie made me feel totally depressed 1 2 3 The movie made me feel scared 1 2 3 The movie made me feel happy 1 2 3 The movie made me feel angry 1 2 3 The movie made me feel distressed 1 2 3 The movie made me feel heartened 1 2 3 The movie made me feel miserable 1 2 3 The movie made me feel horrified 1 2 3 The movie made me feel cross 1 2 3 The movie made me feel good 1 2 3 The movie made me feel dismayed 1 2 3 The movie made me feel comforted 1 2 3 The movie made me feel elated 1 2 3 The movie made me feel on edge 1 2 3 The movie made me feel concerned 1 2 3 The movie made me feel terrified 1 2 3 The movie made me feel enraged 1 2 3 The movie made me feel distraught 1 2 3 The movie made me feel sad 1 2 3 The movie made me feel uplifted 1 2 3

 O              P     !  " # $% &  '    (/03(

PART 3

Now we would like to find out your thoughts about the movie.

For EACH STATEMENT, please circle the number that best describes your thoughts. Note that these questions ask you to agree or disagree Agree with each statement. Disagree Strongly Agree Strongly Strongly Disagree Strongly

I could happily watch unlimited re-runs of this movie 1 2 3 4 I was interested in the storyline 1 2 3 4 My mind wandered at times during the movie 1 2 3 4 I barely watched or listened to any of the movie 1 2 3 4 I developed a real affection towards one or more of the main characters 1 2 3 4 I enjoyed watching this movie 1 2 3 4 I couldn't wait to see what happened next 1 2 3 4 My mind was totally pre-occupied with things other than the movie 1 2 3 4 I felt completely immersed in the story 1 2 3 4 I enjoyed watching this movie more than I have most others 1 2 3 4 I really liked one or more of the main characters 1 2 3 4 There were times that I chose not to watch or listen to the movie 1 2 3 4 The movie starred one of my favourite actors 1 2 3 4 I really like this type of movie 1 2 3 4

PART 4

In this section we would like to know how you made sense of the movie.

For EACH STATEMENT, please circle the number that best describes Agree Disagree your response. Like the last section, these questions ask you to agree Strongly Agree Strongly

or disagree with each statement. Disagree Strongly

I thought that overall the storyline was difficult to understand 1 2 3 4 There were times I needed to pay attention to follow the story 1 2 3 4 I found that following this story was mentally demanding 1 2 3 4 I had no problems following the storyline 1 2 3 4 I thought that parts of the storyline were difficult to understand 1 2 3 4 I thought the storyline was totally incomprehensible 1 2 3 4 Generally I had to concentrate to follow the story 1 2 3 4 I found the storyline very easy to understand 1 2 3 4

 O              P     !  " # $% &  '    (/1,(

PART 5

In this section we would like to gain a more general understanding of your movie-going experience.

For EACH STATEMENT, please circle the number that best describes Agree your thoughts. Again, these questions ask you to agree or disagree Disagree with each statement. Agree Strongly Strongly Disagree Strongly

There were some external circumstances that prevented me from paying full 1 2 3 4 attention to the movie (e.g. there was outside noise, people talking) I was comfortable in my seat 1 2 3 4 The air temperature was pleasant 1 2 3 4 I had a clear view of the screen 1 2 3 4 The sound quality was good 1 2 3 4 The image was good quality 1 2 3 4 I was in a good mood when I came to see this movie 1 2 3 4

Now we are interested in what brands you remember seeing in the movie.

Was the brand DEFINITELY in the movie? Do you think it MAY HAVE BEEN? Or was it NOT in the movie?

Please tick the box that best represents your belief about whether the brand was in the movie or not. The actual brand may have appeared, you may have seen its logo, or the brand name may have been mentioned in conversation.

Please tick one box per brand only.

For example,

This brand was I think this brand This brand was Product Brand DEFINITELY IN the MAY HAVE been in NOT Category movie the movie in the movie Yoghurt Yoplait  Ski  Fruche  Yogo 

Now it’s your turn……  O              P     !  " # $% &  '    (/1-(

This brand was I think this brand This brand was Product Brand DEFINITELY IN MAY HAVE been NOT Category the movie in the movie in the movie Sporting Puma

apparel Nike

adidas

Reebok

Bottled water Mount Franklin

Evian

Aquafina

Perrier

Computers IBM

Dell

Apple

Sony / VAIO

Game consoles Nintendo Wii

Sony Playstation

Microsoft Xbox

Beer VB

Crown Lager

Budweiser

Heineken

Michelob

Technology Popular Mechanics magazines Wired

PC User

 O              P     !  " # $% &  '    (/1*(

This brand was I think this brand This brand was Product Brand DEFINITELY IN MAY HAVE been NOT Category the movie in the movie in the movie Alcoholic Bacardi

Spirits Jim Beam

Jack Daniels

Smirnoff

Absolut

Men’s Ralph

Magazines FHM

Maxim

Esquire

GQ

Women’s Cosmopolitan magazines marie claire

Vogue

Girlfriend

Dolly

American Amtrak transportation Greyhound companies Southwest Airlines

United Airlines

Restaurant McDonalds

Chains Hard Rock Café

Johnny Rockets

TGI Fridays

 O              P     !  " # $% &  '    (/1.(

This brand was I think this brand This brand was Product Brand DEFINITELY IN MAY HAVE been NOT Category the movie in the movie in the movie Cars BMW

Lexus

Bentley

Porsche

Ferrari

Cadillac

Hummer

Jeep

Land Rover

GMC

Chrysler

Chevrolet

Citroën

Nissan

Volvo

Pontiac

Trucks Mack

Isuzu

Mercedes Benz

Advanced Armor

Internet search msn engines Yahoo

Google

AltaVista

 O              P     !  " # $% &  '    (/1/(

This brand was I think this brand This brand was Product Brand DEFINITELY IN MAY HAVE been NOT Category the movie in the movie in the movie Luxury clothing Polo Ralph Lauren / perfume Calvin Klein brands Gucci

Armani

Burberry

Ice cream Gelatissimo

Ben & Jerry’s

Baskin-Robbins

New Zealand Natural

Computer Nortel Network Juniper Hardware Cisco Systems

Avaya

Mobile phones Nokia

Samsung

Motorola

Sony Ericsson

American ABC television NBC networks Fox

ESPN

Soft drinks Coca-Cola

Sprite

Pepsi

Fanta

 O              P     !  " # $% &  '    (/1+(

This brand was I think this brand This brand was Product Brand DEFINITELY IN MAY HAVE been NOT Category the movie in the movie in the movie Accounting PwC

firms KPMG

Deloitte

Sporting NFL Associations NBL

NRL

Watches Rolex

Omega

Tag Heuer

Finally, we want to get an idea of how much you liked this movie (The Island). For EACH STATEMENT, please circle the number that best describes your feelings about the movie. These questions ask you to agree or disagree with each statement.

Strongly Strongly Disagree Disagree Slightly Disagree Neither agree nor disagree Slightly Agree Agree Strongly Agree

I'm glad I had a chance to see this movie 1 2 3 4 5 6 7

I would never watch a re-run of this movie 1 2 3 4 5 6 7

I liked watching this movie 1 2 3 4 5 6 7 If I knew this movie was going to be on 1 2 3 4 5 6 7 television, I would look forward to watching it I disliked watching this movie more than I 1 2 3 4 5 6 7 do most other movies There is something about this movie that 1 2 3 4 5 6 7 appeals to me

Thank you very much for your participation in the two stages of this research!  O              P     !  " # $% &  '    (/10(

APPENDIX 6.9: PEARSON CORRELATIONS BETWEEN PRODUCT PLACEMENT CHARACTERISTICS

star star Star Star Aural Aural Visual Visual Use byUse Low plot Low plot High plot Creative presence Total secsTotal placement placement placement placement Prominent Prominent connection connection No. of times Actual branded brand appeared brand appeared Brand logo / ad / product appeared signage appeared Number of times brand appeared 1.00

Total seconds brand appeared 0.96 1.00 0.00

Brand was a creative placement 0.13 0.11 1.00 0.43 0.52

Brand was a prominent placement 0.06 0.13 -0.51 1.00 0.71 0.45 0.00

Brand and star appeared together -0.01 0.07 -0.15 0.15 1.00 0.95 0.68 0.38 0.36

Brand was used by the star 0.04 0.12 -0.20 0.26 0.85 1.00 0.82 0.45 0.22 0.10 0.00

Brand had low plot connection -0.22 -0.32 0.26 -0.42 -0.34 -0.39 1.00 0.18 0.05 0.11 0.01 0.04 0.01

Brand had high plot connection 0.22 0.32 -0.26 0.42 0.34 0.39 -1.00 1.00 0.18 0.05 0.11 0.01 0.04 0.01 0.00

- Actual branded product appeared 0.18 0.17 0.26 0.02 0.10 0.18 0.02 0.02 1.00 0.28 0.30 0.11 0.89 0.54 0.29 0.93 0.93

Brand logo / ad / signage appeared -0.03 -0.03 -0.24 0.20 0.07 0.06 -0.33 0.33 -0.64 1.00 0.86 0.86 0.14 0.23 0.67 0.72 0.04 0.04 0.00

Brand appeared visually 0.13 0.17 -0.13 0.25 -0.27 -0.31 -0.14 0.14 -0.14 0.23 1.00 0.44 0.31 0.43 0.12 0.10 0.05 0.38 0.38 0.38 0.17

- - Brand appeared aurally -0.08 -0.16 0.18 -0.21 0.23 0.16 0.20 0.20 0.20 0.02 -0.72 1.00 0.61 0.34 0.27 0.20 0.15 0.32 0.22 0.22 0.22 0.89 0.00

 O              P     !  " # $% &  '    (/11(