understanding and measuring light buyer brand equity

179
Are you Keeping Track of your Light Buyers? Understanding and Measuring Light Buyer Brand Equity Samantha Hogan BMktgComm BBus(Honours) Supervisors: Professor Jenni Romaniuk; Dr Margaret Faulkner A thesis submitted for the degree of Masters by Research (Marketing) University of South Australia August 2015

Upload: others

Post on 11-Sep-2021

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Understanding and Measuring Light Buyer Brand Equity

Are you Keeping Track of your Light Buyers?

Understanding and Measuring Light Buyer Brand Equity

Samantha Hogan

BMktgComm BBus(Honours)

Supervisors:

Professor Jenni Romaniuk; Dr Margaret Faulkner

A thesis submitted for the degree of

Masters by Research (Marketing)

University of South Australia

August 2015

Page 2: Understanding and Measuring Light Buyer Brand Equity

II

Acknowledgements

First and foremost, I must thank my wonderful supervisors, Jenni Romaniuk and Margaret Faulkner. I could not have hoped for better mentors. I am so grateful for the amazing academic leadership and personal support you have given me – I have been very lucky to have you both by my side.

Thank you to my colleagues at the Ehrenberg-Bass Institute, both past and present, for your advice, feedback and support. You have made the Institute a wonderful workplace to be involved with.

To my friends, both near and far, for your understanding and continuous encouragement, thank you all.

My final acknowledgement goes to my family. I am incredibly grateful for the patience and support you have given me. Thank you for believing in my abilities, and encouraging me to take every opportunity that presented itself.

I declare that this thesis does not incorporate without acknowledgement any material

previously submitted for a degree or diploma in any University and that to the best of my knowledge it does not contain any materials previously published or written by another person except where due reference is made in text.

Samantha Hogan …………………………………… August 2015

Page 3: Understanding and Measuring Light Buyer Brand Equity

III

AbstractCustomer Based Brand Equity (CBBE) surveys consist of different measures designed

to gauge the relationship between the brand and consumer memory structures over time. These surveys help assess the performance of a brand and the potential for brand growth (Aaker 1992; Keller 1993). Traditionally marketers believe that brand growth can be sourced from heavier buyers, who buy the brand often, generating more sales and profit per customer (Anschuetz 2002). At the aggregate level, however, light brand buyers make up the largest portion of a brand’s customer base (Ehrenberg 1959; Sharp 2010). As a brand grows it acquires new buyers and its existing buyers purchase (slightly) more often (Baldinger, Blair & Echambadi 2002; Ehrenberg, Goodhardt & Barwise 1990). This gain in brand buyers is in line with the distribution of buyers it had the previous period, indicating that most of the buyers gained will be light buyers (Romaniuk 2011). Therefore, determining where and how light brand buyers respond in brand equity measures will help to better understand changes in the brand over time.

This research aims to better understand and measure light buyer brand equity. Two brand equity measures are analysed: brand awareness and brand image. All CBBE

surveys contain brand awareness measures as once a person is aware of the brand information can be linked to the brand in memory (Keller & Davey 2001). This information is measured via brand image, which helps identify the current position of the brand and evaluate advertising effectiveness (Kapferer 2008). Therefore, it is important to assess the measures from a light buyer perspective.

Part One: Brand Awareness

Secondary data is used to analyse light buyer response to three brand awareness measures (top of mind (TOM), unprompted and prompted) for over 10,000 respondents, 52 brands, four product categories and three countries. The research tests light buyer response compared to non and heavier buyers, while also examining light buyer response across awareness measures.

Results from this thesis suggest utilising the prompted awareness measure to assess light buyer absolute level of awareness. Light buyer response increases from TOM to unprompted to prompted awareness, in line with the difficulty of brand retrieval from memory. In prompted awareness, respondents indicate whether the brand name is present in memory networks, allowing brands purchased frequently and infrequently to be selected. Limitations of competitor linkages and accessibility present in the TOM and

Page 4: Understanding and Measuring Light Buyer Brand Equity

IV

unprompted recall measures are removed, demonstrated by a greater light buyer response level for prompted awareness.

Additionally, this thesis determines that light buyers are better able to retrieve the brand name from memory than non-buyers for all awareness measures. This suggests that analysis should occur separately for both non and light buyer groups. In comparison to heavier buyers, light buyer response is significantly lower than heavier buyers for TOM awareness. However, light and heavier buyers respond in a similar manner to unprompted and prompted awareness. Therefore, while analysis should occur separately for light and heavier buyers for TOM awareness, there is no need to separate light and heavier buyer groups for unprompted and prompted awareness analysis.

Part Two: Brand Image

Pick any (PA) and forced choice binary (FCB) image measures are compared for light buyer response using two different methods of prompting for response (attribute or brand). Primary data collected via an online questionnaire in the UK tests two product categories: breakfast cereal and butter/margarine. The study has a sample size of over 2,000 respondents, evenly divided between four sample groups for comparison of measures and prompted method.

The results from this thesis recommend using a PA attribute prompted measure in CBBE surveys. While the FCB measure consistently captures a higher proportion of light buyer response than PA across both prompting methods, light buyers select over half of the brands (or attributes) presented for each attribute (or brand). Results suggest that light buyers may be selecting attributes that are typical of the category rather than only those linked to the brand in memory (Barsalou 1983) and/or that there is an agreement bias present (as suggested in Joyce 1963). In comparison, the PA measure allows light buyers to indicate only what is evoked from the cue, capturing attributes linked to the brand rather than the category in memory. Light buyers associate only around a third of brands (or attributes) to each attribute (or brand). The PA attribute prompted measure also shows greater differentiation between light and non/heavier buyer response, where attributes linked in memory can be assessed separately for these different types of brand buyers.

This research helps better understand light brand buyers, the extent of their brand knowledge and how to better capture their response in brand equity surveys.

Page 5: Understanding and Measuring Light Buyer Brand Equity

V

Table of Contents Chapter 1 INTRODUCTION 1

1.1 Research Background 1 1.1.1 Customer Based Brand Equity 1 1.1.2 How Brands Grow 2

1.2 Research Aim 2 1.3 Research Method 3 1.4 Thesis Structure 4

Chapter 2 LIGHT BRAND BUYERS 5 2.1 The Importance of Light Brand Buyers 5 2.2 Light Brand Buyer Classification 9 2.3 Chapter Summary 11

Chapter 3 BRAND EQUITY & CONSUMER MEMORY 12 3.1 What is Brand Equity? 12

3.1.1 Brand Equity and Light Buyers 13 3.2 Customer Based Brand Equity Measures 13 3.3 Consumer Memory Structures 14

3.3.1 Associative Network Theories of Memory 14 3.3.2 Information Retrieval from Memory 15 3.3.3 Light Buyer Memory Structure 15

3.4 Chapter Summary 16

Part One: BRAND AWARENESS 17

Chapter 4 BRAND AWARENESS 18 4.1 What is Brand Awareness? 18 4.2 The Importance of Brand Awareness 18 4.3 Brand Awareness Measurement 20

4.3.1 Brand Recall 20 4.3.2 Prompted Brand Awareness 22 4.3.3 Research Scope 23

4.4 Light Brand Buyer Awareness 24 4.4.1 Top of Mind Awareness 25 4.4.2 Unprompted Awareness 26 4.4.3 Prompted Awareness 27 4.4.4 Light Buyer Response across Awareness Measures 28

4.5 Chapter Summary 30

Chapter 5 AWARENESS RESEARCH APPROACH 31 5.1 Secondary Data 31 5.2 Product Categories 31 5.3 Data Sets 32 5.4 Operationalisation of Brand Awareness 33 5.5 Selecting an Appropriate Timeframe for Buyer Classification 33 5.6 Operationalisation of Light Brand Buyers 34 5.7 Data Analysis 35

5.7.1 Buyer and Non-Buyer Response 35 5.8 Chapter Summary 36

Chapter 6 AWARENESS RESULTS & DISCUSSION 37 6.1 Awareness Results 37

6.1.1 TOM Awareness Results 37 6.1.2 Unprompted Awareness Results 45 6.1.3 Prompted Awareness Results 52 6.1.4 Light Buyer Response across Awareness Measures 58

Page 6: Understanding and Measuring Light Buyer Brand Equity

VI

6.2 Awareness Discussion 61 6.2.1 TOM Awareness 62 6.2.2 Unprompted Awareness 62 6.2.3 Prompted Awareness 63 6.2.4 Summary 64

6.3 Chapter Summary 65

Part Two: BRAND IMAGE 66

Chapter 7 BRAND IMAGE 67 7.1 What is Brand Image? 67 7.2 The Importance of Brand Image 67 7.3 Brand Image Measurement 68

7.3.1 Scaling Measures 68 7.3.2 Sorting Measures 70

7.3.2.1 Pick-Any Measure (PA) 70 7.3.2.2 Forced-Choice Binary Measure (FCB) 72

7.3.3 Prompting Method 73 7.3.4 Research Scope 75

7.4 Light Brand Buyer Associations 77 7.4.1 Attribute Versus Brand Prompting 78

7.4.1.1 PA Measure 80 7.4.1.2 FCB Measure 80

7.4.2 PA Versus FCB 81 7.5 Chapter Summary 82

Chapter 8 IMAGE RESEARCH APPROACH 83 8.1 Primary Data 83 8.2 Product Categories and Timeframe 83 8.3 Questionnaire Design 84

8.3.1 Brands Tested 84 8.3.2 Attributes Tested 85 8.3.3 Brand Buyer Classification 87

8.4 Operationalisation of Brand Image 87 8.5 Sample Size 90 8.6 Screening Respondents 91 8.7 Data Collection and Demographics 91 8.8 Operationalisation of Light Brand Buyer 92 8.9 Data Analysis 93 8.10 Chapter Summary 93

Chapter 9 IMAGE RESULTS & DISCUSSION 94 9.1 Image Results 94

9.1.1 PA: Attribute versus Brand Prompting 94 9.1.2 FCB: Attribute versus Brand Prompting 97 9.1.3 PA versus FCB Results 99

9.1.3.1 Attribute Prompted 100 9.1.3.2 Brand Prompted 101

9.1.4 Summary 103 9.1.4.1 Average Number of Attributes 104 9.1.4.2 Average Number of Brands 105 9.1.4.3 Distinguishing between light and non/heavier buyer response 107

9.2 Image Discussion 110 9.2.1 Proportion of Light Buyer Response 110 9.2.2 Average Number of Brands/Attributes Selected 111 9.2.3 Distinguishing between Light and Non/Heavier Buyer Response 111 9.2.4 Summary 112

9.3 Chapter Summary 113

Page 7: Understanding and Measuring Light Buyer Brand Equity

VII

Chapter 10 CONCLUSION 114 10.1 Contribution to Marketing Knowledge and Theory 114

10.1.1 Brand Awareness 114 10.1.2 Brand Image 115

10.2 Contribution to Marketing Practice 117 10.2.1 Brand Awareness 117 10.2.2 Brand Image 118

10.3 Strengths of the Present Study 119 10.4 Limitations of the Present Study 119 10.5 Avenues for Future Research 120

List of References 122

Appendix A: Image Questionnaire 131

Appendix B: Income & Education Demographics 136

Appendix C: Buyer Purchase Frequency Distribution 137

Appendix D: PA: Attribute versus Brand Prompted 138

Appendix E: FCB: Attribute versus Brand Prompted 143

Appendix F: PA versus FCB: Attribute Prompted 148

Appendix G: PA versus FCB: Brand Prompted 153

Appendix H: Non, Light and Heavier Buyer Image Response 158

Page 8: Understanding and Measuring Light Buyer Brand Equity

VIII

List of Tables Table 1: Summary of three brand awareness measures. ............................................................................... 30 Table 2: Description of multiple sets of data for Brand Awareness. .............................................................. 32 Table 3: Example of buying distributions and buyer classification. ................................................................ 34 Table 4: Example TOM awareness and buyer type cross-tabulation for Brand A pasta sauce. .................... 35 Table 5: Proportion of non-buyer & light buyer TOM awareness response - Tea. ......................................... 38 Table 6: Proportion of non-buyer & light buyer TOM awareness response - Pasta Sauce. ........................... 38 Table 7: Proportion of non-buyer & light buyer TOM awareness response – Soft Drink. .............................. 39 Table 8: Proportion of non-buyer & light buyers TOM awareness response – Whiskey. ............................... 40 Table 9: Proportion of light buyer & heavier buyer TOM awareness response - Tea. .................................... 41 Table 10: Proportion of light buyer & heavier buyer TOM awareness response - Pasta Sauce. ................... 42 Table 11: Proportion of light buyer & heavier buyer TOM awareness response – Soft Drink. ....................... 43 Table 12: Proportion of light buyer & heavier buyer TOM awareness response – Whiskey. .......................... 43 Table 13: Summary TOM awareness results. ................................................................................................. 44 Table 14: Logistic regression results for non, light and heavier buyer TOM awareness. ............................... 45 Table 15: Proportion of non-buyer & light buyer unprompted awareness response - Tea. ........................... 45 Table 16: Proportion of non-buyer & light buyer unprompted awareness response - Pasta Sauce. ............. 46 Table 17: Proportion of non-buyer & light buyer unprompted awareness response – Soft Drink. ................. 47 Table 18: Proportion of non-buyer & light buyer unprompted awareness response – Whiskey. ................... 47 Table 19: Proportion of light buyer & heavier buyer unprompted awareness response - Tea. ...................... 48 Table 20: Proportion of light buyer & heavier buyer unprompted awareness response - Pasta Sauce. ........ 49 Table 21: Proportion of light buyer & heavier buyer unprompted awareness response – Soft Drink. ........... 49 Table 22: Proportion of light buyer & heavier buyers unprompted awareness response – Whiskey. ............ 50 Table 23: Summary unprompted awareness results. ..................................................................................... 51 Table 24: Logistic regression results for non, light and heavier buyer unprompted awareness. ................... 52 Table 25: Proportion of non-buyer & light buyer prompted awareness response - Tea. ............................... 52 Table 26: Proportion of non-buyer & light buyer prompted awareness response - Pasta Sauce. ................. 53 Table 27: Proportion of non-buyer & light buyer prompted awareness response – Soft Drink. ..................... 53 Table 28: Proportion of non-buyer & light buyer prompted awareness response – Whiskey. ....................... 54 Table 29: Proportion of light buyer & heavier buyer prompted awareness response - Tea. .......................... 55 Table 30: Proportion of light buyer & heavier buyer prompted awareness response - Pasta Sauce. ............ 56 Table 31: Proportion of light buyer & heavier buyer prompted awareness response – Soft Drink. ............... 56 Table 32: Summary prompted awareness results. ......................................................................................... 57 Table 33: Logistic regression results for non, light and heavier buyer prompted awareness. ....................... 57 Table 34: Proportion of light buyer TOM, unprompted & prompted awareness response –Tea. .................. 58 Table 35: Proportion of light buyer TOM, unprompted & prompted awareness response –Pasta Sauce. .... 59 Table 36: Proportion of light buyer TOM, unprompted & prompted awareness response – Soft Drink. ....... 59 Table 37: Proportion of light buyer TOM, unprompted & prompted awareness response –Whiskey. ........... 60 Table 38: Summary light buyer awareness results. ........................................................................................ 60 Table 39: Logistic regression results for non and light buyers across TOM, unprompted and prompted

awareness measures. ............................................................................................................................ 61 Table 40: Logistic regression results for light and heavier buyers across TOM, unprompted and prompted

awareness measures. ............................................................................................................................ 61 Table 41: Summary of key image measurement studies. ............................................................................... 76 Table 42: Summary of brand image measures for testing. ............................................................................ 82 Table 43: Number of brands and attributes tested in image studies. ............................................................ 84 Table 44: Attributes selected for inclusion in brand image questionnaire. ..................................................... 87 Table 45: Respondent allocation to image methods. ..................................................................................... 90 Table 46: Distribution of respondents across image methods by age and gender. ....................................... 92 Table 47: Distribution of respondents across image methods by location. ................................................... 92 Table 48: Proportion of light brand buyer cereal PA response – Good for a treat. ........................................ 95 Table 49: Proportion of light brand buyer cereal PA response – A healthy option. ........................................ 95 Table 50: Proportion of light brand buyer butter/margarine PA response – A healthy option. ...................... 96 Table 51: Proportion of light brand buyer butter/margarine FCB response – Spreads easily. ....................... 98 Table 52: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Good value for

money. ................................................................................................................................................. 100 Table 53: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted –

Helps control cholesterol. .................................................................................................................... 101 Table 54: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Good value for

money. ................................................................................................................................................. 102

Page 9: Understanding and Measuring Light Buyer Brand Equity

IX

Table 55: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Good value for money. .................................................................................................................................. 102

Table 56: Proportion of light brand buyer FCB and PA image response, attribute and brand prompted- Cereal. .................................................................................................................................................. 103

Table 57: Proportion of light brand buyer FCB and PA image response, attribute and brand prompted- Butter/Margarine. ................................................................................................................................. 103

Table 58: Average number of attributes selected per brand, FCB and PA attribute prompted - Cereal. .... 104 Table 59: Average number of attributes selected per brand, FCB and PA brand prompted - Cereal. ........ 104 Table 60: Average number of attributes selected per brand, FCB and PA attribute prompted –

Butter/margarine. ................................................................................................................................. 105 Table 61: Average number of attributes selected per brand, FCB and PA brand prompted –

Butter/margarine. ................................................................................................................................. 105 Table 62: Average number of brands selected per attribute for light buyers - Cereal. ................................ 106 Table 63: Average number of brands selected per attribute for light buyers - Butter/Margarine. ............... 106 Table 64: Average proportion of non, light and heavier buyer cereal response, PA attribute prompted. .... 108

List of Figures Figure 1: Percentage of breakfast cereal buyers purchasing Brand A x times, 2010. ..................................... 6 Figure 2: Percentage of shampoo buyers purchasing Brand A x times, 2010. ................................................ 6 Figure 3: Larger brands approach the category distribution. ........................................................................... 8 Figure 4: Percentage of toothpaste buyers purchasing Brand A x times, 2006 and 2008. ............................. 9 Figure 5: A representation of memory structure under the ANT of memory. ................................................. 14 Figure 6: Example of a prompted awareness question, fast food category. ................................................. 22 Figure 7: Buyer and Non-buyer TOM awareness hypothesis development. .................................................. 26 Figure 8: Buyer and Non-buyer Prompted awareness hypothesis development. ......................................... 28 Figure 9: Light buyer awareness hypothesis development. ........................................................................... 29 Figure 10: Example of a pick-any image measurement question, cereal category. ...................................... 71 Figure 11: Example of a forced-choice binary image measurement question, laundry detergent category. 72 Figure 12: Example of a pick-any, brand prompted image measurement question, cereal category. .......... 73 Figure 13: Example of a forced-choice binary, attribute prompted image measurement question, laundry

detergent category. ............................................................................................................................... 74 Figure 14: Representation of brand and attribute cue activation. .................................................................. 79 Figure 15: Representation of strength between brand and attribute linkages. .............................................. 80 Figure 16: Pick any brand prompted image question. .................................................................................. 88 Figure 17: Pick any attribute prompted image question. .............................................................................. 88 Figure 18: Forced choice binary brand prompted image question. .............................................................. 89 Figure 19: Forced choice binary attribute prompted image question. .......................................................... 90

Page 10: Understanding and Measuring Light Buyer Brand Equity

1

Chapter 1 INTRODUCTION This chapter provides an overview of the research rationale and objective. A description of the research approach is stated, along with contributions to marketing knowledge and practice. The chapter concludes with an outline of the thesis structure.

1.1 Research Background

1.1.1 Customer Based Brand Equity Marketing courses, textbooks and academic literature have a commonality: to help

managers gauge the market performance of brands. One tool that marketers use to assess the performance of a brand and the potential for brand growth is via Customer Based Brand Equity (CBBE) surveys. These surveys consist of different measures designed to gauge the relationship between the brand and consumer memory structures over time (Aaker 1992; Keller 1993).

Tracking a brand’s equity is an essential tool for market research departments, providing insight into areas such as consumer perceptions and attitudes, along with evaluating brand knowledge and purchase behaviour in relation to competitor brands (Christodoulides & De Chernatony 2010; Farquhar 1989; Keller 2005). Results from brand equity surveys inform and are used to evaluate marketing strategies, along with assessing the performance of market practitioners themselves (i.e. reaching KPI sets). Given the importance of these results, research often seeks to better understand and assess measures included in equity surveys.

One area that has received attention is analysing brand equity measures by user and non-user (e.g., Barwise & Ehrenberg 1985; Bird, Channon & Ehrenberg 1970; Bird & Ehrenberg 1966; Driesener & Romaniuk 2006; Romaniuk, Bogomolova & Dall'Olmo Riley 2012; Romaniuk & Wight 2009). Findings from these studies confirm that the two groups of category buyers respond to brand equity measures in a different manner. Wight (2010, p.13) suggests that by analysing equity scores separately, “practitioners will be able to more accurately identify the success of marketing efforts and evaluate the effectiveness of targeted activities”.

Page 11: Understanding and Measuring Light Buyer Brand Equity

2

1.1.2 How Brands Grow A brand’s customer base can be classified into different buyer1 categories according to

their purchase frequency, which is how often the brand has been purchased in a given time period (Ehrenberg 2000a).

Traditionally marketers believe that brand growth can be sourced from heavier buyers, who buy and/or use the brand often, generating more sales and profit per customer (Anschuetz 2002). At the aggregate level, however, buyers who purchase the brand infrequently, i.e. light buyers, are a valuable buyer group for brand growth. Light buyers make up the largest portion of a brand’s customer base (Ehrenberg 1959; Sharp 2010) and contribute around half of brand sales (Schmittlein, Cooper & Morrison 1993). As a brand grows it acquires new buyers and its existing buyers purchase (slightly) more often, increasing both penetration and loyalty (Baldinger, Blair & Echambadi 2002; Ehrenberg, Goodhardt & Barwise 1990). Romaniuk (2011) explains that as a brand gains buyers, this gain is in line with the distribution of buyers it had the in previous period. Light buyers make up the largest portion of a brand’s customer base, indicating that most of the buyers acquired will be light brand buyers (Anschuetz 2002; Ehrenberg, Goodhardt & Barwise 1990).

Given the number of light brand buyers in a brand’s customer base and light brand buyers being the major source of brand growth, understanding how this buyer group responds to brand equity measures is of importance. Assessing where and how light buyers respond in brand equity measures will help to better understand changes in the brand over time.

1.2 Research Aim

The purpose of this research is to better understand and measure light buyer brand

equity. Thus, the overarching research question that this thesis aims to answer is:

How do light buyers respond to brand equity measures?

1 ‘User’ is consistent with prior studies, however this thesis examines brand ‘buyers’. The author acknowledges that a person may purchase the brand but not use/consume it and alternatively that a person may use/consume the brand not having purchased it, i.e. as a gift. However, for the purpose of this thesis, the two terms, user and buyer, are used interchangeably.

Page 12: Understanding and Measuring Light Buyer Brand Equity

3

Two brand equity measures are analysed: brand awareness and brand image. All CBBE surveys contain brand awareness measures as once a person is aware of a brand information can be linked to the brand in memory (Keller & Davey 2001). This information is measured via brand image, which helps identify the current position of the brand. The measure can also be used to evaluate advertising effectiveness in terms of its ability to reinforce or promote the brand’s image (Kapferer 2008). Therefore, it is important to assess the measures from a light buyer perspective. The concepts, brand equity, brand awareness and brand image, are defined and discussed in further detail in following chapters.

The results from this research will contribute to the following areas:

Academia – Results will give new insight into light brand buyer memory structures and

information retrieval. Results will also extend current research on user and non-user response, in turn expanding knowledge on brand equity and its measurement.

Market Practitioner – Results from this thesis will determine which measure(s) should

be utilised to assess light buyer brand knowledge. This research will highlight potential pitfalls when measuring brand performance. Results will provide guidance on equity tracker questionnaire design and data analysis. Overall, this research will lead to more informed strategies in order to grow the brand and provide a more detailed assessment of the brand’s performance.

1.3 Research Method

The categories selected for analysis are repertoire markets, where buyers purchase from multiple brands (Ehrenberg 2000a). The research approach and findings are presented in two sections: Part One: Brand Awareness and Part Two: Brand Image. A brief summary of the research method for each brand equity measure is provided.

Part One: Brand Awareness

Secondary data is used to compare light brand buyer response to three measures of brand awareness (top of mind, unprompted and prompted) for over 10,000 respondents, 52 brands, four product categories and three countries. Awareness is analysed for different brand buyers, placing light buyer awareness response in context of the brand’s entire (and potential, i.e. non-buyers) customer base. A key strength of Part One is analysing matched panel and brand equity survey data, where purchase data and response to awareness questions are available for the same individual.

Page 13: Understanding and Measuring Light Buyer Brand Equity

4

Part Two: Brand Image

Part Two replicates Joyce’s (1963) research, comparing two image measures (pick any and forced choice binary) and for two methods of prompting for response (attribute and brand). The present study extends Joyce’s work to examine light buyer response. Primary data is collected via an online questionnaire in the United Kingdom, testing two product categories: breakfast cereal and butter/margarine. A split sample is used during data collection, where each participant answers questions for two different image measurement approaches and two different product categories. Part two has a sample size of over 2,000 respondents, evenly divided between four sample groups for comparison of measures and prompted method.

1.4 Thesis Structure

This thesis will first introduce and explain the importance of light brand buyers in

Chapter Two. To understand the context of this research, brand equity is detailed in Chapter Three, along with a discussion of consumer memory structures and information retrieval.

Part One: Brand Awareness

Brand awareness is explained and hypotheses are developed in Chapter Four. Chapter Five details the research approach, including data sets, categories analysed and data analysis. Finally, Chapter Six presents the results for the brand awareness, in addition to a discussion of key findings.

Part Two: Brand Image

Part two comprises Chapters Seven to Nine. Brand image is first introduced and hypotheses to be tested are stated in Chapter Seven. Chapter Eight discusses the approach taken for data collection and the analysis process. Findings are presented and discussed in Chapter Nine.

Chapter Ten concludes this thesis. Research contributions, strengths and limitations of the study are detailed, before suggesting avenues for future research.

Page 14: Understanding and Measuring Light Buyer Brand Equity

5

Chapter 2 LIGHT BRAND BUYERS This chapter introduces and explains the importance of light brand buyers.

2.1 The Importance of Light Brand Buyers

Those who purchase the brand frequently are often considered a more valuable buyer

than those who purchase infrequently (Anschuetz 2002). At the individual level, infrequent, lighter buyers are overlooked in favour of more frequent, heavier buyers, whose more frequent purchases contribute more to brand sales and revenue. However, at the aggregate level, light buyers make up the largest portion of a brand’s customer base (Ehrenberg 1959) and contribute around half of brand sales (Schmittlein, Cooper & Morrison 1993).

The light buyer group is also a major source of brand growth. Brands grow by gaining more buyers and this gain is in line with the distribution of buyers it had the previous period. This indicates that most of the buyers gained will be light brand buyers (Ehrenberg, Goodhardt & Barwise 1990; Romaniuk 2011).

Thus, it is the sheer number of buyers, their value in terms of overall sales and importance to brand growth that makes light buyers a valuable buyer group. The following sections expand on these points.

2.1.1 Brand Customer Base Structure

Ehrenberg (1959) describes patterns of category buyer purchases, showing the percentage of buyers who purchase the brand infrequently, frequently and not at all in the time period. Figures 1 and 2 demonstrate this distribution using examples from Kantar TNS household panel data. Those who purchase the brand once in the timeframe make up the largest portion of buyers in the brand’s customer base, i.e. the light brand buyers. This distribution is skewed with many buyers who are lighter than average and a long tail of few heavy buyers resembling a reversed J-shape (Ehrenberg 1959, 1988).

Page 15: Understanding and Measuring Light Buyer Brand Equity

6

Figure 1: Percentage of breakfast cereal buyers purchasing Brand A x times, 2010.

Source: TNS Superpanel, UK

Figure 2: Percentage of shampoo buyers purchasing Brand A x times, 2010.

Source: TNS Superpanel, UK

Romaniuk (2011) uses Kantar Worldwide household panel data to analyse the buying distribution of 93 brands across 10 categories. Results show that light buyers consistently make up at least 50% of brand purchases, regardless of whether the brand grows, declines or remains stable. Therefore, the number of light brand buyers makes this an important buyer group.

Page 16: Understanding and Measuring Light Buyer Brand Equity

7

2.1.2 Brand Sales

In Figures 1 and 2, to overlook light buyers and focus purely on heavier, ‘more profitable’ buyers would mean neglecting 80-90% of the brands customer base. Empirical evidence (e.g. Schmittlein, Cooper & Morrison 1993; Sharp & Romaniuk 2007)

find that this large group of light buyers contribute between 40-60% of brand sales, while the top 20% of brand buyers, i.e. heavier buyers, make up the remaining sales.

Empirical evidence builds on knowledge of the Pareto Law (or 80:20 rule), which claims that heavier buyers account for almost the entirety of brand sales, with 80% of the brand’s sales coming from the top 20% of brand buyers (Persky 1992). Theorised in 1897 by Italian economist Vilfredo Pareto (1848-1923) (Koch 1999), the law suggests that marketers should target the heavy buyer segment to increase brand profitability (e.g., Goldsmith, Flynn & Bonn 1994; Light 1994; Twedt 1964). Schmittlein et al. (1993) and Sharp and Romaniuk’s (2007) findings highlight that, at the aggregate level, light buyers have value in terms of the brand’s overall sales and should not be neglected for heavier buyers.

2.1.3 Brand Growth via Penetration and Loyalty

The Negative Binomial Distribution (NBD) models buying distributions of individual brands and is “a useful planning model for understanding consumer contributions to growth” (Anschuetz 2002, p.16). The NBD model reflects the reversed J-shape distribution shown above, where there are many buyers who are lighter than average and a long tail of few heavy buyers. Although the NBD model is often applied to category purchase frequency (Ehrenberg 1959; Ehrenberg, Uncles & Goodhardt 2004; Goodhardt & Ehrenberg 1967; Uncles, Ehrenberg & Hammond 1995), there is evidence that the NBD also works at the brand level (for examples see Ehrenberg 2000a; Schmittlein, Bemmaor & Morrison 1985).

As a brand grows it acquires more buyers and their buyers purchase (slightly) more often, increasing both penetration and loyalty (Baldinger, Blair & Echambadi 2002; Ehrenberg, Goodhardt & Barwise 1990; Sylvester, McQueen & Moore 1994). As brand share increases, the buyer distribution moves from one NBD closer to the category distribution of buying (refer Figure 3) (Ehrenberg 1988; McDonald & Ehrenberg 2003). Larger share brands therefore skew slightly more to heavier buyers but still have the majority of their customer base buying infrequently.

Page 17: Understanding and Measuring Light Buyer Brand Equity

8

Figure 3: Larger brands approach the category distribution.

Source: Anschuetz (2002, p.17)

Studies using longitudinal data find that brand share growth is greater for penetration than for loyalty. For example, Sylvester et al. (1994) analysed 95 brands across 14 categories using IRI and Nielsen US data and found that for 67% of brands penetration was the major contributor to growth. Similarly, Anschuetz (2002) found a large increase in penetration (36%) compared to a smaller increase in the frequency of buying (16%) for a large dairy brand, which grew 57 per cent over a four year period. Baldinger et al. (2002) provide another example across 353 Canadian packaged goods brands, measured at two points in time, 1992 and 1997. Results show a higher correlation of change in penetration (0.83) with change in share than a correlation with change in

loyalty (0.59). Stern and Ehrenberg (2003) analysed doctors’ Prozac prescriptions over eight years (1889-1997), which showed a 20-fold increase in market share (from 1% to 21%). During this time, the percentage of doctors prescribing the medication increased 10-fold (from 6% to 67%), while prescription frequency only doubled (from 1.8 to 3.8). Finally, in a more recent study, Romaniuk et al. (2014) examined 63 packaged goods categories from 2003 to 2007 using Kantar TNS’s UK consumer panel. Their findings indicate that as market share increases, penetration changes proportionally 2.8 times more than loyalty. These patterns in brand growth indicate that it may be easier to attract new buyers to purchase the brand once (light buyers) than to convince buyers who are already spending money on the brand to purchase more often (heavy buyers).

0

2

4

6

8

10

12

14

16

18

20

0 2 4 6 8 10 12 14 16 18 20

% Households

Frequency of Buying

Category)

18)Share)

7)Share)

4)Share)

3)Share)

Page 18: Understanding and Measuring Light Buyer Brand Equity

9

Romaniuk (2011) explains that as a brand gains buyers, this gain is in line with the distribution of buyers it had the previous period. Light buyers make up the largest portion of a brand’s customer base, indicating that most of the buyers gained will be light brand buyers. For example, Anschuetz (2002) looks at two brands that differ in market share in the hair care category. He finds a 76% difference in the number of light buyers between a four and seven share brand, compared to a 5% difference for heavier buyers. Using Nielsen US data, Figure 4 provides an example of the prominent increase in light brand buyers (from 6.4% to 8.1% for those buying the brand once) compared to frequent buyers (from 0.1% to 0.3% for those buying the brand four times) as market share increases. The large increase in the light buyer segment can be attributed to non-buyers purchasing the brand at least once in the timeframe of interest, entering the light buyer segment.

Figure 4: Percentage of toothpaste buyers purchasing Brand A x times, 2006 and 20082.

Source: Nielsen, US

2.2 Light Brand Buyer Classification

Brand buyers may be defined according to two measures: share of category requirements and purchase frequency. Share of category requirements (SCR), or share loyalty, is calculated as the frequency of brand purchase divided by the frequency of category purchase (Ehrenberg 2000a; Uncles et al. 1994).

2 Brand A is a small market share brand and as such has a high percentage of non-buyers. This is reflected by a low percentage of buyers in the graph.

Page 19: Understanding and Measuring Light Buyer Brand Equity

10

For example, Sarah purchases from the category 12 times and from Brand A three

times in a given period. Her SCR is 25% (3/12=0.25), where the higher a person’s SCR the more loyal they are considered to the brand.

A limitation of using SCR for buyer classification is the minimum number of category purchases required to classify light buyers. For instance, if a light buyer is defined as someone who has less than 25% share loyalty, they must purchase from the category a minimum of four times. As the number of category purchases decreases, the buyer’s SCR increases (e.g. 1 brand purchase / 3 category purchases = 33%; 1/2 = 50%), despite purchasing the brand once in both cases. Furthermore, using SCR is an inappropriate classification if customer value is of interest, as ‘once only’ category buyers would have low sales value but high SCR.

The second measure to classify light brand buyers is by their number of brand purchases made in a given timeframe. In line with consumer packaged goods studies (e.g., Anschuetz 2002; Ehrenberg 2000a; Ehrenberg, Uncles & Goodhardt 2004), light buyers are classified in this study as having purchased the brand once in a given period.

2.2.1 Brand and Category Purchase Frequency

Depending on the number of category purchases made in a given period, a light brand buyer can fall into one of two categories (Ehrenberg 2000a):

• Light brand buyer and light category buyer

• Light brand buyer and heavy category buyer

Similarly, a heavy brand buyer can be a heavy category buyer. It is unlikely however that a heavy brand buyer is a light category buyer and this becomes impossible if the category is purchased fewer times than specified in the heavy brand buyer classification (Romaniuk & Wight 2014). For example, a buyer cannot purchase the brand twice, but from the category only once.

The introduction of different brand and category buyer groups is important when considering buyers’ memory structures. Discussion of different types of brand and category buyers will continue in the following chapters.

Page 20: Understanding and Measuring Light Buyer Brand Equity

11

2.3 Chapter Summary

This chapter has introduced and explained the importance of light brand buyers. Key

take-outs from this chapter are:

• Despite being less profitable individually, together the total number of light brand buyers contributes substantially to brand growth and revenue.

• Light brand buyers make up the largest portion of buyers a brand has and account for 40-60% of brand sales.

• As brand growth occurs primarily by gaining buyers in line with the brand’s buyer distribution, it is inevitable that a brand will grow by acquiring more light buyers than heavy buyers.

The following chapter explains the need for research on light buyers in a brand equity context.

Page 21: Understanding and Measuring Light Buyer Brand Equity

12

Chapter 3 BRAND EQUITY & CONSUMER MEMORY This chapter defines brand equity, in particular Customer Based Brand Equity (CBBE). An overview of consumer memory structures is given and how this differs for brand

buyers and non-buyers.

3.1 What is Brand Equity?

There are two streams of brand equity research identified in the literature: Firm or

Financial Based Brand Equity (FBBE) and Consumer Based Brand Equity (CBBE). FBBE is well documented in the literature and is used for areas such as stock replacements and return on investment studies (Farquhar 1990; Myers 2003; Simon & Sullivan 1993). This research focuses on Customer Based Brand Equity.

Many definitions of CBBE exist, with Keller (2005, p.4) noting that “all definitions typically implicitly or explicitly rely on brand knowledge structures in the minds of consumers as the source or foundation of brand equity”. This thesis draws on Farquhar’s (1989) definition of brand equity, described as the value that the brand provides the product and/or service. Aaker (1992) explains that this value is created from key brand equity assets, including brand name awareness, perceived brand quality and brand associations – the thoughts, words and actions of consumers (Keller 2005). Brand equity assets can help customers interpret, process, store and retrieve information about a given brand (Aaker 1992; Christodoulides & de Chernatony 2010).

Marketers aim to develop strong brand equity through marketing actions and activities. A number of benefits are associated with having strong brand equity, including:

• Building store traffic, ensuring consistent volume and reducing risk in allocating shelf space (Cobb-Walgren, Ruble & Donthu 1995; Crimmins 1992).

• Strengthening brand loyalty through increased awareness and associations with

the brand (Neal & Strauss 2008).

• Allowing firms to more effectively employ other elements of the marketing mix (e.g. advertising and promotions) (Keller & Lehmann 2009).

• Providing a platform for brand growth and extension (Farquhar 1989).

Page 22: Understanding and Measuring Light Buyer Brand Equity

13

3.1.1 Brand Equity and Light Buyers Brand equity is assessed via a combination of different measures, which provide an

indication of the performance and potential of the brand that can be tracked over time (Keller 1993). Brand equity measures should be sensitive; when brand equity changes from brand growth or decline, the measures should detect that change (Aaker 1996b). This change is best detected by examining light buyers as more buyers enter this segment during brand growth, as detailed in Chapter Two. Aggregating over different groups of buyers (i.e. non, light and heavier buyers) will provide a less sensitive measurement, clouding the strategic interpretation of the results (Aaker 1996a).

Past studies have tested measures, such as brand advertising awareness and brand image (e.g., Barwise & Ehrenberg 1985; Bird & Ehrenberg 1966), but this is for user and non-users of the brand and category. This thesis, which examines the brand equity of different brand buyers, particularly light buyers, addresses the call in the literature for further research into the ‘relative strength of band equity by type of user’

(Christodoulides & De Chernatony 2010; Knox & Walker 2001; Romaniuk 2013).

3.2 Customer Based Brand Equity Measures

CBBE is measured by looking at the network of associations linked to the brand in

consumer memory (Keller 1993). These mental associations are assessed over time using brand equity surveys (trackers) (Kapferer 2008). Brand equity surveys are important for brand managers, either producing information that can be used to make an assessment of the likely future of the brand or monitoring the health of the brand to detect indications of weakness (Aaker 1972; Keller 2005).

Literature on brand equity detail a number of measures used for testing, creating confusion as to which ones should be included in equity surveys. According to Keller

(1993), brand knowledge is a key antecedent of CBBE and is composed of two separate constructs: brand awareness and brand image (associations). This aligns with Farquhar’s (1990) comments stating that elements essential in building a strong brand with the consumer include: positive brand evaluations, accessible brand attitude and a consistent brand image. Similarly, Agarwal and Rao (1996) develop a framework that provides measures linked to the stages a consumer passes through when making a purchase: 1) awareness of the brand, 2) perceptions of the brand, assessed through information linked to the brand in memory (associations), and 3) preference, judged as the overall evaluation of the brand.

Page 23: Understanding and Measuring Light Buyer Brand Equity

14

The focus of this study is not to determine which measures should be included in brand equity surveys, rather how light buyers respond to these measures. For this reason two measures are selected that are commonly used in brand equity surveys and are highly cited in literature: brand awareness and brand image. Further explanation of these measures appears in the following chapters.

3.3 Consumer Memory Structures

Brand equity measures require respondents to draw on their knowledge of brands and

categories, highlighting the need for marketing academics and practitioners to understand the structure and retrieval of information from memory.

3.3.1 Associative Network Theories of Memory The Associative Network Theories of Memory (ANT) explain the structure of consumer

memory networks and retrieval of brand information from these structures (e.g., Anderson & Bower 1979). The ANT propose that information is represented as nodes connected via linkages, forming a complex network (see Figure 5).

From a branding perspective, information contained within memory structures encompasses brand knowledge and is reflected as the strength of a brand node in memory and the consumer’s ability to retrieve this information in different conditions (Keller 1993). Links to the brand are created and maintained through exposure to the brand (Nedungadi 1990). Exposure can occur via purchase or consumption, advertising, or word of mouth.

Figure 5: A representation of memory structure under the ANT of memory.

Source: Hogan (2012)

Page 24: Understanding and Measuring Light Buyer Brand Equity

15

3.3.2 Information Retrieval from Memory The Spreading Activation Theory of Memory (SAT) explains that a piece of information is

first ‘activated’ and retrieved, where activation then spreads out along paths of the networks to activate further nodes for retrieval (Anderson & Bower 1979; Collins & Loftus 1975). As more connections are activated, the more the activation is divided and the weaker retrieval becomes, as per the fan effect (Anderson 1974; Heil, Rösler & Hennighausen 1994).

Retrieval is cue dependent, where the cue given to consumers determines the ability of information retrieval from the network. The degree to which one node will activate another for retrieval is a function of the strength of the link between the two nodes (Collins & Loftus 1975), where linkages are strengthened via the frequency and recency of exposure to the cue (Anderson 1983). Over time, the strength of links between nodes that are not encountered frequently and/or recently will decrease, making it harder for this information to be retrieved from the network (Tulving & Pearlstone 1966).

In brand equity measures, respondents are prompted with a category and/or brand cue for activation and retrieval of information in memory. The more frequently and/or recently a brand name co-occurs with the category and/or brand information, the stronger the linkage between the brand name and information will become (Van Osselaer & Janiszewski 2001). As linkage strength increases, retrieval from memory becomes less difficult, thus brands encountered more frequently/recently will be recalled over brands encountered infrequently (Nedungadi & Hutchinson 1985).

3.3.3 Light Buyer Memory Structure As discussed in Chapter Two, buyers are classified according to their purchase frequency (PF). As links in memory to the brand are developed from exposure, including prior purchases and marketing efforts, this suggests that light buyers hold less brand knowledge than heavy buyers in memory. In addition, having fewer and weaker linkages between the brand and information (or category and brand name) makes it harder for light buyers to retrieve this brand knowledge during brand equity measures than for heavy buyers (as per frequency and recency effects, Anderson 1983; Nedungadi & Hutchinson 1985).

Page 25: Understanding and Measuring Light Buyer Brand Equity

16

Non-buyers are only exposed to the brand’s marketing activities and word of mouth. As such, non-buyers will have less brand information stored in memory and will have more difficulty retrieving this information when cued. Though infrequent, light buyer brand purchase increases the amount of brand knowledge stored in memory, resulting in more information accessible than non-buyers.

Romaniuk and Sharp (2008) note that a brand’s total knowledge (total number of associations held by all respondents) is distributed according to a heavy half and a light half. The top 20% of buyers hold almost half of the brand’s total knowledge, while the remaining 80% of buyers hold the other 50% of brand knowledge. This highlights the need for brand equity measures to effectively access and retrieve brand information from light buyers – “many of whom do not recall a great deal of knowledge about the brand and aren’t far away from forgetting the brand altogether” (Romaniuk & Sharp 2008, p.3).

For each brand equity measure, discussion of expectations for light brand buyer (and other buyer and non-buyer) responses will stem from memory concepts introduced in this chapter.

3.4 Chapter Summary

This chapter has presented a background on Customer Based Brand Equity and

consumer memory networks. Key take-outs from this chapter are:

• CBBE is the network of associations linked to the brand in consumer memory.

Brand equity surveys assess the performance and potential of the brand relative to competitors via a combination of different measures.

• This research extends previous studies that test brand equity measures for users and non-users to examine buyers and non-buyers in line with how brands grow (i.e., through light brand buyers).

• Light buyers have less brand information in memory that is harder to retrieve

than heavy buyers.

Two brand equity measures are selected as a focus of this research: brand awareness and brand image. Subsequent chapters will discuss these measures, with a specific focus on light brand buyer ability to respond.

Page 26: Understanding and Measuring Light Buyer Brand Equity

17

Part One: BRAND AWARENESS Part one comprises Chapters Four to Six. Chapter Four introduces brand awareness and the hypotheses this research will test. The research approach is discussed in Chapter

Five, with results presented in Chapter Six.

Page 27: Understanding and Measuring Light Buyer Brand Equity

18

Chapter 4 BRAND AWARENESS This chapter examines brand awareness, its importance and measures included in brand equity surveys. The chapter draws on memory literature to state the hypotheses this

research will test, which are presented separately for each of the three awareness measures: top of mind, unprompted and prompted awareness.

4.1 What is Brand Awareness?

As other aspects of brand equity cannot be formed without first being aware of the

brand (Keller & Davey 2001), awareness is the first measure discussed. Once a brand is known, a person can form associations with the brand (Nedungadi 1990), detailed in Part Two: Brand Image.

Brand awareness is a person’s ability to recall and/or recognise a brand is a member of a product category (Bird & Ehrenberg 1966; Percy & Rossiter 1992). Anything that causes the consumer to experience (through purchase and/or consumption) or be to exposed to the brand has the potential to create brand awareness (Keller 1993). This exposure can occur from aspects such as its presence in store (passing the brand on the supermarket shelf), brand communication efforts (seeing advertising at home or out of home) and/or through word of mouth (someone mentions the brand online or in person) (Aaker 1991; Huang & Sarigöllü 2012).

Brand awareness provides an anchor in memory from which nodes can be connected (Keller 1993), as per the Associative Network Theories of Memory (Anderson & Bower 1979). For example, once aware of the brand name ‘Starbucks’, information can be linked to the name in memory. This can include descriptive features (Lefkoff-Hagius & Mason 1993), such as ‘serves coffee’ or ‘café’, situational information (Ratneshwar & Shocker 1991), such as ‘the place I meet my friends’, or overall evaluations (Farquhar 1990), such as ‘a brand I like’.

4.2 The Importance of Brand Awareness

For marketers, it is important to build brand awareness to be considered for purchase (Nedungadi & Hutchinson 1985). When making a purchase, consumers do not consider all brands in the category. Instead, information on a select number of brands is evoked from consumer memory for comparison (Howard & Sheth 1969; Macdonald & Sharp 1996). Brand awareness also offers the brand a competitive advantage as retrieval of

Page 28: Understanding and Measuring Light Buyer Brand Equity

19

one brand from memory may inhibit the retrieval of other brands because the strength of activation decreases as more information is retrieved (Alba & Chattopadhyay 1986).

From a consumer perspective, brand awareness aids the decision making process when selecting a brand for purchase and/or consumption. In an attempt to minimise the time taken to make a decision for purchase, many buyers resort to selecting known brands over unknown brands (Macdonald & Sharp 1996). Hoyer and Brown (1990) demonstrate this by analysing in a controlled experiment the influence of brand awareness on purchase decisions for a common repeat purchase product, peanut butter spread. The authors find that 94% of respondents select the known brand out of three options when asked to make a decision for purchase. The study was conducted using non-buyers and light brand buyers, concluding that brand awareness is a prevalent choice tactic for inexperienced consumers or alternatively, for low involvement decisions where little thought is given to the purchase (Hoyer & Brown 1990).

Macdonald and Sharp (2000) later replicate and extend Hoyer and Brown’s (1990) study, using the cordial fruit drink category and sampling experienced and

inexperienced buyers. Experienced buyers were defined as those who buy cordial at least once every few months, i.e. heavy buyers, and inexperienced as those who had never bought cordial or only a few times at most, i.e. non-buyers or light buyers. Results are similar to Hoyer and Brown (1990), with 86% of respondents selecting a known brand over an unknown brand. Macdonald and Sharp (2000) also find that consumers presented with at least one known brand make purchase decisions quicker than consumers where there are no known brands (9.8 seconds compared to 15.1 seconds). Brand awareness therefore eases the decision making process, helping buyers make a purchase decision more efficiently.

The studies mentioned above highlight the role brand awareness plays in choice situations. However, it is important to note that the studies discussed analyse respondent decisions when they are aware of one brand and unaware of another. In purchase situations, a person can be aware of multiple brands, which may affect the decision process undertaken.

Page 29: Understanding and Measuring Light Buyer Brand Equity

20

4.3 Brand Awareness Measurement

Three awareness measures identified in marketing literature are: top of mind (TOM),

unprompted and prompted awareness (e.g., Bird & Ehrenberg 1966; Laurent, Kapferer & Roussel 1995; Romaniuk & Wight 2009). Brand awareness measures rely on recall or recognition for retrieval of brand information from memory (Lynch & Srull 1982; Percy & Rossiter 1992). The three awareness measures tap into consumer semantic memory to assess the brand names held in memory networks (Romaniuk et al. 2004). Semantic memory is a highly structured network of concepts, words and images, detached from reference to events or episodes in time (Tulving 1972; Tulving & Thomson 1973). An explanation of the differences between recall and recognition (prompted brand) awareness measures is provided, along with a definition and example of the three awareness measures. Information on the research scope follows.

4.3.1 Brand Recall Brand recall measures use the category as a cue to retrieve brand names linked in memory (Nedungadi & Hutchinson 1985; Zinkhan, Locander & Leight 1986). For example:

Which brands come to mind when you think of fast food?

Bird and Ehrenberg (1966) note that when constructing recall awareness measures, time should be spent considering how the category is presented to respondents. The way the category is described can influence response overall and for different brands. In the example provided, respondents may link unhealthy brands with ‘fast food’, therefore excluding brands in the category considered as ‘healthier’ options.

Respondents recall brand names until they feel that the question has been sufficiently answered or until a consecutive number of recall attempts produces no previously unrecalled brands, whichever occurs first (Alba & Chattopadhyay 1986; Holden 1993). Past studies rarely specify the time taken and/or given to respondents for recall or if a maximum number of recall attempts are permitted. One exception is Nedungadi and Hutchinson (1985) who allow respondents five minutes to name all brands that come to mind when cued with the product categories, magazines and beverages.

Recall measures activate an initial cue (category name) in memory structures and require the respondent to search and retrieve nodes (brand names) linked to this cue (Holden 1993). Brands that have strong links in memory to the category cue will be recalled first, with respondent ability to recall decreasing with each attempt, as per the

Page 30: Understanding and Measuring Light Buyer Brand Equity

21

fan effect (Anderson 1974; Heil, Rösler & Hennighausen 1994). Recall measures therefore indicate to marketers which brands are not only stored (available) in the minds of consumers but those that are readily retrieved (accessible) (Romaniuk & Wight 2009).

Lynch and Srull (1982, p.20) explain that “once information is fully comprehended and encoded into long-term memory, it is thought to always be available”. However, a respondent’s ability to retrieve this information is dependent on the strength of the cue used for activation (Tulving & Pearlstone 1966). For example, as consumers encounter different brands, this information is encoded in memory networks and becomes ‘available’ for retrieval. Over time, brands purchased frequently and recently form strong linkages between the brand and category cue, while links for brands not encountered begin to weaken (Anderson 1983; Van Osselaer & Janiszewski 2001). When provided with the category cue and asked to recall brands linked in memory, brands with stronger links are accessible and easily retrieved, while those purchased infrequently and/or not recently are difficult to retrieve and are not accessible from memory. Tulving and Pearlstone (1966) note that inferences about what is available in memory cannot be made on the basis of what is accessible. Therefore, while some brands may be recalled, there are most likely other brands linked to the category cue in memory that are not captured in recall awareness measures.

Awareness can be assessed according to two recall measures:

Top of mind (TOM) awareness, which provides respondents with a category cue and

asks respondents to name the brands they know. TOM is measured as the first brand recalled (Gruber 1969).

TOM awareness is often used interchangeably with brand salience, defined as the prominence or ‘level of activation’ of the brand in memory (Alba & Chattopadhyay 1986). Therefore a brand recalled first is thought to have a higher position in the consumer’s mind, increasing the probability of selection in choice situations (Nedungadi 1990)3.

3 Romaniuk and Sharp (2004, p.327) build on salience definitions that simultaneously refer to TOM awareness, defining the concept as the “propensity of the brand to come to mind in buying situations, reflecting the quality and quantity of the network of memory structures buyers’ hold about the brands”.

Page 31: Understanding and Measuring Light Buyer Brand Equity

22

Unprompted awareness4 (or unaided awareness) is analysed using data drawn from

TOM awareness questions, measured as all brands recalled when provided with a

category cue (Bird & Ehrenberg 1966).

In purchase/consumption situations, buyers often aim to fulfil a goal or need, which evokes a brand, or several brands, from memory to make a decision (Macdonald & Sharp 1996; Percy & Rossiter 1992). Unprompted awareness measures all brands recalled from the category cue, providing a measure of brands accessible from memory. Many of these brands could be included in the respondent’s consideration set depending on the cue used for activation and retrieval. “While they might be able to name 20 brands of chocolate bars when they are probed, rarely in ‘real life’ will they think of more than a few at any one time” (Romaniuk & Sharp 2004, p.332).

4.3.2 Prompted Brand Awareness Prompted awareness (or aided awareness) is measured as the respondent’s ability to

confirm prior exposure when given the brand as a cue (refer Figure 6) (Keller 1993; Laurent, Kapferer & Roussel 1995).

Figure 6: Example of a prompted awareness question, fast food category.

Please indicate the fast food brands you are aware of.

☐ McDonalds ☐ Hungry Jacks ☐ Kentucky Fried Chicken (KFC) ☐ Subway ☐ Dominos Pizza ☐ None of these

The prompted awareness measure provides the respondent with brand and category cues, where the respondent states whether or not the brand is present in their memory structure. The brand name is the first node to be activated, rather than activation spreading from category to brand as in recall measures. Prompted awareness is therefore considered less difficult than recall measures (Laurent, Kapferer & Roussel 1995; Romaniuk & Wight 2010).

4 There is variability in the terminology used for recall measures. Bird and Ehrenberg (1966) and Laurent et al. (1995) use the term ‘spontaneous awareness’ to describe all brands recalled. This thesis uses terminology consistent with Romaniuk and Wight (2009) and Wight (2010) study, who analyse user and non-user awareness: top of mind, unprompted and prompted awareness.

Page 32: Understanding and Measuring Light Buyer Brand Equity

23

The brand prompted measure overcomes limitations of recall measures, particularly when respondents may have seen the brand but not processed the exposure deeply enough to recall it from the stimulus provided (Zinkhan, Locander & Leight 1986). Lynch and Srull (1982) state that recognition almost always produces a higher response than recall. The authors explain that recall is often a two-stage process, where the respondent must first retrieve the information from memory, then make a judgement about whether the item answers the question posed. Whereas, recognition merely involves a ‘discrimination’ check to assess whether the brand is present in their memory network. Thus, the time taken to formulate an answer is reduced, along with the lower level of cognitive effort required, permitting responses for more brands than recall measures.

4.3.3 Research Scope Over time, research has examined brand awareness and buyer behaviour (e.g., Assael &

Day 1968; Bird & Ehrenberg 1966; Hutchinson, Raman & Mantrala 1994; Nedungadi & Hutchinson 1985). Sharp, Beal and Romaniuk (2001) take the initial steps into quantifying the link between usage and awareness. From an advertising awareness perspective, the authors determine that users are at least twice as likely than non-users5 to recall a brand and that any brand’s advertising recall scores tends to be around 20 percentage points higher for users than for non-users. Romaniuk and Wight (2009)

similarly test advertising awareness scores for users and non-users of the personal financial services industry. Findings again show that brand users are more likely than non-users to recall (2.5 times more likely), and recognise the brand (1.7 times).

Wight (2010) examines brand awareness for users and non-users outside of an advertising context, testing whiskey, toothpaste and hair care categories. His findings support Sharp, Beal and Romaniuk (2001) and Romaniuk and Wight’s (2009) results. Brand user recall and recognition levels are higher than non-users, with the difference between scores decreasing from TOM to prompted awareness measures. Users are 18 times more likely to respond than non-users to TOM, 2.7 times to unprompted, and 1.1 times to prompted awareness.

5 As stated in Chapter One, previous research looks at ‘users’, while this thesis focuses on ‘buyers’. Therefore, the two terms, user and buyer, are used interchangeably.

Page 33: Understanding and Measuring Light Buyer Brand Equity

24

Results from previous studies (i.e., Romaniuk & Wight 2009; Sharp, Beal & Romaniuk 2001; Wight 2010) suggest analysing user and non-user awareness separately, where “practitioners will be able to more accurately identify the success of marketing efforts and evaluate the effectiveness of targeted activities” (Wight 2010, p.13). The purpose of a brand’s strategy, whether acquisition and/or retention, may alter the awareness of one group without affecting the other. Therefore marketers are advised to track user and non-user awareness scores separately, looking to measures that are sensitive to changes for each buyer group of interest.

No studies to date have examined brand awareness from a light buyer perspective. This research aims to understand light buyer awareness structures and determine the awareness measure that should be utilised to assess light buyer response. Results from this thesis will aid marketers by providing guidance on analysis approaches and interpretation of results for the three measures, building on and extending prior recommendation for users and non-users (i.e., those detailed by Wight, 2010). This research will examine top of mind, unprompted and prompted awareness measures for light brand buyers.

4.4 Light Brand Buyer Awareness

As brands are encountered via purchase/consumption they are encoded and stored as

category knowledge in memory networks, i.e. the brand node is connected to the category node (Barsalou 2003). Awareness measures aim to uncover these brand name linkages connected with the category. Light buyers may have the brand name stored in memory from prior purchase, however the linkage to the category cue is weak from infrequent exposure to the brand. Brand to category linkages are stronger for brands purchased more frequently, which are therefore retrieved more readily from memory, as per frequency and recency effects (Anderson 1983). As light buyers can be heavier buyers of other brands in the category (Ehrenberg 2000a), competitor nodes may interfere with the accessibility of light buyer brand information during retrieval (Alba & Chattopadhyay 1986). To account for competitor interference during retrieval from memory, a better awareness measure will therefore capture a higher proportion of light buyer response. Results will indicate the number of light buyers where the brand node is linked to the category node in memory structures.

Page 34: Understanding and Measuring Light Buyer Brand Equity

25

The three brand awareness measures tested in this research differ in terms of the cue provided (category/brand), retrieval of brand information from memory (recall/recognition) and response permitted (first brand/all brands). It is important to first understand how light brand buyers respond to each awareness measure, before comparing results across measures.

The following sections will draw from memory theory literature. Light brand buyer response is first discussed in relation to non-buyers and heavier buyers. Assessing different buyer and non-buyer groups will provide context for light buyer results, and highlight best practice for future analysis, i.e. whether to analyse different brand buyer groups separately.

4.4.1 Top of Mind Awareness Though this research is focused on examining light brand buyer response specifically, to provide context for light buyer awareness, analysis of other buyer and non-buyer groups is needed. Furthermore, to gauge a complete picture of the performance and

potential of the brand, marketers require an understanding of where awareness measures will better capture light buyer response and response for the brand’s entire customer base. This includes the brand’s potential customers, non-brand buyers, as growth in the light buyer segment can be attributed to non-buyers purchasing the brand once, entering the light buyer segment (Anschuetz 2002; Goodhardt & Ehrenberg 1967). Light brand buyer awareness is analysed in contrast to non-buyer and those who purchase more frequently, i.e. heavier buyers.

TOM awareness is measured as the fist brand recalled when given the category as a cue. Brands that have strong links in memory to the activated node (category cue) will be recalled first, with respondent ability to recall decreasing with each attempt (Anderson 1974; Heil, Rösler & Hennighausen 1994). Respondents are more likely to recall a brand that has been purchased, followed by brands that are familiar due to advertising and in-store exposure (Alba & Chattopadhyay 1986; Hutchinson 1983; Hutchinson, Raman & Mantrala 1994; Nedungadi & Hutchinson 1985). Therefore, brands purchased frequently will be recalled in TOM awareness over brands purchased infrequently or not at all in the timeframe.

Page 35: Understanding and Measuring Light Buyer Brand Equity

26

Drawing from discussion of memory theory, for those who do not purchase a brand in the timeframe (non-buyers), links between the category and brand are only formed via marketing efforts (i.e. advertising, store displays). Light buyers who, though infrequently, purchase the brand have stronger links between the category and brand than those with no purchase experience with the brand (non-buyers). As frequency of repeat purchase increases, buyers encounter the brand more regularly, which helps form stronger links to the category cue in memory, aiding recall.

Therefore, it is expected that response to TOM awareness will increase from non-buyer to light buyer to heavier buyer, in line with respondent level of prior experience with the brand. This leads to the first hypothesis (represented in Figure 7):

H1a: Light buyers will have a higher response to TOM awareness than non-

buyers.

H1b: Heavier buyers will have a higher response to TOM awareness than

light buyers.

Figure 7: Buyer and Non-buyer TOM awareness hypothesis development.

4.4.2 Unprompted Awareness Unprompted awareness is measured as all brands recalled when given the category as a cue (Bird & Ehrenberg 1966). Buyers are therefore not only able to recall brands they have purchased frequently, but also brands they know or have purchased infrequently (Wight 2010).

Page 36: Understanding and Measuring Light Buyer Brand Equity

27

Like TOM awareness, unprompted awareness is a recall question. Recall is dependent on the strength of the link between the category cue and brand name in memory, with those who purchase the brand more frequently more likely to retrieve the brand over those who only encounter the brand via marketing efforts (Hutchinson 1983). As the category cue is activated in memory, this activation spreads along linkages, with brands recalled in line with their strength to the category cue. As the activation spreads, recall weakens (Anderson 1974; Heil, Rösler & Hennighausen 1994).

Thus, unprompted response will follow TOM awareness results (as show in Figure 7), with response highest for brands purchased frequently (heavier buyers), followed by brands purchased infrequently (light buyers) and finally those exposed to the brand only via marketing efforts (non-buyers). The second hypothesis to be tested is:

H2a: Light buyers will have a higher response to unprompted awareness

than non-buyers.

H2b: Heavier buyers will have a higher response to unprompted awareness

than light buyers.

4.4.3 Prompted Awareness Prompted awareness is measured as the respondent’s ability to confirm prior exposure

when given the brand as a cue (Keller 1993; Laurent, Kapferer & Roussel 1995). As a recognition measure, prompted awareness provides marketers with an indication of brands available in memory structures (Lynch & Srull 1982).

In prompted awareness, brands are more accessible from memory, as opposed to recall measures where retrieval depends on the activation of a brand name node from the category cue (Laurent, Kapferer & Roussel 1995; Lynch & Srull 1982). Both non-buyer and buyers of the brand have a higher probability of recognising the brand with prompted awareness (Wight 2010).

However, Romaniuk and Wight (2009) find that brand users are 1.7 times more likely to recognise the brand than non-brand users. Indeed, those with little exposure to the brand, via marketing efforts and word of mouth, may have seen the brand previously, yet not processed the brand name enough to form a link in memory. This leads to the first prompted awareness hypothesis:

H3a: Light buyers will have a higher response to prompted awareness than

non-buyers.

Page 37: Understanding and Measuring Light Buyer Brand Equity

28

When buyers make a purchase, brand information is encoded in memory, with subsequent purchases strengthening the link between the brand and information (Van Osselaer & Janiszewski 2001). Retrieval of brand information from memory requires the respondent to indicate whether the node is present in memory structures, rather than activate and search between nodes. Less cognitive effort is required than recall measures (Lynch & Srull 1982), allowing information for brands purchased frequently and infrequently to be retrieved. This indicates that both light and heavier buyers will be able to recognise the brand, leading to the next hypothesis. Figure 8 provides an example of expectations for buyer prompted awareness response levels.

H3b: Prompted awareness response will not differ for light and heavier

buyers.

Figure 8: Buyer and Non-buyer Prompted awareness hypothesis development.

4.4.4 Light Buyer Response across Awareness Measures The three awareness measures, TOM, unprompted and prompted awareness have been discussed for light brand buyers in comparison with non-buyers and heavier buyers.

Previous empirical studies compare results for TOM, unprompted and prompted awareness. In Laurent, Kapferer and Roussel’s (1995) study, a relationship is found between the three measures, whereby the results from one measure can be used to predict the results for another measure. Romaniuk et al. (2004) replicate and extend this work, with analysis conducted at the brand level, comparing results over time, rather than across brands and measures at a single point in time. However, there are discrepancies between the two studies. Romaniuk et al. (2004) determine that the relationship stated previously may be due to variance between brands, rather than a close relationship between measures. Regardless, results for the two studies show a similar relationship between awareness measures. Romaniuk et al.’s (2004) findings support Laurent, Kapferer and Roussel’s (1995) research that awareness measures

Page 38: Understanding and Measuring Light Buyer Brand Equity

29

differ in terms of difficulty of retrieval from memory in the same structural way across categories. Both studies show that TOM awareness is more difficult than unprompted awareness, and in turn unprompted awareness is more difficult than prompted awareness. Similarly, Romaniuk and Wight (2009) and Wight (2010) note that the difference between user and non-user response decreases from TOM to prompted awareness, where both users and non-users are able to recognise the brand.

Therefore, as the difficulty of the awareness measure decreases, light buyer ability to recall and/or recognise the brand will increase. TOM awareness is measured as a subset of unprompted awareness. Light buyers who recall the brand TOM will be included in unprompted awareness results, increasing the proportion of responses captured. For light buyers who respond to recall measures, the brand name becomes salient (has a higher propensity to come to mind) (Alba & Chattopadhyay 1986), thus when given the brand name it is expected that most light buyers will be able to recognise the brand. Prompted awareness is therefore a reflection of light buyers who are able to recall and recognise the brand. This leads to the final awareness hypotheses (depicted in Figure 9).

H4a: Light buyers will have a higher response for unprompted awareness

than for TOM awareness.

H4b: Light buyers will have a higher response for prompted awareness than

for unprompted awareness.

Figure 9: Light buyer awareness hypothesis development.

Page 39: Understanding and Measuring Light Buyer Brand Equity

30

4.5 Chapter Summary

This chapter has introduced the first brand equity measure of interest, brand

awareness. Key take-outs include:

• Brand awareness is the building block for other brand equity measures.

• For marketers, brand awareness is important to be considered for brand purchase. For consumers, brand awareness simplifies the decision making process.

• Brand awareness measures can be classified as recall or recognition measures. This thesis analyses three brand awareness measures: top of mind (recall), unprompted (recall) and prompted (recognition) awareness (refer Table 1).

Table 1: Summary of three brand awareness measures. Awareness measure Classification Activation cue Memory

retrieval

Top of mind First brand mentioned from a category cue. Category name Recall

Unprompted All brands mentioned from a category cue. Category name Recall

Prompted Those selected from a list of brands provided.

Brand name/ visual Recognition

The following chapter details the research approach taken to test brand awareness hypotheses.

Page 40: Understanding and Measuring Light Buyer Brand Equity

31

Chapter 5 AWARENESS RESEARCH APPROACH Chapter Five details the research approach for brand awareness, the first of two brand equity measures analysed in this thesis. The chapter provides a detailed description of

the data sets used for analysis, buyer classification, operationalisation of brand awareness measures and techniques for data analysis.

5.1 Secondary Data

Secondary data was used for brand awareness analysis, provided by the Ehrenberg-

Bass Institute for Marketing Science. Using secondary data removes limits of scope and time constraints, permitting more extensive analysis across multiple countries and categories. This is important to develop empirically generalisable results (Ehrenberg 1995).

However, the researcher has no control over the survey design and the variables available within each data set. Limitations associated with secondary data include missing data, data not collected in the initial studies or issues of booking, where variables are coded in a way that is not transparent to the current researcher (Neuman 2011). As Ehrenberg-Bass researchers were involved in data collection for the categories tested, any discrepancies or issues of booking were resolved with discussion.

5.2 Product Categories

Consumer packaged goods categories were chosen for testing. Studies that quantify

the relationship between awareness and usage have largely tested from an advertising awareness perspective (e.g., Romaniuk & Wight 2009; Sharp, Beal & Romaniuk 2001). This research extends Wight’s (2010) research, which analyses brand awareness for user and non-user for three consumer packaged goods categories.

In addition, the categories selected for analysis in this research are repertoire markets, where buyers purchase from multiple brands (Ehrenberg 2000a). A higher purchase frequency allows greater ability to distinguish between light and heavier buyers of different brands during analysis. Furthermore, the prevalence of data available that were collected as part of brand equity surveys suggest that results for this type of category will be of immediate use to marketers.

Page 41: Understanding and Measuring Light Buyer Brand Equity

32

5.3 Data Sets

This thesis used a multiple sets of data (MSoD) approach. That is, testing for patterns

across different categories, brands and countries for brand awareness. A multiple sets of data approach was used to produce results that are generalisable, lawlike and predictable, which can be replicated and extended in future studies (Ehrenberg 1990; Uncles & Wright 2004). Table 2 provides an overview of the data sets analysed for brand awareness.

Table 2: Description of multiple sets of data for Brand Awareness.

# Category Country Year Total Sample

Brands Analysed

Panel / Claimed Usage

1 Tea UK N/A 7005 9 Panel 2 Pasta Sauce Australia 2012 2095 13 Panel 3 Soft Drink UK 2012 974 14 Claimed 4 Whiskey US 2013 785 16 Claimed

A key strength of brand awareness analysis was the scope of research, where analysis was conducted for over 10,000 respondents, four categories, three countries, and 52 brands. A second strength was analysing matched panel and brand equity survey data, where purchase data and response to awareness questions were available for the same individual.

Panel data is collected via a representative sample of households who regularly provide their purchase information in a longitudinal study, providing a means to classify respondents into different buyer categories. This study was cross-sectional, analysing buyer purchase frequency for a single point in time. Though responses captured in panel data may suffer from attrition bias (participants ‘dropping out’ from the study), data is comprised of observed behaviour rather than claimed behaviour, providing more accurate and reliable results (Ehrenberg 2000a; Sudman & Wansink 2002).

For soft drink and whiskey categories, response to awareness questions and claimed usage, which is typically collected and analysed in brand equity surveys, was analysed. Claimed usage refers to asking respondents to recall their number of purchases made within a given period. I.e., How many times have you bought each of these brands in the past 12 months? Respondents are presented with a range of numbers to select from for

each brand, i.e. from 1 to 10+. There are limitations with using claimed usage, including memory biases, such as: encoding and retrieval failure, memory decay, or telescoping, where respondents have difficulty keeping to the time period asked (Nenycz-Thiel et al. 2012). Results influenced by response bias can lead to overestimation or

Page 42: Understanding and Measuring Light Buyer Brand Equity

33

underestimation in purchase frequency, which has been found to be particularly prevalent for light buyers (Ludwichowska 2013).

Including the two types of purchase data in the research design allowed comparison between data that is more accurate, panel data, and data that is traditionally used when analysing brand equity surveys, claimed usage. Therefore, results reflect the ‘true’ light buyer awareness response, along with results practitioners can expect in marketing practice.

5.4 Operationalisation of Brand Awareness

Brand awareness was measured according to three variables:

Top of mind (TOM) – first brand recalled when provided with the category cue.

I.e. Please list all the brands of <product category> that you can think of.

Unprompted – all brands recalled when provided with the category cue.

Prompted – brands selected when provided with the brand name. I.e. Looking

at the following brands, can you indicate which ones you have seen before today? Respondents were instructed to select all brands that apply.

5.5 Selecting an Appropriate Timeframe for Buyer Classification

A period of 12-months was used for analysis in order to sufficiently capture the brand’s

light buyers (as outlined in Schmittlein, Cooper & Morrison 1993; Sharp & Romaniuk 2007). For categories where buyer purchases were not captured for a 12 month period, a limitation of using secondary data, the largest period was used for analysis. For soft drink, a one month period sufficed, along with a three month period for whiskey.

A longer timeframe allowed greater ability to capture a representative sample of all types of buyers, from non-buyers to heavier buyers. The length of the timeframe for analysis affects which classification someone falls into. A longer timeframe allows for more purchase occasions, which means greater opportunity to distinguish between heavy and light buyers (Ehrenberg 2000a). As the timeframe increases, buyers make further purchases becoming heavier buyers and non-buyers purchase once entering the light buyer segment. It is important to note that while increasing the timeframe increases

the number of purchases from light buyers, it also increases the number of purchases from heavier buyers. Thus, light buyers’ purchase frequency still looks relatively light in comparison to heavy buyer purchase frequency.

Page 43: Understanding and Measuring Light Buyer Brand Equity

34

Romaniuk and Wight (2014) provide a useful example (see Table 3) that shows how buyers can be classified differently depending on the timeframe chosen for analysis. In their example, a heavy buyer is defined as buying five or more times. In option one, the shaded timeframe (Jan-Jun) classifies Joe as a heavy buyer, whereas in option two (May-Oct) Mary is considered the heavy buyer, not Joe, despite both purchasing ten times in the period. This highlights the natural variation in purchase timing and the need to examine periods that exceed average inter-purchase times (Ehrenberg, Uncles & Goodhardt 2004; Goodhardt, Ehrenberg & Chatfield 1984).

Table 3: Example of buying distributions and buyer classification. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Total Option 1 Joe 0 2 2 2 0 0 0 0 1 1 1 1 10 Mary 1 0 1 1 0 1 1 2 1 1 0 1 10 Option 2 Joe 0 2 2 2 0 0 0 0 1 1 1 1 10 Mary 1 0 1 1 0 1 1 2 1 1 0 1 10 Source: Romaniuk and Wight (2014, p.7)

5.6 Operationalisation of Light Brand Buyers

In line with consumer packaged goods studies (e.g., Anschuetz 2002; Ehrenberg 2000a;

Ehrenberg, Uncles & Goodhardt 2004), light brand buyers were classified as having purchased the brand once. Classification of buyer and non-buyers follows Anschuetz’s (2002) study, who categorised buyers and non-buyers for a 12-month period. Light brand buyers were compared to non-buyers (zero purchases) and other, heavier, buyers (two or more purchases).

For soft drink, which is a frequently purchased category, light buyer classification was adjusted to account for the small sample of ‘once only’ buyers. Light brand buyer classification therefore included those who purchased the brand once or twice, with heavier buyers classified as those who purchased the brand three or more times in a one month period.

Page 44: Understanding and Measuring Light Buyer Brand Equity

35

5.7 Data Analysis

To avoid sampling error, only brands with an adequate sample within each buyer and

non-buyer group were included in analysis. Here, brands with sample below 30 were excluded from analysis. Given brands have many buyers who purchase the brand once or twice and few buyers who purchase the brand frequently (Ehrenberg 1959, 1988), brand exclusion occurred primarily for having few heavier buyers. This sample cut off was used for all analyses.

To understand the findings easily, data was reduced using elements drawn from Ehrenberg (2000b), including: rounding to two significant figures, using row and column averages, ordering data by size, describing the main pattern or findings in a summary sentence, and effective use of layout and spacing. Data reduction allows researchers to see patterns in the data and minimise unnecessary complexities to assist in interpretation.

5.7.1 Buyer and Non-Buyer Response Analysis tested awareness response for over 10,000 respondents, across four product

categories, three countries and for 52 brands. For each product category, respondents were categorised as non-buyer, light, and heavier buyer using brand purchase frequency. Cross-tabulation descriptive statistics were used to determine the proportion of each buyer group who responded to the brand awareness measure of interest. A cross-tabulation creates a row for each value of one variable and a column for each value in a second variable. The intersecting cell indicates the number of respondents that fill both criteria. Table 4 provides an example of a cross-tabulation output for TOM awareness using the statistical program, SPSS. Analysis shows that 19% of non-buyers recall Brand A6 pasta sauce in TOM awareness.

Table 4: Example TOM awareness and buyer type cross-tabulation for Brand A pasta sauce.

%

Brand A Purchase Frequency

Total 0 (Non-buyer)

1 (Light buyer)

2+ (Heavier buyer)

Brand A TOM

Not Aware 81 75 65 74 Aware 19 25 35 26 Total 100 100 100 100

6 Brands are masked for confidentiality reasons.

Page 45: Understanding and Measuring Light Buyer Brand Equity

36

At the brand level, inferential statistics of independent samples t-tests were used to test

for statistical significance at the 0.05 level. This gives the researcher 95% confidence that the results were large enough to indicate a relationship truly exists between two variables and was not due to chance alone (Neuman 2011). In order to merge results and come to a conclusive finding across the four product categories, further statistical testing was employed. Therefore, binary logistic regression was used to determine if awareness results could be statistically differentiated for two groups of buyers, i.e. light and non/heavier.

5.8 Chapter Summary

This chapter has discussed the method used for Part One: Brand Awareness, including:

• Information on the data sets analysed.

• Justification for buyer classification and the analysis timeframe selected.

• Operationalisation of brand awareness measures.

• Analysis techniques, including the presentation of results.

The following chapter will present the results for awareness and provide discussion of key findings.

Page 46: Understanding and Measuring Light Buyer Brand Equity

37

Chapter 6 AWARENESS RESULTS & DISCUSSION Results from awareness analysis are presented, answering the hypotheses posed by this study. Discussion of the findings allows recommendations to be made for marketer and

academic use.

6.1 Awareness Results

Results are presented according to the order of hypotheses in Chapter Four. Results

are tested for statistical significance using independent samples t-tests at the 0.05 level. Light buyer response is presented in comparison to non-buyer and heavier buyer response for TOM, unprompted and prompted awareness. Analysis of light brand buyer response across the three awareness measures follows.

6.1.1 TOM Awareness Results H1a: Light buyers will have a higher response to TOM awareness than non-

buyers.

To address the first hypothesis, which states that light buyers will have a higher response to TOM awareness than non-buyers, cross-tabulations are calculated of buyer purchase frequency and the first brand recalled when provided with the category cue. Results are presented at the brand level for the four product categories: tea, pasta sauce, soft drink and whiskey (see Tables 5 to 8). Brands are ordered by penetration.

Observation of the data shows two patterns. TOM awareness response increases from non-buyers to light buyers. For tea (refer Table 5), 7 out of 9 brands show statistically significant differences, where light buyer TOM awareness response is higher than non-buyer response. Secondly, TOM awareness response decreases within each buyer group in line with brand penetration. Exceptions to this brand size effect are noted for each product category. Further discussion of brand size remains outside the scope of research.

In addition, it is interesting to note lower awareness response for private label brands (i.e. Brand B and Brand H). Private label brands are distributed less than national brands and rarely advertised, which limit non-buyer and light buyer exposure to the brand.

Page 47: Understanding and Measuring Light Buyer Brand Equity

38

Table 5: Proportion of non-buyer & light buyer TOM awareness response - Tea.

Tea (n=7005) % Pen7

% Non – Buyer

% Light Buyer

Brand A 44 14 27* Brand B 42 2 6* Brand C 41 20 39* Brand D 28 1 6* Brand E 18 8 15* Brand F 9 0 3* Brand G 9 1 12* Brand H 8 0 1 Brand I 2 0 3 Average 5 12

*Statistically significantly higher than non-buyer at p<0.05.

For pasta sauce, private label brands (i.e. Brand E, Brand F, Brand G and Brand H) also

exhibit lower awareness response than expected given brand penetration (see Table 6).

Table 6: Proportion of non-buyer & light buyer TOM awareness response - Pasta Sauce.

Pasta Sauce (n=2095)

% Pen

% Non – Buyer

% Light Buyer

Brand A 59 19 25* Brand B 54 24 37* Brand C 44 3 7* Brand D 30 2 5* Brand E 17 0 1 Brand F 12 1 3 Brand G 10 0 3 Brand H 9 1 4 Brand I 9 1 2 Brand J 7 0 1 Brand K 5 0 0 Brand L 4 1 9 Brand M 4 0 6 Average 4 8

*Statistically significantly higher than non-buyer at p<0.05.

There are four pasta sauce brands that show statistical differences, where light buyer TOM awareness response is significantly higher than non-buyer response. For the remaining brands, TOM awareness response for both non and light buyers is below 5%. There are exceptions for two brands, Brand L and Brand M, which though not

statistically significantly higher than non-buyer, have a slightly higher light buyer response than non-buyer response given brand penetration. 7 Brand penetration: % of sample that have purchased the brand at least once in the timeframe.

Page 48: Understanding and Measuring Light Buyer Brand Equity

39

Despite few brands showing statistically significant differences, a trend appears, where light buyer TOM awareness response is higher than non-buyer response for all brands. This trend is present in results for other categories, such as soft drink (see Table 7).

Table 7: Proportion of non-buyer & light buyer TOM awareness response – Soft Drink.

Soft Drink (n=974) % Pen

% Non – Buyer

% Light Buyer

Brand A 39 48 61* Brand B 28 0 8* Brand C 18 0 3 Brand D 18 2 13* Brand E 16 8 27* Brand F 14 1 12* Brand G 14 0 0 Brand H 13 0 0 Brand I 11 0 0 Brand J 11 1 3 Brand K 11 1 11* Brand L 11 0 2 Brand M 9 1 5 Brand N 7 0 13 Average 4 11

*Statistically significantly higher than non-buyer at p<0.05.

Again, despite light buyer TOM awareness response being higher than non-buyer

response across brands, few brands show statistically significant difference in results. Six out of 14 brands have light buyer TOM awareness response statistically significantly higher than non-buyer response. Given that TOM awareness permits one brand to be recalled, the high percentage of both non and light buyers who respond with the largest brand in the category, Brand A, accounts for the low response for other brands in the category.

Similarly, results for the final product category, whiskey (refer Table 8), show low TOM awareness response for both buyer groups. TOM awareness response is highest for Brand G, with light buyer response statistically significantly higher than non-buyer

response. However, unlike Brand A (soft drink), light buyer response for Brand G (whiskey) reaches only 10%. Eleven out of 16 brands are not recalled by non or light buyers, which can be explained by a low level of advertising for whiskey brands in the US. The remaining brands show results trending in favour of light buyer response.

Page 49: Understanding and Measuring Light Buyer Brand Equity

40

Table 8: Proportion of non-buyer & light buyers TOM awareness response – Whiskey.

Whiskey (n=785) % Pen

% Non – Buyer

% Light Buyer

Brand A8 40 2 4 Brand B 35 0 1 Brand C 35 0 0 Brand D 33 0 0 Brand E 31 2 6 Brand F 27 1 4 Brand G 26 1 10* Brand H 25 2 6 Brand I 23 0 0 Brand J 18 0 0 Brand K 18 0 0 Brand L 15 0 4 Brand M 14 0 0 Brand N 14 0 0 Brand O 14 0 0 Brand P 12 0 3 Average 1 2

*Statistically significantly higher than non-buyer at p<0.05.

Across the four product categories, there are statistically significant differences for 18 out of 52 brands (35%), where light buyer TOM awareness response is higher than non-buyer response. Twelve brands have no response for non or light buyers, with results for the remaining 22 brands trending in favour of light buyer response.

Further testing of binary logistic regression is used to determine if TOM awareness response can be significantly discriminated between non and light buyers. A dependent variable is created, with non-buyers coded as 0 and light buyers coded as 1. The independent variable is TOM awareness response. Results find response for non and light buyers statistically significantly different at p<0.05 (Nagelkerke R2=8%). The test has a Wald statistic of 4.3 and Exp(B) of 1.06. At the end of this section, regression results are presented and compared to light versus heavier buyer test statistic.

Overall, the weight of evidence supports hypothesis 1a that light buyer TOM awareness response is higher than non-buyer response.

8 Brand A is the Nett figure for a whiskey brand.

Page 50: Understanding and Measuring Light Buyer Brand Equity

41

H1b: Heavier buyers will have a higher response to TOM awareness than

light buyers.

Following from non and light buyer response to TOM awareness, light buyer response is compared to heavier buyer ability to recall the brand. Here, it is expected that light buyer response will be lower than heavier buyer TOM awareness response. Cross-tabulations are again used to test the hypothesis, with independent samples t-tests and binary logistic regression used to assess statistical significance. Tables 9 to 12 present results for the four product categories.

For tea (see Table 9), there are statistically significant differences in TOM awareness response for all brands, where heavier buyer response is higher than light buyer response at p<0.05.

Table 9: Proportion of light buyer & heavier buyer TOM awareness response - Tea.

Tea (n=7005) % Pen

% Light Buyer

% Heavier Buyer

Brand A 44 27 47* Brand B 42 6 26* Brand C 41 39 62* Brand D 28 6 18* Brand E 18 15 29* Brand F 9 3 9* Brand G 9 12 44* Brand H 8 1 10* Brand I 2 3 15* Average 12 29

*Statistically significantly higher than light buyer at p<0.05.

Similarly, for pasta sauce (see Table 10), with the exception of three brands, heavier buyer TOM awareness response is statistically significantly higher than light buyer response. Brands that do not show a significant difference in response are private label brands, i.e. Brand H, or brands with lower penetration, i.e. Brand K and Brand L. However, results for the three brands still show heavier buyer response higher than light buyer response.

Page 51: Understanding and Measuring Light Buyer Brand Equity

42

Table 10: Proportion of light buyer & heavier buyer TOM awareness response - Pasta Sauce.

Pasta Sauce (n=2095)

% Pen

% Light Buyer

% Heavier Buyer

Brand A 59 25 35* Brand B 54 37 49* Brand C 44 7 21* Brand D 30 5 17* Brand E 17 1 7* Brand F 12 3 9* Brand G 10 3 5 Brand H 9 4 15* Brand I 9 2 27* Brand J 7 1 0 Brand K 5 0 9 Brand L 4 9 49* Brand M 4 6 25* Average 8 21

*Statistically significantly higher than light buyer at p<0.05.

For soft drink, there are two brands, Brand A and Brand C, with statistically significant

differences in light and heavier buyer response, while no whiskey brands show statistically significantly different results for light and heavier buyers (refer Table 11 and 12). Of the remaining 30 soft drink and whiskey brands, results for 16 brands trend in favour of heavier buyer response, 5 trend in favour of light buyer response, 2 brands show no difference in non and light buyer response, while there is no response for non or light buyers for 5 brands.

Page 52: Understanding and Measuring Light Buyer Brand Equity

43

Table 11: Proportion of light buyer & heavier buyer TOM awareness response – Soft Drink.

Soft Drink (n=974) % Pen

% Light Buyer

% Heavier Buyer

Brand A 39 61 72* Brand B 28 8 9 Brand C 18 3 14* Brand D 18 13 17 Brand E 16 27 22 Brand F 14 12 7 Brand G 14 0 1 Brand H 13 0 0 Brand I 11 0 0 Brand J 11 3 2 Brand K 11 11 12 Brand L 11 2 11 Brand M 9 5 16 Brand N 7 13 9 Average 11 14

*Statistically significantly higher than light buyer at p<0.05.

Table 12: Proportion of light buyer & heavier buyer TOM awareness response – Whiskey.

Whiskey (n=785) % Pen

% Light Buyer

% Heavier Buyer

Brand A 40 4 10 Brand B 35 1 1 Brand C 35 0 2 Brand D 33 0 2 Brand E 31 6 6 Brand F 27 4 12 Brand G 26 10 7 Brand H 25 6 11 Brand I 23 0 1 Brand J 18 0 0 Brand K 18 0 0 Brand L 15 4 7 Brand M 14 0 0 Brand N 14 0 7 Brand O 14 0 7 Brand P 12 3 5 Average 2 5

Page 53: Understanding and Measuring Light Buyer Brand Equity

44

Overall, for 19 out of 52 brands (37%), heavier buyer TOM awareness response is

statistically significantly higher than light buyer response. A further 20 brands trend in favour of heavier buyer response. This provides a total of 75% of brands with either statistically significant or trending results in favour of heavier buyer TOM awareness response.

Binary logistic regression testing similarly finds a significant difference between light and heavier buyer TOM awareness response at p<0.05 (Nagelkerke R2=9%). A dependent variable is used, where light buyers are coded as 0 and heavier buyers coded as 1. The independent variable is again TOM awareness response. Results show a Wald statistic of 5.7 and Exp(B) of 1.04.

Therefore, hypothesis 1b is supported that heavier buyer response to TOM awareness is higher than light buyer response.

Do light buyers look more like non-buyers or heavier buyers?

Results indicate that as purchase frequency increases, response to TOM awareness increases. Thus, response increases from non to light to heavier buyers (see Table 13 for a summary of results across product categories). These findings are notable, however they do not provide information about whether light buyer response to TOM awareness reflects response levels of non-buyers, heavier buyers, or is different to both buyer groups.

Table 13: Summary TOM awareness results.

TOM Awareness % Non-Buyer

% Light Buyer

% Heavier Buyer

Tea (n=7005) 5 12 29 Pasta Sauce (n=2095) 4 8 21 Soft Drink (n=974) 4 11 14 Whiskey (n=785) 1 2 5 Average 4 9 17

Binary logistic regression tests previously used to assess the difference between light

and non/heavier buyer response are compared. Summary of logistic regression results is presented in Table 14.

Page 54: Understanding and Measuring Light Buyer Brand Equity

45

Table 14: Logistic regression results for non, light and heavier buyer TOM awareness.

TOM Nagelkerke R2 Wald Sig. Exp(B)

Non to light 0.08 4.3 <0.05 1.06 Light to heavier 0.09 5.7 <0.05 1.04

Light buyer TOM awareness response is statistically significantly different from both non-buyer and heavier buyer response at p<0.05, where the distance between light buyer response is slightly larger for non-buyers than for heavier buyers.

6.1.2 Unprompted Awareness Results As brands are included in analysis for TOM awareness on the basis of buyer sample,

the same brands are analysed for unprompted awareness.

H2a: Light buyers will have a higher response to unprompted awareness

than non-buyers.

The first unprompted awareness hypothesis that light buyer response will be higher than non-buyer response is tested using cross-tabulations of buyer purchase frequency and unprompted awareness response. Tables 15 to 18 present unprompted awareness results for the four product categories.

For all tea brands (refer Table 15), light buyer unprompted awareness response is statistically significantly higher than non-buyer response. As with TOM awareness, private label brands (Brand B and Brand H) show lower awareness given brand penetration for both buyer groups.

Table 15: Proportion of non-buyer & light buyer unprompted awareness response - Tea.

Tea (n=7005) % Pen

% Non – Buyer

% Light Buyer

Brand A 44 50 63* Brand B 42 13 25* Brand C 41 61 76* Brand D 28 14 27* Brand E 18 37 49* Brand F 9 2 8* Brand G 9 25 54* Brand H 8 1 5* Brand I 2 1 21* Average 23 36

*Statistically significantly higher than non-buyer at p<0.05.

Page 55: Understanding and Measuring Light Buyer Brand Equity

46

Pasta sauce results show a similar pattern with those for tea (see Table 16), with light

buyer unprompted awareness response statistically significantly higher than non-buyer response for 11 out of 13 brands. Response for the two brands, Brand J and Brand K, is again low. One explanation could be that the brands are not advertised often, which makes recall more difficult for non and light buyers.

Table 16: Proportion of non-buyer & light buyer unprompted awareness response - Pasta Sauce.

Pasta Sauce (n=2095)

% Pen

% Non – Buyer

% Light Buyer

Brand A 59 40 56* Brand B 54 50 62* Brand C 44 13 26* Brand D 30 14 33* Brand E 17 1 6* Brand F 12 6 13* Brand G 10 5 15* Brand H 9 7 29* Brand I 9 3 17* Brand J 7 0 2 Brand K 5 0 0 Brand L 4 14 48* Brand M 4 1 19* Average 12 25

*Statistically significantly higher than non-buyer at p<0.05.

For soft drink, there are statistically significant differences for 11 out of 14 brands (refer Table 17), where light buyer response is higher than non-buyer response. Brands that do not show statistical differences, Brand C, Brand G and Brand I, are brand variants.

The lower unprompted awareness response may be due to respondents recalling the brand name, moving on to recall other brand names rather than variants of the brand. Despite lower awareness scores for the three brands, a trend is present, where light buyer response is higher than non-buyer unprompted awareness response.

Finally, for whiskey (see Table 18), of the eight brands that have unprompted awareness response over 10%, six brands show light buyer unprompted awareness response statistically significantly higher than non-buyer response. For the remaining brands, results trend in favour of light buyer unprompted response.

Page 56: Understanding and Measuring Light Buyer Brand Equity

47

Table 17: Proportion of non-buyer & light buyer unprompted awareness response – Soft Drink.

Soft Drink (n=974) % Pen

% Non – Buyer

% Light Buyer

Brand A 39 82 91* Brand B 28 3 13* Brand C 18 2 4 Brand D 18 45 58* Brand E 16 61 76* Brand F 14 29 54* Brand G 14 1 2 Brand H 13 25 52* Brand I 11 2 5 Brand J 11 23 37* Brand K 11 10 45* Brand L 11 20 40* Brand M 9 13 39* Brand N 7 2 34* Average 23 39

*Statistically significantly higher than non-buyer at p<0.05.

Table 18: Proportion of non-buyer & light buyer unprompted awareness response – Whiskey.

Whiskey (n=785) % Pen

% Non – Buyer

% Light Buyer

Brand A 40 6 13* Brand B 35 1 4 Brand C 35 3 3 Brand D 33 1 7 Brand E 31 8 24* Brand F 27 4 19* Brand G 26 9 28* Brand H 25 9 14 Brand I 23 1 1 Brand J 18 0 0 Brand K 18 0 4 Brand L 15 1 13* Brand M 14 0 0 Brand N 14 3 9 Brand O 14 2 15 Brand P 12 2 20* Average 3 11

*Statistically significantly higher than non-buyer at p<0.05.

Across the four product categories, 39 out of 52 brands (75%) have light buyer unprompted awareness response statistically significantly higher than non-buyer response, with remaining brands trending in favour of light buyers.

Page 57: Understanding and Measuring Light Buyer Brand Equity

48

To summarise, as with TOM awareness, binary logistic regression is used to test whether unprompted awareness response can be statistically significantly differentiated for non and light buyers. The dependent variable remains the same, with non-buyers coded as 0 and light buyers coded as 1. The independent variable is unprompted awareness response. Overall, results are statistically significant at p<0.05 (Nagelkerke R2=11%), with Wald statistic 7.7 and Exp(B) 1.03. Statistical testing supports hypothesis 2a, where light buyer unprompted awareness response is higher than non-buyer response.

H2b: Heavier buyers will have a higher response to unprompted awareness

than light buyers.

Next, hypothesis 2b, which states that heavier buyer response will be higher than light buyer response, is tested. In a similar format to previous hypothesis testing, Tables 19 to 22 present light brand buyer unprompted awareness against heavier buyer awareness response.

For tea and pasta sauce categories (see Tables 19 and 20), heavier buyer unprompted awareness response is statistically significantly higher than light buyer response for 20 out of 22 brands. Though not statistically significantly different, results for two pasta sauce brands, Brand J and Brand K, trend in favour of heavier buyer response.

Table 19: Proportion of light buyer & heavier buyer unprompted awareness response - Tea.

Tea (n=7005) % Pen

% Light Buyer

% Heavier Buyer

Brand A 44 63 76* Brand B 42 25 51* Brand C 41 76 84* Brand D 28 27 49* Brand E 18 49 60* Brand F 9 8 29* Brand G 9 54 79* Brand H 8 5 17* Brand I 2 21 44* Average 36 54

*Statistically significantly higher than light buyer at p<0.05.

Page 58: Understanding and Measuring Light Buyer Brand Equity

49

Table 20: Proportion of light buyer & heavier buyer unprompted awareness response - Pasta Sauce.

Pasta Sauce (n=2095)

% Pen

% Light Buyer

% Heavier Buyer

Brand A 59 56 67* Brand B 54 62 76* Brand C 44 26 54* Brand D 30 33 54* Brand E 17 6 17* Brand F 12 13 25* Brand G 10 15 31* Brand H 9 29 43* Brand I 9 17 48* Brand J 7 2 7 Brand K 5 0 14 Brand L 4 48 77* Brand M 4 19 41* Average 25 43

*Statistically significantly higher than light buyer at p<0.05.

For the remaining product categories, soft drink and whiskey (see Tables 21 and 22), there are few statistically significant differences in results for the two buyer groups.

For soft drink, Brand C heavier buyer unprompted awareness response is statistically significantly higher than light buyer response. Across the remaining 13 brands, eight brands trend in favour of heavier buyer response.

Table 21: Proportion of light buyer & heavier buyer unprompted awareness response – Soft Drink.

Soft Drink (n=974) % Pen

% Light Buyer

% Heavier Buyer

Brand A 39 91 94 Brand B 28 13 16 Brand C 18 4 21* Brand D 18 58 67 Brand E 16 76 69 Brand F 14 54 62 Brand G 14 2 7 Brand H 13 52 51 Brand I 11 5 7 Brand J 11 37 35 Brand K 11 45 44 Brand L 11 40 44 Brand M 9 39 53 Brand N 7 34 12 Average 39 41

*Statistically significantly higher than light buyer at p<0.05.

Page 59: Understanding and Measuring Light Buyer Brand Equity

50

Five soft drink brands trend in favour of light buyer response. Light buyer response is,

on average, seven percentage points higher than heavier buyer response. Removing the private label brand, Brand N, which shows light buyer response 22 percentage points higher than heavier buyer response, the average difference in unprompted response for the two buyer groups decreases to three percentage points. For some brands, light buyers have a higher response than heavier buyers to unprompted awareness. However, overall light buyer response remains similar to heavier buyer response.

Table 22: Proportion of light buyer & heavier buyers unprompted awareness response – Whiskey.

Whiskey (n=785) % Pen

% Light Buyer

% Heavier Buyer

Brand A 40 13 16 Brand B 35 4 2 Brand C 35 3 8* Brand D 33 7 5 Brand E 31 24 23 Brand F 27 19 28* Brand G 26 28 31 Brand H 25 14 20* Brand I 23 1 2 Brand J 18 0 1 Brand K 18 4 1 Brand L 15 13 22* Brand M 14 0 0 Brand N 14 9 21* Brand O 14 15 12 Brand P 12 20 28 Average 11 14

*Statistically significantly higher than light buyer at p<0.05.

For whiskey, as with TOM awareness, unprompted awareness response for non and

light buyers is low (see Table 22). Five brands show heavier buyer unprompted awareness response statistically significantly higher than light buyer response. A further five brands have results that trend in favour of heavier buyers, while five brands trend in favour of light buyers. Brand M has no response for non or light buyers. It is important to note, however, that of the 11 brands that do not show statistically significant differences in response, six brands have unprompted awareness levels below 10%.

Page 60: Understanding and Measuring Light Buyer Brand Equity

51

Across the four product categories, 26 out of 52 brands (50%) show heavier buyer unprompted awareness response statistically significantly higher than light buyer response. Of the remaining 26 brands, 15 brands (56%) trend in favour of heavier buyer response. A total of 41 out of 52 brands (79%) show statistical differences or trend in favour of heavier buyer response.

Further testing using binary logistic regression is employed to determine if unprompted awareness response can be differentiated for light and heavier buyer groups. Results are not statistically significant (P=0.07; Nagelkerke R2=4%). Results show a Wald statistic of 3.4 and Exp(B) of 1.02. Therefore, though heavier buyer response to unprompted awareness is higher, there is overlap between light and heavier buyer response.

Hypothesis 2b is partially supported, where half of the brands tested show heavier buyer response is statistically significantly higher than light buyer response, yet overall, response for the two groups cannot be statistically significantly differentiated.

Do light buyers look more like non-buyers or heavier buyers?

As with TOM awareness, response increases in line with purchase frequency, from non-buyer to heavier buyer (refer Table 23).

Table 23: Summary unprompted awareness results.

Unprompted Awareness

% Non-Buyer

% Light Buyer

% Heavier Buyer

Tea (n=7005) 23 36 54 Pasta Sauce (n=2095) 12 25 43 Soft Drink (n=974) 23 39 41 Whiskey (n=785) 3 11 14 Average 15 28 38

To summarise across product categories, unprompted awareness binary logistic regression results are compared. The tests were performed for two dependent variables: 1) non-buyers coded as 0 and light buyers coded as 1, and 2) light buyers

coded as 0 and heavier buyers coded as 1. The independent variable is buyer unprompted awareness response.

Findings indicate that light buyer unprompted awareness response is statistically significantly different to non-buyer response (see Table 24). Findings are not statistically significant for light and heavier buyers, suggesting that light buyers respond to unprompted awareness in a similar manner to heavier buyers.

Page 61: Understanding and Measuring Light Buyer Brand Equity

52

Table 24: Logistic regression results for non, light and heavier buyer unprompted awareness.

Unprompted Nagelkerke R2 Wald Sig. Exp(B)

Non to light 0.11 7.7 <0.05 1.03 Light to heavier 0.04 3.4 0.07 1.02

6.1.3 Prompted Awareness Results H3a: Light buyers will have a higher response to prompted awareness than

non-buyers.

Hypothesis 3a that light buyers will have a higher response to prompted awareness than non-buyers is tested using cross-tabulations of buyer purchase frequency and response when provided with the brand name. Results for non and light buyers for the four product categories are presented in Tables 25 to 28.

For tea (see Table 25), seven out of nine brands show light buyer prompted awareness response statistically significantly higher than non-buyer response. Brand A and Brand C, though not statistically significantly different, have light buyer prompted awareness response higher than non-buyer response.

Table 25: Proportion of non-buyer & light buyer prompted awareness response - Tea.

Tea (n=7005) % Pen

% Non – Buyer

% Light Buyer

Brand A 44 96 97 Brand B 42 44 59* Brand C 41 97 98 Brand D 28 87 91* Brand E 18 96 98* Brand F 9 20 61* Brand G 9 83 96* Brand H 8 46 68* Brand I 2 14 78* Average 44 65 83

*Statistically significantly higher than non-buyer at p<0.05.

Similarly, 11 out of 13 pasta sauce brands show statistical differences, where light buyer response is higher than non-buyer response (see Table 26). For the two remaining brands, Brand A and Brand B, results trend in favour of light buyer response.

For both tea and pasta sauce categories, brands that do not show statistical differences have higher penetration. Larger brands are better known than smaller brands. Therefore, it is expected that both non-buyers and light buyers will recognise the brand, resulting in a high prompted awareness score.

Page 62: Understanding and Measuring Light Buyer Brand Equity

53

Table 26: Proportion of non-buyer & light buyer prompted awareness response - Pasta Sauce.

Pasta Sauce (n=2095)

% Pen

% Non – Buyer

% Light Buyer

Brand A 59 93 94 Brand B 54 93 96 Brand C 44 76 89* Brand D 30 67 83* Brand E 17 21 42* Brand F 12 37 52* Brand G 10 51 71* Brand H 9 47 67* Brand I 9 41 65* Brand J 7 25 46* Brand K 5 6 28* Brand L 4 79 95* Brand M 4 38 63* Average 52 68

*Statistically significantly higher than non-buyer at p<0.05.

For soft drink, almost all non-buyers and light buyers have a prompted awareness

response above 90% (refer Table 27). There are statistically significant differences, however, where for six out of 14 brands, light buyer prompted awareness response is significantly higher than non-buyer response. For brands that do not show statistical differences, results trend in favour of light buyer response.

Table 27: Proportion of non-buyer & light buyer prompted awareness response – Soft Drink.

Soft Drink (n=974) % Pen

% Non – Buyer

% Light Buyer

Brand A 39 96 97 Brand B 28 91 100* Brand C 18 90 99* Brand D 18 94 99 Brand E 16 95 100* Brand F 14 94 99 Brand G 14 90 97 Brand H 13 94 98 Brand I 11 86 98* Brand J 11 93 97 Brand K 11 89 95 Brand L 11 91 98 Brand M 9 86 97* Brand N 7 56 100* Average 89 98

*Statistically significantly higher than non-buyer at p<0.05.

Page 63: Understanding and Measuring Light Buyer Brand Equity

54

For the final product category, whiskey (see Table 28), all brands show light buyer

prompted awareness response is statistically significantly higher than non-buyer response. All light buyers are able to recognise the brand when prompted, while on average, approximately 60% of non-buyers recognise the brand.

Table 28: Proportion of non-buyer & light buyer prompted awareness response – Whiskey.

Whiskey (n=785) % Pen

% Non – Buyer

% Light Buyer

Brand A 40 94 100* Brand B 35 87 100* Brand C 35 90 100* Brand D 33 78 100* Brand E 31 67 100* Brand F 27 41 100* Brand G 26 62 100* Brand H 25 82 100* Brand I 23 81 100* Brand J 18 50 100* Brand K 18 55 100* Brand L 15 23 100* Brand M 14 28 100* Brand N 14 36 100* Brand O 14 22 100* Brand P 12 16 100* Average 57 100

*Statistically significantly higher than non-buyer at p<0.05.

Across the four product categories, 40 out of 52 brands (77%) have light buyer

prompted awareness response statistically significantly higher than non-buyer response. A trend is present for all other brands, where light buyer response is higher than non-buyer response.

Significance testing across brands is supported by binary logistic regression results, which finds prompted awareness response is significantly different for non and light buyer groups at p<0.001 (Nagelkerke R2=25%), with Wald statistic 16 and Exp(B) 1.04. Hypothesis 3a that light buyer prompted awareness response is higher than non-buyer response is supported.

Page 64: Understanding and Measuring Light Buyer Brand Equity

55

H3b: Prompted awareness responses will not differ for light and heavier

buyers.

To address the last section of hypothesis 3 that heavier buyer response to prompted awareness will not differ to light buyer response, results from cross-tabulations are presented in Tables 29 to 31.

For tea, six out of nine brands show heavier buyer prompted awareness response statistically significantly higher than light buyer response (refer Table 29). There is, at most, a two-percentage point difference in light and heavier buyer response for the remaining brands. Across all brands, heavier buyer prompted awareness response is nine percentage points higher than light buyer response. Removing the two private label brands, Brand B and Brand H, reduces the difference between light and heavier buyer

response to seven percentage points.

Table 29: Proportion of light buyer & heavier buyer prompted awareness response - Tea.

Tea (n=7005) % Pen

% Light Buyer

% Heavier Buyer

Brand A 44 97 98 Brand B 42 59 78* Brand C 41 98 99* Brand D 28 91 96* Brand E 18 98 98 Brand F 9 61 80* Brand G 9 96 98 Brand H 8 68 86* Brand I 2 78 97* Average 83 92

*Statistically significantly higher than light buyer at p<0.05.

For pasta sauce, six out of 13 brands show statistically different results, with heavier buyer response higher than light buyer response (see Table 30). Again, the difference between the two buyer groups is low. Across all brands, heavier buyer response is 10 percentage points higher than light buyer response. Removing private label brands, this decreases to an eight percentage point difference in light and heavier buyer prompted awareness response.

Page 65: Understanding and Measuring Light Buyer Brand Equity

56

Table 30: Proportion of light buyer & heavier buyer prompted awareness response - Pasta Sauce.

Pasta Sauce (n=2095)

% Pen

% Light Buyer

% Heavier Buyer

Brand A 59 94 97* Brand B 54 96 96 Brand C 44 89 93* Brand D 30 83 92* Brand E 17 42 49 Brand F 12 52 65* Brand G 10 71 86* Brand H 9 67 77 Brand I 9 65 84* Brand J 7 46 57 Brand K 5 28 34 Brand L 4 95 97 Brand M 4 63 82 Average 68 78

*Statistically significantly higher than light buyer at p<0.05.

For both soft drink and whiskey categories, there is no statistically significant difference between light and heavier buyer prompted awareness response. Whiskey results are not presented, as prompted awareness response for both light and heavier buyer groups is 100% for all brands.

Table 31: Proportion of light buyer & heavier buyer prompted awareness response – Soft Drink.

Soft Drink (n=974) % Pen

% Light Buyer

% Heavier Buyer

Brand A 39 97 98 Brand B 28 100 98 Brand C 18 99 98 Brand D 18 99 94 Brand E 16 100 99 Brand F 14 99 96 Brand G 14 97 94 Brand H 13 98 97 Brand I 11 98 99 Brand J 11 97 96 Brand K 11 95 96 Brand L 11 98 100 Brand M 9 97 92 Brand N 7 100 94 Average 98 96

Page 66: Understanding and Measuring Light Buyer Brand Equity

57

Two out of four categories tested show statistical differences between light and heavier buyer prompted awareness response. However, for both tea and pasta sauce categories, the difference between the two buyer groups is less than 10 percentage points. Only 12 out of 52 brands (23%) across the four product categories show significant differences between light and heavier buyer prompted awareness. In addition, binary logistic regression results show no statistically significant difference between light and heavier buyer prompted awareness response (P=0.28; Nagelkerke R2=2%). Results show a Wald statistic of 1.2 and Exp(B) of 1.01. Hypothesis 3b that light and heavier buyer response does not differ is supported.

Do light buyers look more like non-buyers or heavier buyers?

Table 32 provides a summary of prompted awareness results across the four product categories.

Table 32: Summary prompted awareness results.

Prompted Awareness

% Non-Buyer

% Light Buyer

% Heavier Buyer

Tea (n=7005) 65 83 92 Pasta Sauce (n=2095) 52 68 78 Soft Drink (n=974) 89 98 96 Whiskey (n=785) 57 100 100 Average 66 87 92

As with unprompted awareness results, binary logistic regression results indicate that

light buyer prompted awareness response is statistically significantly different to non-buyer response (see Table 33). There is no statistical difference between light and heavier buyer groups for prompted awareness, as supported by brand level statistical testing. Light buyer prompted awareness response reflects heavier buyer response.

Table 33: Logistic regression results for non, light and heavier buyer prompted awareness.

Prompted Nagelkerke R2 Wald Sig. Exp(B)

Non to light 0.25 16 <0.001 1.04 Light to heavier 0.02 1.2 0.28 1.01

Page 67: Understanding and Measuring Light Buyer Brand Equity

58

6.1.4 Light Buyer Response across Awareness Measures Light brand buyer results for the three awareness measures have been examined in

comparison to non-buyers and heavier buyers. Here, light buyer response is compared across the three awareness measures.

H4a: Light buyers will have a higher response for unprompted awareness

than for TOM awareness.

H4b: Light buyers will have a higher response for prompted awareness than

for unprompted awareness.

The hypothesis posed by this part of the study suggests that light buyer response will increase from TOM to unprompted to prompted awareness. Tables 34 to 37 summarise light buyer results for the three awareness measures for each product category.

For all tea brands, light buyer prompted awareness response is statistically significantly higher than unprompted awareness response, with unprompted awareness response statistically significantly higher than TOM awareness response (refer Table 34).

Table 34: Proportion of light buyer TOM, unprompted & prompted awareness response –Tea.

Tea (n=7005) % Light Buyer TOM Unprom Promp

Brand A 27 63* 97** Brand B 6 25* 59** Brand C 39 76* 98** Brand D 6 27* 91** Brand E 15 49* 98** Brand F 3 8* 61** Brand G 12 54* 96** Brand H 1 5* 68** Brand I 3 21* 78** Average 12 36 83

Statistically significantly higher than *TOM; **Unprompted at p<0.05.

While all pasta sauce brands show light buyer prompted awareness response

statistically significantly higher then unprompted awareness response, 10 out of 13 pasta sauce brands have unprompted awareness response statistically significantly higher than TOM awareness response (see Table 35). For Brand M, unprompted awareness response is higher than TOM awareness response, while for Brand J and

Brand K, very few light buyers respond to both TOM and unprompted awareness measures.

Page 68: Understanding and Measuring Light Buyer Brand Equity

59

Table 35: Proportion of light buyer TOM, unprompted & prompted awareness response –Pasta Sauce.

Pasta Sauce (n=2095)

% Light Buyer TOM Unprom Promp

Brand A 25 56* 94** Brand B 37 62* 96** Brand C 7 26* 89** Brand D 5 33* 83** Brand E 1 6* 42** Brand F 3 13* 52** Brand G 3 15* 71** Brand H 4 29* 67** Brand I 2 17* 65** Brand J 1 2 46** Brand K 0 0 28** Brand L 9 48* 95** Brand M 6 19 63** Average 8 25 68

Statistically significantly higher than *TOM; **Unprompted at p<0.05.

Consistent with the first two categories, light buyer response for all soft drink brands (refer Table 36) shows prompted awareness is statistically significantly higher than unprompted awareness response. Unprompted awareness response is statistically significantly higher than TOM awareness response for 10 out of 14 brands. While not significantly different, the results for the remaining four brands show light buyer

response is higher for unprompted than for TOM awareness.

Table 36: Proportion of light buyer TOM, unprompted & prompted awareness response – Soft Drink.

Soft Drink (n=974)

% Light Buyer TOM Unprom Promp

Brand A 61 91* 97** Brand B 8 13 100** Brand C 3 4 99** Brand D 13 58* 99** Brand E 27 76* 100** Brand F 12 54* 99** Brand G 0 2 97** Brand H 0 52* 98** Brand I 0 5 98** Brand J 3 37* 97** Brand K 11 45* 95** Brand L 2 40* 98** Brand M 5 39* 97** Brand N 13 34* 100** Average 11 39 98

Statistically significantly higher than *TOM; **Unprompted at p<0.05.

Page 69: Understanding and Measuring Light Buyer Brand Equity

60

Again, for all whiskey brands, light buyer prompted awareness response is statistically

significantly higher than unprompted awareness response (see Table 37). Five out of 16 brands show unprompted awareness response statistically significantly higher than TOM awareness response. There is a trend present, where all other brands have unprompted awareness response higher than TOM awareness response.

Table 37: Proportion of light buyer TOM, unprompted & prompted awareness response –Whiskey.

Whiskey (n=785) % Light Buyer TOM Unprom Promp

Brand A 4 13* 100** Brand B 1 4 100** Brand C 0 3 100** Brand D 0 7 100** Brand E 6 24* 100** Brand F 4 19* 100** Brand G 10 28* 100** Brand H 6 14 100** Brand I 0 1 100** Brand J 0 0 100** Brand K 0 4 100** Brand L 4 13 100** Brand M 0 0 100** Brand N 0 9 100** Brand O 0 15 100** Brand P 3 20* 100** Average 2 11 100

Statistically significantly higher than *TOM; **Unprompted at p<0.05.

Across the four product categories (see Table 38 for a summary of results), for all brands, light buyer prompted awareness response is statistically significantly higher than unprompted awareness response. For 34 out of 52 brands (65%), light buyer unprompted awareness response is statistically significantly higher than TOM awareness response. All other brands, however, show light buyer response is higher for unprompted than TOM awareness. Results therefore support hypothesis 4a and 4b.

Table 38: Summary light buyer awareness results.

% Light Buyer TOM Unprom Promp

Tea (n=7005) 12 36 83 Pasta Sauce (n=2095) 8 25 68 Soft Drink (n=974) 11 39 98 Whiskey (n=785) 2 11 100 Average 9 28 87

Page 70: Understanding and Measuring Light Buyer Brand Equity

61

Building on the analysis of light buyer response across the three awareness measures,

binary logistic regression results for each awareness measure are compared. Table 39 presents results for the difference in non and light buyer response for each awareness measure.

All three awareness measures statistically significantly distinguish between response for non-buyers and light buyers. The strength of discrimination is highest for TOM awareness, while lowest for unprompted awareness. However, there is overlap in confidence intervals for each awareness measure. Results therefore indicate that while there is a difference in non and light buyer response, this difference is similar across awareness measures. The findings suggest that response for the two groups differs in line with the difficulty of the measure.

Table 39: Logistic regression results for non and light buyers across TOM, unprompted and prompted awareness measures.

Non to Light Wald Sig. Exp(B) 95% C.I.

Lower Upper TOM 4.3 <0.05 1.06 1.00 1.12 Unprompted 7.7 <0.05 1.03 1.01 1.05 Prompted 16 <0.05 1.05 1.02 1.07

Next, binary logistic regression results are compared across awareness measures for the difference between light and heavier buyer response (see Table 40). Only TOM awareness significantly distinguishes response for the two buyer groups. For unprompted and prompted awareness, light buyer response is similar to heavier buyer response.

Table 40: Logistic regression results for light and heavier buyers across TOM, unprompted and prompted awareness measures.

Light to Heavier Wald Sig. Exp(B)

95% C.I. Lower Upper

TOM 5.7 <0.05 1.04 1.01 1.08 Unprompted 3.4 0.07 1.02 1.00 1.03 Prompted 1.2 0.28 1.01 0.99 1.04

6.2 Awareness Discussion

This research is focused on understanding how light buyers respond to different brand awareness measures. In particular, this study aims to provide marketers with guidance

on which measure better captures light brand buyer response. Three awareness measures have been assessed for light brand buyer response in comparison with non-buyer and heavier buyer groups. Discussion is presented according to awareness

Page 71: Understanding and Measuring Light Buyer Brand Equity

62

measure, TOM, unprompted, then prompted awareness. The chapter concludes with a summary of key findings.

6.2.1 TOM Awareness Results indicate that buyers who have more experience with a brand via purchase/

consumption are more easily able to retrieve the brand name when given the category as a cue. Results are consistent with Wight’s (2010) study, which determines that for three consumer packaged goods categories, the proportion of respondents who recall a brand in TOM awareness is greater for users relative to non-users.

TOM allows respondents to recall only one brand. Light buyers may have the brand name stored in memory from prior purchase, however the linkage to the category cue is weak from infrequent exposure to the brand. Brand to category linkages are stronger for brands purchased more frequently, which are therefore retrieved more readily from memory, as per frequency and recency effects (Anderson 1983). As light buyers purchase from a repertoire of brands in the category (Ehrenberg 2000a), they are likely

heavier buyers of other brands. Therefore, competitor brands interfere with the accessibility of light buyer brand retrieval from memory (Alba & Chattopadhyay 1986). This is evidenced by, on average, less than a quarter of light buyers able to recall the brand in TOM awareness.

Though light buyer response is lower than for heavier buyers, light buyers are better able to recall the brand in TOM awareness than non-buyers. Logistic regression analysis significantly differentiates TOM awareness response for light buyers, non and heavier buyers. As light buyer response is not in line with either buyer group, future analysis should be performed separately for non, light and heavier buyers.

6.2.2 Unprompted Awareness Unprompted awareness response increases in line with buyer purchase frequency, from non-buyer to light to heavy buyers. Therefore, while more brands can be recalled than in TOM awareness, retrieval from memory remains the same. Brands purchased more frequently are recalled first, followed by those purchased infrequently, then brands encountered only via marketing efforts (Alba & Chattopadhyay 1986; Hutchinson 1983; Hutchinson, Raman & Mantrala 1994).

Approximately a third of light buyers are able to recall the brand when given multiple opportunities for response (i.e., in unprompted awareness). Given that brands are recalled from memory in line with frequency and recency of exposure (Anderson 1983),

Page 72: Understanding and Measuring Light Buyer Brand Equity

63

light buyers are likely to respond with heavier brands in their repertoire. Competitor brands linked to the category cue limit the accessibility and retrieval of brands purchased infrequently. For light buyers, recall measures may provide an indication of brands with stronger linkages to the category cue in memory that are accessible and more easily retrieved from memory. However, the presence of competitor linkages means that unprompted awareness may not capture all brands linked to the category node in light buyer memory (in line with Tulving & Pearlstone's discussion on accessibility and availability).

Logistic regression analysis significantly differentiates unprompted awareness response for non and light buyers, with unprompted awareness response higher for light than for non-buyers. Results for unprompted awareness cannot, however, significantly distinguish between light and heavier buyer groups, despite results trending to heavier buyer response. Therefore, when multiple brands can be recalled, light and heavier buyers both have the ability to retrieve the brand from memory. Given a lack of statistically significant difference between light and heavier buyer unprompted awareness response, analysis should be performed separately for two groups: non-buyers and buyers.

6.2.3 Prompted Awareness As with unprompted awareness, light buyer response to prompted awareness is

statistically significantly higher than non-buyer response. This is as expected and consistent with Romaniuk and Wight’s (2009) findings that users are 1.7 times more likely to respond to prompted awareness than non-users. The present study finds, on average across all brands, a 21-percentage point difference between light and non-buyer prompted awareness response. Results range from a nine percentage point difference for soft drink to a 43-percentage point difference between non and light buyers for whiskey. Logistic regression confirms that prompted awareness response is significantly different for the two groups.

It is important to note that light buyers respond to prompted awareness in a similar manner to heavier buyers. Prompted awareness is less difficult than recall measures, where respondents are presented with the brand name and merely indicate brands linked to the category cue in memory structures (Tulving & Pearlstone 1966). Prompted awareness reduces limitations of competitor interference present in TOM and unprompted recall measures, allowing information for brands purchased frequently and infrequently to be retrieved. As both light and heavier buyers’ responses are similar for

Page 73: Understanding and Measuring Light Buyer Brand Equity

64

the measure, there is no need to conduct separate analyses for the two brand buyer groups. However, marketers should perform analysis separately for non-buyers and buyers.

6.2.4 Summary Results from this study are in line with prior research, where light buyer response

increases according to the difficulty of the measure, from TOM to unprompted to prompted awareness (Laurent, Kapferer & Roussel 1995; Romaniuk et al. 2004; Romaniuk & Wight 2009; Wight 2010).

Results from analysis of non and light buyer awareness demonstrate that whilst there is a difference in response for the two groups, the extent of this difference is similar across awareness measures. Findings from this thesis suggest that response for buyer groups increases at a similar rate, in line with the difficulty of the measure. Thus, awareness measures merely differ in the level of buyer response captured. When the marketer’s objective is to determine absolute light buyer brand awareness level,

prompted awareness should be examined. Limitations of competitor linkages and accessibility present in the TOM and unprompted recall measures are removed, resulting in a greater light buyer response level for prompted awareness.

Secondly, this thesis determines that light buyers are better able to retrieve the brand name from memory than non-buyers across all awareness measures. This suggests that analysis should occur separately for both non and light buyer groups. In comparison to heavier buyers, light buyer response is significantly lower than heavier buyers for TOM awareness. However, light and heavier buyers respond in a similar manner to unprompted and prompted awareness. Therefore, while analysis should occur separately for light and heavier buyers for TOM awareness, there is no need to separate light and heavier buyer groups for unprompted and prompted awareness analysis. Results are in line with previous findings by Wight (2010), who suggests separating brand awareness analysis for users and non-users. This research extends Wight’s work to provide a deeper understanding of analysis and interpretation of results for light buyers.

Page 74: Understanding and Measuring Light Buyer Brand Equity

65

6.3 Chapter Summary

This chapter has analysed and compared light buyer results for top of mind,

unprompted and prompted awareness measures. Key points from this chapter are:

• For all three awareness measures, light buyer response is significantly higher than non-buyer response.

• For TOM awareness, light and heavier buyer response is significantly different. However, for unprompted and prompted measures, light buyers respond in a similar manner to heavier buyers.

• For TOM awareness, analysis should occur separately for non, light and heavier buyers. For unprompted and prompted awareness, analysis should occur separately for non-buyers and buyers.

• Light buyer response increases from TOM to prompted awareness. Therefore, to assess light buyer absolute awareness level, marketers should look to prompted awareness.

The next chapter will present Part Two: Brand Image.

Page 75: Understanding and Measuring Light Buyer Brand Equity

66

Part Two: BRAND IMAGE The first brand equity measure of interest, brand awareness, has been examined with the results of the present study stated. This thesis now invites the reader to consider a

second brand equity measure, brand image. Given the difference in literature and the research approach taken for brand awareness and brand image, the measures are presented separately.

Part Two comprises Chapters Seven to Nine. Brand image is introduced and hypotheses stated in Chapter Seven. Chapter Eight presents the research approach, while Chapter

Nine presents the results for brand image.

Page 76: Understanding and Measuring Light Buyer Brand Equity

67

Chapter 7 BRAND IMAGE This chapter introduces and explains the importance of the second brand equity measure of interest, brand image. The chapter details different brand image measures s

and justification for those included in this study. Measures are then discussed from a light brand buyer perspective, with hypotheses to be tested stated.

7.1 What is Brand Image?

Brand image is how consumers perceive the brand, assessed by the associations

(operationalised as brand attributes) held in consumer memory that are linked to the brand node (Aaker 1992; Keller 2005). Brand attributes are “descriptive features that characterise a product/ service – what a consumer thinks the product is or has and what is involved with its purchase/consumption” (Keller 1993, p.4). Figure 5 in Chapter Three provides an example of different attributes linked to the brand node, Cadbury.

Brand associations are formed and strengthened through co-presentation of information and the brand name (Anderson & Bower 1979; Van Osselaer & Janiszewski 2001). This co-presentation can occur through brand purchase and usage and exposure to marketing activities (Nedungadi 1990). For example, advertising depicting people drinking Coca Cola on a beach can build a linkage between the two concepts. Consequently, when cued with the situation ‘going to the beach’, there is a higher probability that the brand will be retrieved from memory than brands with weaker linkages to the situation in memory.

7.2 The Importance of Brand Image

In purchase/consumption situations, consumers use attributes to retrieve the brand

from memory (Nedungadi 1990), identify the brand from competitors (Zaichkowsky 2010) or compare brands within a category (Ratneshwar et al. 2001). Brand attributes encompass desires that the product/service aim to fulfil (Ratneshwar & Shocker 1991), functional aspects associated with using the brand and benefits provided to the user (Keller 1993). The set of associations linked to the brand can also help consumers form an attitude of the product/brand for future reference (Aaker 1992; Keller 2005). As consumers can have none, one or multiple associations linked to the brand in memory (Driesener & Romaniuk 2006), this measure of Customer Based Brand Equity helps gauge associative network structures, along with the strength of the association in memory.

Page 77: Understanding and Measuring Light Buyer Brand Equity

68

As brand attributes are often depicted in marketing efforts, such as advertising, brand image can be used to not only identify the current position of the brand, but to evaluate advertising effectiveness in terms of its ability to reinforce or promote the brand’s image (Driesener & Romaniuk 2006).

7.3 Brand Image Measurement

Though research has examined brand image response for brand users and non-users

(e.g., Barwise & Ehrenberg 1985; Bird, Channon & Ehrenberg 1970; Bird & Ehrenberg 1970; Castleberry & Ehrenberg 1990; Dall'Olmo Riley 2012; Romaniuk, Bogomolova & Dall'Olmo Riley 2012), no research to date has explored how light brand buyers respond to these different measures. Determining which brand image measure better captures light buyer response will aid market researchers when designing brand equity surveys.

Brand image measures fall into two main categories: scaling and sorting (Joyce 1963). Scaling measures identify the extent that the respondent feels the brand has the association, whereas sorting measures require the respondent to indicate if the brand has the association or not (Barnard & Ehrenberg 1990). The following sections will outline the two types of image measures, before stating the scope of research.

7.3.1 Scaling Measures As comparative image measurement studies have previously included ranking and

rating scale measures (e.g., Barnard & Ehrenberg 1990; Dolnicar, Grün & Leisch 2011; Driesener & Romaniuk 2006), it is important to discuss these methods in terms of light buyers.

A common scaling technique is asking respondents to rank a list of brands. For each attribute presented, the respondent is required to rank the brands, from the first being most strongly associated with the attribute to the last being least associated with the

attribute (Barnard & Ehrenberg 1990; Driesener & Romaniuk 2006). Research shows that brands purchased frequently have more associations linked to the brand than brands purchased infrequently or not at all in the timeframe (Bird, Channon & Ehrenberg 1970; Dall’Olmo Riley 2012; Romaniuk, Bogomolova & Dall’Olmo Riley 2012). In addition, Ehrenberg (2000a) notes that buyers purchase from a repertoire of brands, meaning that light buyers of a brand can also be heavier buyers of other brands. When asked to rank brands, those purchased frequently are more likely to be ranked higher than brands purchased infrequently. Therefore, it becomes difficult to determine the cause of a low ranking, either due to the brand having the association (but to a lesser extent than other

Page 78: Understanding and Measuring Light Buyer Brand Equity

69

brands) or that the association does not exist in the buyer’s associative network structure.

It is for this reason that the ranking measure is not included in this research. This decision is supported by Joyce’s (1963, p.54) comments that “if only one or a very few brands (or attributes) have to be selected or ranked using a comparative technique, there is clearly a danger of the most obvious brand or attributes being mentioned first by most people and of movements among the less obvious ones, which could be potentially important, being blurred or altogether lost”.

A second scaling measure of brand image is asking respondents to rate brands based on their level of agreement that the brand is associated with the attribute presented (Driesener & Romaniuk 2006; Haley & Case 1979). Respondents are typically asked to respond on a 5- or 7- point scale, e.g. from strongly agree to strongly disagree (Likert 1932). Respondents are forced to make a selection in scale questions. Respondents must interpret the question, search their memory for relevant information, form a judgement and decide where this judgement is best represented on the scale (Krosnick 1999). This process requires a lot of cognitive effort, which induces respondent fatigue

(Dolnicar & Rossiter 2008; Shulman 1973).

A respondent who has less knowledge of the brand (i.e. light or non-buyer) may not have the motivation to go through these cognitive steps, lack brand information in memory, and/or be unable to retrieve this information from memory. This may lead to response biases including acquiescence (agreement bias), extreme responding and midpoint responding, all common in rating questions (Baumgartner & Steenkamp 2001; Diamantopoulos, Reynolds & Simintiras 2006; Rossiter, Dolnicar & Grun 2015).

Finally, Joyce (1963), and Barnard and Ehrenberg (1990) note that a scale battery of questions takes (on average) twice as long to administer than sorting questions. Driesener and Romaniuk (2006) test this statement for two categories (car market and financial services market) for ranking and rating scale measures and a pick any sorting measure. Findings show that the sorting measure outperformed both scale measures, being 40% quicker for overall measurement and 50% quicker to administer per attribute tested. Similarly, Dolnicar, Grun and Leisch (2011), in a later study, find that a forced choice sorting measure is 50% quicker than a rating scale measure.

Page 79: Understanding and Measuring Light Buyer Brand Equity

70

Scaling measures therefore have limitations that could restrict light buyer response and/or induce response bias from long administering times and the amount of cognitive effort required from respondents. Given the limitations of scaling techniques, these image measures will not be included in the research design.

7.3.2 Sorting Measures Results from this thesis will provide marketers with guidance on brand equity survey

design. Brand equity surveys are time-consuming and consist of a number of different measures, including awareness, brand evaluation, and product use, along with brand image measurement (Aaker 1992; Keller 1993). Measures within equity surveys that are quick to administer will help decrease the overall length of the survey. From a managerial perspective, a shorter survey will mean reduced costs. While from a researcher perspective, a shorter administration time is likely to decrease response bias, such as fatigue effects, which will in turn improve the quality of results (Galesic & Bosnjak 2009).

As discussed in the previous section, sorting techniques are quicker and the preferred image measurement approach, requiring the respondent to only select attributes linked to the brand in memory (Barnard & Ehrenberg 1990). Prior comparative image measurement studies (i.e., Driesener & Romaniuk 2006; Dolnicar & Grün 2007) determine that sorting techniques capture similar information as scaling techniques, indicating that no information will be lost by omitting these measures from further testing.

This research draws from Joyce’s (1963) comparison of brand image measures, where two sorting measures are analysed: pick any and forced choice binary. Pick any and forced choice binary approaches are introduced with the measures’ advantages and limitations in terms of capturing light buyer associations detailed.

7.3.2.1 Pick-Any Measure (PA)

In the pick-any (PA) approach respondents are given an attribute (or brand) and asked

which brands (or attributes) they associate with the cue (Holbrook, Moore & Winer 1982) (e.g., Figure 10). Respondents are not required to make a selection if they feel the brand does not have the association, instead selecting a ‘none of these’ option (Joyce 1963). As multiple brands can be selected for each attribute, buyers are able to respond for all brands in their repertoire (Driesener & Romaniuk 2006), regardless of whether the brand is purchased frequently or infrequently.

Page 80: Understanding and Measuring Light Buyer Brand Equity

71

Figure 10: Example of a pick-any image measurement question, cereal category.

Good value for money ☐ Kellogg’s Cornflakes ☐ Frosties ☐ Alpen ☐ Weetabix ☐ Shredded Wheat ☐ Special K ☐ Rice Krispies ☐ All Bran ☐ None of these

Source: Barnard & Ehrenberg (1990)

Advantages

Research comparing the PA, rating and ranking measures find similar brand-level results (0.90 correlation between measures) (Driesener & Romaniuk 2006). PA, however, has long since been determined the quicker method and consequently is suggested for marketer use (Joyce 1963). As such, the PA technique is used by “nearly all brand-tracking companies, including the worldwide leader, Millward Brown” (Rossiter, Dolnicar & Grun 2015, p.3).

The PA measure requires that respondents indicate brand and attribute linkages present in memory structures. Respondents are not forced to make a selection for each brand and attribute. Therefore, light buyers, who have less brand information in memory from

infrequent purchase, are less likely to overthink the task and respond with associations that are not contained in memory structures. The quick nature of the measure also better reflects what happens in choice situations, where buyers assess only a small portion of brands (and attributes) (Nedungadi 1990).

Limitations

The PA approach is a free choice measure that allows respondents to select brands linked to the association, omitting all other brands. Rossiter, Dolnicar and Grun (2015) state that this is a limitation of the technique, where omission leads to under-reporting. The authors note that it is difficult to determine if the respondent does not have the association in memory or whether they feel they have fulfilled the requirements of the question by selecting some associations, skipping over others and are moving forward in the survey.

Evasion and under-reporting is particularly relevant when considering light brand buyer response (as detailed in Ludwichowska 2013). Light buyers have fewer and weakly established brand to association linkages in memory from infrequent purchase, which increases the difficulty of information retrieval (Anderson & Bower 1979). This may result in light buyers not giving a response for a particular brand and/or attribute.

Page 81: Understanding and Measuring Light Buyer Brand Equity

72

7.3.2.2 Forced-Choice Binary Measure (FCB)

In the forced choice binary (FCB) measure respondents are provided with a list of

attributes for each brand and asked to select a position in the binary answer options, yes/no9 (see Figure 11) (Dolnicar, Rossiter & Grun 2012).

Figure 11: Example of a forced-choice binary image measurement question, laundry detergent category.

Radiant

Gets clothes very clean ☐ Yes ☐ No Whitens whites ☐ Yes ☐ No Attractively priced ☐ Yes ☐ No Noticeably freshens clothes and bed linen ☐ Yes ☐ No Brightens colours ☐ Yes ☐ No Removes tough stains ☐ Yes ☐ No Cleans well in cold water ☐ Yes ☐ No

Source: Dolnicar, Rossiter & Grun (2012)

Advantages

Similar to the PA approach, Dolnicar and Grun (2007) compare the FCB measure with a 7-point scale measure, concluding that both give similar results, are equally pleasant, yet the FCB measure is quicker to complete. Later testing (Dolnicar, Grün & Leisch 2011) again finds that respondents take 50% longer to complete a 7-point scale approach than FCB questions.

The PA method allows light buyers to omit, and under-report beliefs about the brand (Rossiter, Dolnicar & Grun 2015). The forced-choice binary technique instead requires that respondents make a judgement for all associations and brands, therefore making sure that time is spent considering each attribute, in turn capturing all brand associations held in light buyer memory.

9 In a recent paper, Rossiter et al. (2015) have built on the FCB approach to establish the Doubly Level-Free with Individual Inferred Thresholds (DFL IIST) method. In their method, the respondent is presented with attributes that are free of any indication of intensity, i.e. removes tough stains instead of removes tough stains well. The respondent uses their own threshold to assess whether it removes tough stains well enough to justify a ‘yes’ response. This method is noted, however is not incorporated in the research design as the researcher chose to limit the scope of analysis to more established techniques within brand image literature. Drawing from established measures in the literature gives the researcher more guidance on questionnaire design and prior results for comparison for a range of product categories and conditions.

Page 82: Understanding and Measuring Light Buyer Brand Equity

73

Limitations

Forcing a response can be particularly difficult for those with little to no information about the brand in memory, i.e., non or light buyers. By forcing a response, it becomes unclear if a respondent has selected ‘no’ because they do not link the association to the brand or whether the respondent does not have the relevant information in memory to access and make a judgement (Holbrook, Moore & Winer 1982). The method therefore has an underlying assumption that all respondents 1) have sufficient knowledge of the brand from exposure through purchase, marketing efforts or word of mouth, and 2) are able to retrieve this information when prompted.

7.3.3 Prompting Method Within each image measure, there are two alternative methods of presentation: attribute or brand prompting (Joyce 1963). The two methods differ in the cue provided to respondents as the initial activation point in memory. I.e. attribute prompted methods present the respondent with a list of brands and ask for a response for each attribute; brand prompted methods present the respondent with a list of attributes and ask for a response for each brand. The examples in previous sections demonstrate an attribute prompted approach for PA and a brand prompted approach for FCB. Figure 12 and Figure 13 provide examples of the two remaining image measurement conditions: PA

brand prompted and FCB attribute prompted.

Figure 12: Example of a pick-any, brand prompted image measurement question, cereal category.

Kellogg’s Cornflakes ☐ Sort of cereal you come back to ☐ Popular with all the family ☐ Reasonably priced ☐ Very easy to digest ☐ Good value for money ☐ Tastes nice ☐ Stays crispy in milk ☐ None of these

Page 83: Understanding and Measuring Light Buyer Brand Equity

74

Figure 13: Example of a forced-choice binary, attribute prompted image measurement question, laundry detergent category.

Gets clothes very clean

Omo ☐ Yes ☐ No Spree ☐ Yes ☐ No Radiant ☐ Yes ☐ No Cold Power ☐ Yes ☐ No Surf ☐ Yes ☐ No Dynamo ☐ Yes ☐ No

Past studies on PA and FCB measures have used a combination of attribute and brand prompted approaches. Joyce (1963) first compares attribute and brand prompting for PA and FCB. Studies following Joyce’s (1963) comparative study have incorporated a

PA attribute prompted design (e.g., Barnard & Ehrenberg 1990; Bird, Channon & Ehrenberg 1970). While for the FCB measure, researchers have used a brand prompted approach (e.g., Dolnicar, Grün & Leisch 2011; Dolnicar, Rossiter & Grun 2012). Sharp and Romaniuk (2002) examine attribute to brand and brand to attribute links for the PA approach, however this is from a response stability perspective.

No studies to date have examined the affect different methods of prompting have on light brand buyer response. Light buyers have less information in memory from infrequent purchase and this makes it difficult to retrieve from memory (Anderson & Bower 1979). Therefore, it is important to identify the prompting method that activates and retrieves more information from light buyer memory structures. This will provide marketers with information on the brand to attribute linkages present in memory.

Page 84: Understanding and Measuring Light Buyer Brand Equity

75

7.3.4 Research Scope Both PA and FCB image measures have advantages of a quicker administration time

and ease of response, important for use in brand equity surveys. The two methods differ in terms of response type (free or forced), which affect light buyer ability to respond. Table 41 highlights key brand image studies mentioned, the method(s) analysed and the context of analysis. Though research has emphasized the importance of distinguishing between brand buyer and non-buyer response (and variations of this, i.e. infrequent and frequent; current, former, never-trieds), an important aspect that has yet to be examined is how the measures capture responses from light brand buyers. In addition, only Joyce (1963) incorporates an attribute and brand prompted design for the two image measures.

To extend current literature, this research will analyse light buyer response to PA and FCB image measures using attribute and brand methods of prompting for response. The four conditions tested are:

1. Pick Any - Attribute Prompted

2. Pick Any - Brand Prompted

3. Forced Choice Binary - Attribute Prompted

4. Forced Choice Binary - Brand Prompted

This research will update Joyce’s (1963) study, analysing the four measurement conditions together in a modern, online context. Results will help understand light buyer brand equity and update literature for two established image measures and two methods of prompting.

Page 85: Understanding and Measuring Light Buyer Brand Equity

76

Table 41: Summary of key image measurement studies.

Study Measure Prompt Country Category Context Survey Method

Joyce (1963)

PA, Forced

Choice & Rating

Attribute & Brand UK Toilet

Paper Response

Level Face-to-

face

Bird & Ehrenberg (1970) PA Attribute UK 10

Categories Usage N/A*

Bird, Channon & Ehrenberg (1970) PA Attribute UK

7 different food/non-

food

Former, Current &

Never-tried buyer

N/A*

Barwise & Ehrenberg (1985) PA Attribute UK

Cereal & Washing Powder

Users & Non-users N/A*

Barnard and Ehrenberg (1990)

PA, Rating & Ranking

Attribute (PA &

Ranking) Brand

(Rating)

UK 5 different food/non-

food

Infrequent/Frequent/

Non-Buyer

Face-to-face

Driesener and Romaniuk (2006)

PA, Rating & Ranking Attribute Australia Cars

User/Non-

user

CATI (Comp. assisted

telephone interview)

Dolnicar, Grun and Leisch (2011)

FCB & Rating Brand Australia Fast Food

Response Level /

Stability N/A

Dall Olmo’Reilly (2012) PA Attribute UK & US

5 grocery (UK) 2

grocery, 2 services

(US)

Stability Infrequent/ Frequent

Online

Dolnicar, Rossiter and Grun (2012)

PA, FCB & Rating Brand Australia Laundry

Detergent

Response Level/

Stability Online

Romaniuk, Bogomolova & Dall’Olmo Riley (2012)

PA Attribute

15 countries

(developed &

emerging)

30 categories

(15 packaged

goods)

Former, Current &

Never-tried

buyers

Online; Face-to-

face; Telephone

* Data is derived from the Advertising Planning Index (API) of the British Market Research Bureau

Page 86: Understanding and Measuring Light Buyer Brand Equity

77

7.4 Light Brand Buyer Associations

The aim of this research is to determine which brand image measure better captures

light brand buyer response. The ideal measure for use in CBBE surveys should:

1. Be easy for light buyers to respond to. Light buyers have less knowledge

about the brand in memory and have difficulty retrieving this information from memory networks compared to heavier buyers (consistent with recency and frequency effects, Anderson 1983; Nedungadi 1990). A measure that provides multiple entry points in memory will increase the probability that more associations will be activated and retrieved in image measures (Ratneshwar et al. 2001). Thus, a measure that captures a higher proportion of light buyer response is valuable. However, the measure must also;

2. Reflect associations of the brand rather than those typical of the category.

Light buyers should respond with attributes that are linked to the brand in memory, which will help identify the brand’s strengths and weaknesses. However, this becomes difficult to assess if light buyer response is high for all attributes tested. A measure where light buyers select a majority of attributes per

brand may reflect associations typical of the category rather than those linked to the brand in memory (Barsalou 1983). This may occur when light buyers are forced to select a response across a number of attributes for a brand they have less knowledge about in memory. In addition, as light buyers purchase the brand once in a given period, they have less opportunity to develop linkages between different attributes and the brand (Desai & Hoyer 2000). Therefore, a measure that captures fewer attributes per brand will more accurately reflect information stored in light buyer memory networks and the probability of retrieval in purchase/consumption situations.

An image measure that fulfils the two criteria outlined should consequently be better able to detect change in light buyer response over time. The following sections draw from these criteria of a good image measure for light buyers to develop hypotheses that this research will test. PA and FCB measures are discussed separately for different methods of prompting, with PA and FCB measures then compared.

Page 87: Understanding and Measuring Light Buyer Brand Equity

78

Brand image analysis tests different measures and different methods of prompting for response. Therefore, the scope of research is limited to examining light brand buyer response rather than a comparison across different buyer groups.

7.4.1 Attribute Versus Brand Prompting Discussion draws from the framework introduced in Chapter Three, which details

knowledge of light buyer memory structures and retrieval of information from these structures. From this discussion, conclusions are drawn for light buyer response10 to PA and FCB measures.

Cue Activation in Memory Structures

Associations are cue dependent, where the cue provided to the respondent determines the information retrieved from memory structures. When prompted with a brand name with attributes presented below, the brand name becomes the initial node activated in memory. This provides a single entry point in memory for retrieval, as per the Spreading Activation Theory of Memory (Collins & Loftus 1975), where activation spreads along paths of the network to activate further nodes for retrieval. Given that light buyers have less knowledge of the brand in memory from infrequent purchase, linkages from the brand name to associations are weak, making it harder for associations to be retrieved from the network (Anderson 1983). If the brand itself is hard to retrieve in memory, then any information linked to the brand will also be harder to retrieve (Barsalou 1983). This decreases light buyer ability to respond.

Prompting via attributes provides multiple entry points in memory for activation and retrieval (Barsalou 1983; Ratneshwar et al. 2001). As each attribute activates different

parts of consumer memory, the number and composition of brands evoked differs in each instance (Holden & Lutz 1992). Therefore, there is a greater opportunity to capture more light buyer associations.

10 Response is operationalised for PA as selecting an attribute or brand, while for FCB as selecting ‘yes’ for an attribute or brand.

Page 88: Understanding and Measuring Light Buyer Brand Equity

79

Figure 14 provides a representation of this discussion, listing hypothetical brands and attributes retrieved when activated via an attribute or brand cue. This example demonstrates that when attribute prompted the cereal brand, Rice Krispies, may be evoked from cues ‘fun for kids to eat’, ‘stays crispy in milk’ and ‘popular with all the family’. However, when prompted with the Rice Krispies brand, which the respondent

purchases infrequently, only ‘fun for kids to eat’ and ‘stays crispy in milk’ may be evoked.

Figure 14: Representation of brand and attribute cue activation.

Linkage Pathway: Attribute-to Brand or Brand-to-Attribute

Furthermore, the link between a brand and attribute is bi-lateral, where either the brand name or the cue can begin activation in consumer memory structure (Aaker 1996a). However, the strength of the links differs based on how often the pathway is reinforced (Collins & Loftus 1975). In purchase/consumption situations, buyers often aim to fulfil a goal or need and rely on attributes to retrieve brands from memory for consideration (Nedungadi 1990; Ratneshwar et al. 2001). The pathway from attribute to brand is therefore reinforced and strengthened over time.

Alternatively, buyers may activate the brand node first (e.g. by seeing it in the supermarket) or store window, which may evoke associations linked to the brand name. As light buyers do not purchase the brand often they spend little time thinking specifically about the brand. In addition, for everyday product categories, where habitual purchasing is common (Ehrenberg 2000a), little thought might be given to a brand regardless of how frequently it is purchased.

Page 89: Understanding and Measuring Light Buyer Brand Equity

80

Therefore, image measures that replicate thought processes that occur in purchase/ consumption situations, which have stronger linkage pathways from reinforcement, have a greater likelihood of retrieving information from memory, i.e. attribute rather than brand prompted. Figure 15 provides a representation of the linkages between brand and attribute cue, where the weight of the arrow depicts the perceived linkage strength.

Figure 15: Representation of strength between brand and attribute linkages.

7.4.1.1 PA Measure

The PA measure allows the respondent to indicate only what is evoked from the cue, giving the marketer a more realistic indication of what happens in purchase/consumption situations. Light buyers have less knowledge of brands in memory. Prompting with a brand activates a single node in memory from which to retrieve associations. When given the option to select ‘no response’, rather than searching for weak and/or non-existent linkages in memory, light buyers are likely to make this selection.

Thus, for the PA measure, as there are more opportunities for the brand to be activated and come to mind, response for attribute prompted will be greater than for brand prompted, where activation in memory is limited to a single cue. This leads to the first image hypothesis.

H5: Light brand buyer response to PA will be higher when attribute

prompted than when brand prompted.

7.4.1.2 FCB Measure

Forced response requires that buyers search memory structures for brand to attribute

linkages that are weakly established. When forced to make a response for a brand they know little about, light buyers can select ‘no’ for all associations or brands. However, there is a higher probability as respondents make their way down the attribute and/or brand list that they will make at least one selection. Joyce (1963) notes that forcing response can result in ‘polite’ response, or the respondent making a selection, ‘giving the brand the benefit of the doubt’, i.e. don’t really know much about this brand, but I’m sure it would be ‘good for kids’.

Attribute Brand

Page 90: Understanding and Measuring Light Buyer Brand Equity

81

Regardless of the prompting method, FCB requires that light brand buyers consider all options presented. Therefore, the probability of response, whether due to response bias or giving no response (selecting ‘no’) is the same whether attribute or brand prompted. In addition, Joyce’s (1963) results show no statistically significant difference between forced choice response when brand (M=85%) or attribute prompted (M=87%). This leads to the next hypothesis.

H6: There will be no difference in light brand buyer response to FCB

whether brand prompted or attribute prompted.

7.4.2 PA Versus FCB Joyce (1963) finds for both brand and attribute prompting methods that FCB has a higher response than PA (38 percentage point difference attribute prompted; 25 percentage point brand prompted). When testing for response stability for the two approaches, Dolnicar, Rossiter and Grun (2012) similarly find that the average response for a combined buyer and non-buyer sample is higher for FCB (M=71%) than for PA (M=36%) when using a brand prompted method.

In the FCB measure, given respondents are required to consider all brands (or attributes) as they move down the list, there is a higher probability that a brand (or attribute) not previously considered in PA will be selected. This could be due to a response bias, with respondents selecting the brand (attribute) either out of fatigue, or ‘politeness’ (Joyce 1963). Light buyers may be particularly prone to response bias, given they have less information in memory to draw from, which requires more cognitive effort compared to brands purchased frequently. Alternatively, the extra time spent considering each brand may activate and retrieve information that the PA approach misses (Rossiter, Dolnicar & Grun 2015).

Results from past studies suggest that FCB will capture a higher percentage of response, regardless of prompting approach. This leads to the final hypothesis.

H7a: Light brand buyer response will be higher for FCB than for PA when

attribute prompted.

H7b: Light brand buyer response will be higher for FCB than for PA when

brand prompted.

Page 91: Understanding and Measuring Light Buyer Brand Equity

82

7.5 Chapter Summary

This chapter has introduced the second brand equity measure of interest, brand image.

Key take-outs include:

• Brand image is how consumers perceive a brand, measured by examining the brand attributes linked to the brand name in memory.

• Measuring brand image allows marketers to assess the brand’s position in the market and evaluate advertising effectiveness.

• This study tests two image measures and two methods of prompting for light brand buyer response (refer Table 42). Image measures selected are highly cited in literature and commonly included in brand equity survey design.

Table 42: Summary of brand image measures for testing. Image measure Approach Example

Prompt Response

type

Pick-any Which brands do you associate with ‘Tastes great’? Select all brands that apply. Attribute Free

Forced-choice binary

Thinking about the brand, Omo, select a response for each attribute. Yes/No Brand Forced

The following chapter details the research approach taken for data collection and analysis.

Page 92: Understanding and Measuring Light Buyer Brand Equity

83

Chapter 8 IMAGE RESEARCH APPROACH This chapter details the research approach for Part Two: Brand Image. It provides a detailed description of the data sets used for analysis, buyer classification,

operationalisation of equity measures and techniques for data analysis.

8.1 Primary Data

In Joyce’s (1963) study, a split sample is collected in order to compare between PA and

FCB image measures using attribute and brand prompting methods. Part two replicates and extends Joyce’s work to examine light buyers, where primary data was collected via an online survey11. Using primary data allowed analysis for the same product categories across image measures and prompting methods for the same time period. Limitations associated with using secondary data, such as no control over survey design and misinterpretation of variables, were removed.

Data was collected in the United Kingdom (UK). Early research on brand image measurement has been conducted in the United Kingdom (e.g., Barnard & Ehrenberg 1990; Joyce 1963). More recent development in the field by Driesener and Romaniuk (2006) and Dolnicar et al. (2011; 2012) have chosen to sample consumers residing in Australia. Collecting data in the UK allows comparison between studies conducted in Australia for brand buyer and non-buyer response for each method. In addition, the results from this study expand those in Barnard and Ehrenberg’s (1990) research, who test an attribute prompted pick any method for the same product category, breakfast cereal, across a similar set of attributes in the UK.

8.2 Product Categories and Timeframe

Two consumer packaged goods categories were selected for testing: breakfast cereal

and butter/margarine. Drawing from TNS UK Kantar 2012 panel data across a 12-month period, breakfast cereal and butter/margarine categories have an average purchase frequency of 16.6 and 17.5 respectively. As both categories are repertoire markets, where buyers purchase from multiple brands (Ehrenberg 2000a), a higher purchase frequency allowed greater ability to distinguish between light and heavier buyers of

11 This study was approved by the University of South Australia’s Human Ethics Review Committee. This study was deemed to pose no risks, deceptions, or harm to potential respondents.

Page 93: Understanding and Measuring Light Buyer Brand Equity

84

different brands during analysis. In addition, the two categories have penetration above 90%, indicating that there will be high respondent eligibility to participate in the survey.

A 12-month timeframe was selected for this study. As detailed in section 5.5, a longer timeframe allows greater ability to capture a representative sample of all types of buyers, ranging from non-buyers to heavy buyers. A longer timeframe allows greater distinction between light and heavier buyers (Ehrenberg 2000a).

8.3 Questionnaire Design

This research tests nine brands and 15 attributes for each product category, with detail

on their selection provided. Previous image measurement studies (see Table 43) have included an average of seven brands and ten attributes for testing. As detailed in Chapter Seven, many of these studies collect data via face-to-face or telephone interviews, which may limit the number of items presented to respondents to reduce the interview time or respondents forgetting brands/attributes presented. Data for the present study is collected via an online survey, which allows a greater number of brands and attributes to be displayed at one time. Furthermore, the online survey does not contain questions outside the scope of this research, where the addition of more brands and attributes will not greatly affect the survey time or induce respondent fatigue.

Table 43: Number of brands and attributes tested in image studies.

Study Country Category # Brands

# Attributes

Joyce (1963) UK Toilet Paper 6 9

Barnard and Ehrenberg (1990) UK Five different food/non-food 8-9 12-13

Driesener and Romaniuk (2006) Australia Cars 6 10

Dolnicar, Grun and Leisch (2011) Australia Fast Food 6 11 Dolnicar, Rossiter and Grun (2012) Australia Laundry

Detergent 6 7

Average - - 7 10

8.3.1 Brands Tested In addition to having a high average purchase frequency, the categories selected were required to have multiple brands with high penetration. If the category selected contained a clear market leader, with the remainder of the category comprising smaller

brands, results would be limited in the number of heavy buyer response. A brand’s customer base is made up of many light buyers and few heavier buyers (Ehrenberg 1959). As brands grow they gain more customers who purchase slightly more frequently

Page 94: Understanding and Measuring Light Buyer Brand Equity

85

(Ehrenberg, Goodhardt & Barwise 1990). Larger share brands will provide more appropriate response levels to compare across light and other types of brand buyers (i.e. non and heavier buyers) within and across image methods.

The top nine brands in terms of penetration (number of buyers) were identified using TNS UK Kantar household panel data for inclusion in the questionnaire. Brand names are omitted for confidentiality reasons.

8.3.2 Attributes Tested Keller (1993) states that there are three types of brand associations that should be included in brand image questions to assess Customer Based Brand Equity: attributes, benefits and attitudes. Each type of association is defined and discussed in detail:

Attributes: “Descriptive features that characterise a product/service – what a consumer

thinks the product is or has, and what is involved with its purchase/consumption” (Keller 1993, p.4).

Product-Related Attributes – Aspects of the product required to perform the

function sought by consumers. Attributes also refer to characteristics related to the product’s physical composition, including descriptive features of the product that help define the brand’s identity (Lefkoff-Hagius & Mason 1993; Zaichkowsky 2010).

Non-product Related Attributes – Four external aspects of the product that

relate to its purchase/consumption.

Price information. These attributes are transferable across product categories and include attributes such as, ‘reasonably priced’ and ‘good value for money’ (Keller 1993).

Packaging or product appearance information. For example, colours or

package design (Zaichkowsky 2010). These aspects can be tested in separate ‘distinctive asset’ surveys (Romaniuk 2013).

User imagery. Who or what type of person purchases and/or consumes

the product. This type of attribute can include demographic factors (i.e. age, gender) and psychographic factors (i.e. attitude towards environment) (Lefkoff-Hagius & Mason 1993).

Page 95: Understanding and Measuring Light Buyer Brand Equity

86

Usage imagery. Purchase and/or consumption situations (Ratneshwar & Shocker 1991). Using the fast food category as an example, an attribute may be ‘for when I am in a rush’.

Benefits: Tangible and intangible benefits that the brand/product provides to the

consumer (Hirschman 1980).

Functional Benefits – Intrinsic benefits that provide a solution to a consumer’s

problem or need (Keller 1993). Attributes highlight the core purpose or function of the product and how this benefits consumers. I.e. toothpaste for sensitive teeth.

Experiential Benefits – Benefits associated with using and/or purchasing the

product (Ratneshwar et al. 2001). For instance, the taste of food or beverage products.

Symbolic Benefits – Extrinsic advantages of the product. Desires that the

product aims to fulfil. For example, social acceptance associated with keeping up with the latest trends. Attributes represent perceptions about what the brand is doing or providing for the user (Steenkamp & Van Trijp 1997). Symbolic attributes link the consumer with a group, role or image and can include personality traits (Aaker 1997), for example, a fun brand.

Brand attitudes: Consumers’ overall evaluations of the brand, which encompass brand

attributes and benefits (Farquhar 1990).

Limiting attributes to one or two types could restrict light buyer response and the ability to perform further analysis. Therefore, this research includes a mixture of Keller’s (1993) three types of attributes (see Table 44). Attribute type and wording are in line with those used in previous studies, particularly for breakfast cereal (i.e., Barnard & Ehrenberg 1990).

Page 96: Understanding and Measuring Light Buyer Brand Equity

87

Table 44: Attributes selected for inclusion in brand image questionnaire.

# Breakfast Cereal Butter/ Margarine Type

1 Stays crispy in milk Spreads easily Product

2 Natural Natural Product

3 Good value for money Good value for money Price

4 Good for the whole family Good for the whole family User

5 Kids would like it Kids would like it User

6 Good for a treat Good for cooking / baking Usage

7 Would give me energy for the day Helps control cholesterol Functional

8 A healthy option A healthy option Functional

9 Would taste great Would taste great Experiential

10 Innovative Innovative Symbolic

11 A modern brand A modern brand Symbolic

12 A brand I feel positively about A brand I feel positively about Attitude

13 Better quality than other brands Better quality than other brands Attitude

14 Appeals to you more than other brands

Appeals to you more than other brands Attitude

15 Offers something other brands do not

Offers something other brands do not Attitude

8.3.3 Brand Buyer Classification Claimed usage variables were included in the questionnaire design to differentiate

between different types of brand buyers and non-buyers. Following the brand image question for each product category, respondents were given a list of nine brands and asked to identify brands they had bought in the past 12 months. Brands were randomised for each respondent to avoid order effects.

Brands selected were presented to respondents in a follow up question, where they were asked to select from a drop down menu of 1 to 10+ times, the number of times they had bought each of the brands selected in the past 12 months. Brands were presented in the same order as the prior question.

8.4 Operationalisation of Brand Image

Brand image was measured according to four conditions. The conditions are a

combination of response type (free-pick any/ forced- forced choice binary) and prompting method (attribute/ brand). Figure 16 to Figure 19 detail question wording and response format for each condition (see Appendix A for the full questionnaire).

Page 97: Understanding and Measuring Light Buyer Brand Equity

88

Figure 16: Pick any brand prompted image question.

Figure 17: Pick any attribute prompted image question.

Page 98: Understanding and Measuring Light Buyer Brand Equity

89

Figure 18: Forced choice binary brand prompted image question.

Page 99: Understanding and Measuring Light Buyer Brand Equity

90

Figure 19: Forced choice binary attribute prompted image question.

8.5 Sample Size

A split sample was used during data collection, where each respondent was randomly

placed into one of four groups. Each group completed a pick any and a forced choice binary question, prompted with brand and attribute cues for different product categories (see Table 45). The use of different groups allowed comparison of image measures across different categories and between methods for the same category.

Table 45: Respondent allocation to image methods.

Group Breakfast Cereal Butter/Margarine

Measure Prompting Method Measure Prompting

Method A Pick Any Attribute FC Binary Brand B FC Binary Attribute Pick Any Brand C Pick Any Brand FC Binary Attribute D FC Binary Brand Pick Any Attribute

The use of different groups also provided a reasonable measure for determining an appropriate sample size for the study, both in terms of allocated time and resources and for data analysis. This study had a sample size of over 2,000 respondents, evenly divided between the four sample groups. Each group contained approximately 500 respondents, allowing comparison between 1,000 respondents for each response type and prompting method. To avoid order effects, both PA and FCB measures were presented first and second, along with respondents being prompted with an attribute

Page 100: Understanding and Measuring Light Buyer Brand Equity

91

and brand cue first and second. For the four conditions, questions for breakfast cereal were presented first, followed by questions for butter/margarine.

8.6 Screening Respondents

To be eligible to participate in the study, respondents:

1. Must have purchased both breakfast cereal and butter/margarine in the last 12

months.

This research aims to analyse consumer response for different brand buyer groups.

Therefore respondents must be buyers of the category to participate in the survey. A 12-month timeframe is consistent with classification in consumer packaged goods studies (e.g., Anschuetz 2002).

2. Must by over 18 years of age

In accordance with ethics approval and conduct, respondents must be able to give consent themselves. In addition, this research is interested in those who make purchases from the categories. In most instances, those under 18 years of age live with a parent or guardian and do not partake in grocery shopping frequently.

8.7 Data Collection and Demographics

An external research panel provider, Cross Tabs, was used to recruit participants for data collection across the United Kingdom. Regions were selected to match those captured in TNS Kantar Panel data. The researcher designed and programmed the survey using the research software, Qualtrics, and provided a survey link to the panel provider for collection, along with specification for sample demographics.

Quotas were set for demographic variables across the four groups to ensure a comparable sample. The purpose of this thesis is not to provide generalisable findings to the population. Instead, the focus is on comparing results across techniques to provide insight and clarity on the topic for use in further research. Each group had an even split across gender variables.

Page 101: Understanding and Measuring Light Buyer Brand Equity

92

Age and location categories were selected based on those used in TNS Kantar panel data. As such, quotas for the two variables were determined by looking at category buyer distributions in the panel data. Table 46 shows the distribution of respondents by age and gender for the four groups. Table 47 shows the distribution of respondents by location. Appendix B provides justification of image and education variables selected and details the distribution of respondents across the variables.

Table 46: Distribution of respondents across image methods by age and gender.

% Gender % Age M F 18-28 29-34 35-44 45-54 55+

Group A 45 55 12 12 21 21 34 Group B 46 54 12 12 20 21 35 Group C 45 55 13 12 21 21 33 Group D 46 54 11 12 19 22 36 Average 46 55 12 12 20 21 35

Table 47: Distribution of respondents across image methods by location.

% Location Group A Group B Group C Group D Average London 13 11 12 14 13 Midlands 16 19 18 16 17 North East 7 5 4 5 5 Yorkshire & The Humber 8 10 8 11 9 Lancashire 11 11 12 13 12 South 15 16 15 14 15 Scotland 12 6 8 9 9 East England 9 11 11 9 10 Wales & West England 9 12 12 8 10

8.8 Operationalisation of Light Brand Buyer

Consistent with awareness analysis, for image analysis, light brand buyer classification draws from Anschuetz’s (2002) study. As primary data was collected for this section, light buyer classification was via claimed usage. Awareness analysis using claimed

usage classified light brand buyers according to one purchase within a 12-month timeframe. This classification was possible given the large sample size available for the categories tested. For image, however, given the smaller sample size, issues of light buyer over-reporting (Ludwichowska 2013) limits the number of participants in the once-only group, restricting further analysis. Appendix C shows the purchase frequency distribution for each product category, demonstrating light buyer over-reporting, inflating the proportion of buyers who purchase the brand two or more times. Light brand buyer classification was therefore adjusted to include those who purchased the brand twice in the timeframe.

Page 102: Understanding and Measuring Light Buyer Brand Equity

93

8.9 Data Analysis

Data was collected for 2,113 respondents. Before analysis, data was cleaned to remove

all respondents who gave no response. No response was defined as selecting ‘no’, ‘not sure’ or ‘no brands’ for all questions. A total of 100 (5%) respondents were removed from the data set, resulting in 2,013 useable responses. Of those who were removed, 70% completed the survey in less than seven minutes, indicating that respondents may not have spent time thinking about each question, instead trying to complete the survey as quickly as possible.

In a similar format to awareness analysis, respondents were first categorised as light brand buyers using brand purchase frequency. The proportion of light brand buyer response was identified using cross-tabulations of buyer purchase frequency and brand/attributes selected for each image measurement approach. Principles of data reduction were used, as outlined in Section 5.7.

Finally, inferential statistic independent samples t-tests were used to test for statistical significance at the 0.1 and 0.05 level. Analysis was performed at the attribute and brand level. See appendices D through H for details of analysis results.

8.10 Chapter Summary

This chapter has discussed the method used for the study, including:

• Information on the data sets analysed.

• Justification for light buyer classification and the analysis timeframe selected.

• Operationalisation of image measures.

• Analysis techniques, including the presentation of results.

The following chapter will present the results for brand image and provide discussion of key findings.

Page 103: Understanding and Measuring Light Buyer Brand Equity

94

Chapter 9 IMAGE RESULTS & DISCUSSION Results from image analysis are presented, answering the hypotheses posed by this study. Discussion of the findings follows.

9.1 Image Results

Results are presented according to the order of hypotheses in Chapter Seven. A total of

eight and four brands are analysed for cereal and butter/margarine categories, respectively. Remaining brands are excluded from analysis due to the small sample size of light brand buyers.

Four image measurement groups are tested: pick any (PA) attribute prompted, PA brand prompted, forced choice binary (FCB) attribute prompted, and FCB brand prompted. For each image measurement group, the proportion of light brand buyer response is determined using cross-tabulations of buyer purchase frequency (1-2 purchases) and brands/attributes selected.

9.1.1 PA: Attribute versus Brand Prompting H5: Light brand buyer response to PA will be higher when attribute

prompted than when brand prompted.

To address hypothesis 5, light buyer response to the PA measure is compared when attribute and brand prompted. The hypothesis suggests that light buyer response to PA will be higher when attribute prompted than when brand prompted. Results are presented according to each attribute, detailed in Appendix D.

For cereal, 53 out of 120 cases (15 attributes x 8 brands) (44%) have light buyer PA response statistically significantly higher (at p<0.1) when attribute prompted than when brand prompted. In 37 cases, results are statistically significant at p<0.05. In addition, in 44 cases, results trend in favour of attribute prompting, where results for 33 of these cases show a five or higher percentage point difference between attribute and brand prompted response. Thus, in 97 out of 120 cases (81%), results show statistically significant differences or trend in favour of PA attribute prompted.

Page 104: Understanding and Measuring Light Buyer Brand Equity

95

When brand prompted, light buyer response to the attribute ‘good for a treat’ for Brand E and ‘a healthy option’ for Brand F is statistically significantly higher than when attribute prompted (see Tables 48 and 49). Seventeen cases also show results trending in favour of brand prompting, and eight of these cases having a 5+ percentage point difference between attribute and brand prompted response.

Table 48: Proportion of light brand buyer cereal PA response – Good for a treat.

Good for a treat %Ave Pen12

PA Attribute

PA Brand

Brand A 15 9 10 Brand B 15 22 26 Brand C 14 53 51 Brand D 13 26 26 Brand E 13 17 31* Brand F 13 23 26 Brand G 11 69 63 Brand H 11 54 50 Average 34 36

Statistically significantly higher than PA Attribute at *p<0.1.

Table 49: Proportion of light brand buyer cereal PA response – A healthy option.

A healthy option %Ave Pen

PA Attribute

PA Brand

Brand A 15 77 71 Brand B 15 53 47 Brand C 14 17 22 Brand D 13 41 36 Brand E 13 58 60 Brand F 13 20 39** Brand G 11 10 10 Brand H 11 4 10 Average 35 37

Statistically significantly higher than PA Attribute at **p<0.05.

For butter/margarine, 24 out of 60 cases (15 attributes x 4 brands) (40%) show statistically significant differences in light buyer response for PA, where response when attribute prompted is higher than when brand prompted. As with cereal, most cases (17 out of 24) are statistically significant at p<0.05. A further 26 cases trend in favour of attribute prompting, with 19 cases showing a five or higher percentage point difference in response.

12 % Average light brand buyer penetration across the four survey groups.

Page 105: Understanding and Measuring Light Buyer Brand Equity

96

In one instance, for the attribute ‘a healthy option’, Brand F PA brand prompted response is statistically significantly higher than PA attribute prompted response (refer Table 50). Seven additional cases show results trending in favour of brand prompting.

Table 50: Proportion of light brand buyer butter/margarine PA response – A healthy option.

A healthy option %Ave Pen

PA Attribute

PA Brand

Brand B13 11 30 36 Brand C 10 16 22 Brand D 10 20 13 Brand F 10 16 40** Average 20 28

Statistically significantly higher than PA Attribute at **p<0.05.

Overall, results for the two product categories find in 77 out of 180 cases (43%) light

buyer response to PA is statistically significantly higher when attribute prompted than when brand prompted. A further 70 cases show results trending in favour of attribute prompting, with a total of 147 out of 180 cases (82%) with statistical differences or results trending in favour of the PA measure when attribute prompted.

In comparison, three out of 180 cases (2%) find brand prompted results statistically significantly higher than attribute prompted. A total of only 27 out of 180 cases (15%) show statistically significant differences or results trending in favour of brand prompting.

The weight of evidence therefore supports hypothesis 5 that light buyer response to the PA measure is higher when attribute prompted than when brand prompted.

Type of attribute

It is interesting to note different types of attributes where many brands have statistically significantly results or, alternatively, very little difference in results across the two prompting methods.

For cereal and butter/margarine, at least half of the brands tested find PA results when attribute prompted statistically significantly higher than when brand prompted for the symbolic attributes ‘innovative‘ and ‘a modern brand’. Statistical differences also occur for over half of the brands tested for three out of four attitude attributes: ‘a brand I feel positively about’, ‘better quality than other brands’, and ‘appeals to you more than other

brands’.

13 Brands are ordered alphabetically in terms of overall penetration.

Page 106: Understanding and Measuring Light Buyer Brand Equity

97

For both categories, the functional attribute, ‘a healthy option’, however, shows statistically significant differences in the opposite direction, where brand prompted is higher than attribute prompting. For cereal, the usage attribute ‘good for a treat’ also shows statistical differences in favour of brand prompting.

Attributes that show little difference in light buyer response to PA attribute and brand prompted are noted. The product related attribute ‘natural’ (both categories) and usage attribute ‘good for cooking/baking’ (butter) show few brands with significant differences in light buyer response. In addition, for butter, the functional attribute, ‘helps control

cholesterol’, and price attribute, ‘good value for money’, does not show statistically significant difference in response for any brand.

Similarities across the two product categories suggest that light buyer response for symbolic and attitude attributes tend to be in favour of attribute prompting. For usage

and functional attributes, light buyers either respond in a consistent manner to both attribute and brand prompted methods or trend in favour of the brand prompted approach. However, findings for usage and functional attributes are not consistent across the two product categories tested. This research tests only a small number of each type of attribute, which makes it difficult to draw overall conclusions about light buyer response for different types of attributes. Future research is encouraged to test if findings hold for a type of attribute across multiple attributes.

9.1.2 FCB: Attribute versus Brand Prompting H6: There will be no difference in light brand buyer response to FCB

whether brand prompted or attribute prompted.

To address the next hypothesis, which states that there will be no difference in light buyer response for FCB for the two prompting methods, results are presented at attribute level for both product categories in Appendix E.

For cereal, unlike results for the PA measure, there are few statistically significant differences between brand and attribute prompting for the FCB measure. Results for 20 out of 120 cases (17%) show light buyer FCB response statistically significantly higher (at p<0.1) when brand prompted than when attribute prompted. Of the 20 cases with statistical differences, 16 are statistically significantly different at p<0.05. There are 60 cases where results trend towards brand prompting, yet only 33 of these cases show a five or higher percentage point difference in brand and attribute prompted response.

Page 107: Understanding and Measuring Light Buyer Brand Equity

98

There are no cases where light buyer response is statistically significantly higher when attribute prompted than when brand prompted. However, there are 34 cases where results trend in favour of attribute prompting. Again, few cases show large differences, with nine out of the 34 cases showing a difference in response of five or more percentage points. Results for cereal suggest that light buyer response to FCB is similar whether brand or attribute prompted.

For butter/margarine, 15 out of 60 cases (25%) find light buyer response to FCB when brand prompted statistically significantly higher than when attribute prompted. Thirteen of the 15 cases are statistically significant at p<0.05. There are an additional 36 cases where results trend in favour of brand prompting. In contrast, there are seven cases where results trend in favour of attribute prompting, with only four cases showing a five or higher percentage point difference in response.

There is one case for butter/margarine where there are conflicting results. For the attribute, ‘spreads easily’ (see Table 51), light buyer response for Brand F is statistically

significantly higher for brand prompted, while for Brand B, attribute prompted results are statistically significantly higher than brand prompted. One explanation may be that when attribute prompted, respondents cannot differentiate between the two types of butter/margarine products, linking both to the attribute. When presented with the brand name individually, the respondent considers each brand separately, where Brand F is

more strongly linked to the attribute ‘spreads easily’, consistent with brand advertising.

Table 51: Proportion of light brand buyer butter/margarine FCB response – Spreads easily.

Spreads easily %Ave Pen

FCB Brand

FCB Attribute

Brand B 11 85 95** Brand C 10 91 94 Brand D 10 53 39 Brand F 10 98* 91 Average 82 80

*Statistically significantly higher than FCB Attribute at p<0.1. **Statistically significantly higher than FCB Brand at p<0.1.

Overall, for both product categories, only 35 out of 180 cases (19%) show light buyer FCB response statistically significantly higher for brand prompted than when attribute prompted. In comparison, one case finds response for attribute prompted statistically significantly higher than brand prompted. However, in 96 cases, results trend in favour of brand prompted, with 63 out of the 96 cases having a five or higher percentage point

Page 108: Understanding and Measuring Light Buyer Brand Equity

99

difference in brand and attribute prompted response. Therefore, 131 out of 180 cases show statistical differences or results trending in favour of brand prompting.

Given that under 20% of total light buyer response is statistically significantly different, hypothesis 6 is partially supported, where there is little statistical difference in FCB results, but with a clear trend towards brand prompted.

Type of attribute

Provided that there are few statistical differences for each category, there are a number of attributes where all brands tested show no statistically significantly different results. However, only the attitude attribute, ‘appeals to you more than other brands’, is common to cereal and butter with no statistical difference in brand and attribute prompted response.

For the FCB measure, there are few attributes where light buyer response for a majority of brands tested shows statistically significant differences. Over half of the brands tested for each product category show statistical differences in response for the attitude attribute, ‘offers something other brands do not’, where response is higher when brand

prompted than when attribute prompted. In addition, both categories show more brands with statistical differences in response for usage attributes ‘good for a treat’ (cereal) and ‘good for cooking/baking’ (butter). Given patterns in light buyer response

are not present for all attitude and usage attributes, no conclusions can be draw about how light brand buyers respond to different types of attributes.

9.1.3 PA versus FCB Results To address hypothesis 7, where it is expected that light buyer response will be higher

for FCB than PA regardless of the prompting method used, cross-tabulations of buyer purchase frequency and image response are calculated. Results are presented for each attribute, for eight cereal brands and four butter/margarine brands.

Results for PA and FCB methods are first presented for attribute prompting, then for brand prompting. Following this, results for the two prompting methods are summarised and compared across PA and FCB approaches.

Page 109: Understanding and Measuring Light Buyer Brand Equity

100

9.1.3.1 Attribute Prompted

H7a: Light brand buyer response will be higher for FCB than for PA when

attribute prompted.

To test hypothesis 7a that light buyer response for FCB will be higher than for PA when attribute prompted, cross-tabulations of light buyer response to the two image measures are calculated, detailed in Appendix F.

For cereal, 114 out of 120 cases (95%) find light buyer response for FCB statistically significantly higher than for PA at p<0.1. In 109 cases, results are statistically significant at p<0.05. For the attribute ‘good value for money’, response for Brand E does not differ for the two image measures (refer Table 52). The remaining five cases that do not have statistically significant differences in results show results trending in favour of FCB, two of which are shown in Table 52.

Table 52: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Good value for money.

Good value for money

%Ave Pen

FCB Attribute

PA Attribute

Brand A 15 78 74 Brand B 15 81* 70 Brand C 14 65* 52 Brand D 13 77 68 Brand E 13 71 71 Brand F 13 68** 46 Brand G 11 67** 47 Brand H 11 70** 52 Average 72 60

Statistically significantly higher than PA Attribute at *p<0.1 **p<0.05.

For butter/margarine, 46 out of 60 cases (77%) show light buyer FCB response is

statistically significantly higher (at p<0.1) than PA response when attribute prompted. As with cereal, almost all brands (41 out of 46) have results statistically significant at p<0.05. Results for all other brands/attributes show results trending in favour of FCB.

There are minor variations in results across attributes. For example, there is no statistically significant difference between FCB and PA light buyer response for Brand D

butter for three out of four attitude attributes, i.e. ‘better quality than other brands’, ‘appeals to you more than other brands’, and ‘offers something other brands do not’. Results for the three attitude attributes, however, trend in favour of FCB.

Page 110: Understanding and Measuring Light Buyer Brand Equity

101

A second noticeable difference in responses is for the attribute ‘helps control cholesterol’ (see Table 53). Here, only Brand C shows statistical differences in light buyer response. Results across all brands, however, trend in the same direction, with FCB response higher than PA response.

Table 53: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Helps control cholesterol.

Helps control cholesterol

%Ave Pen

FCB Attribute

PA Attribute

Brand B 11 40 28 Brand C 10 31** 12 Brand D 10 15 13 Brand F 10 29 18 Average 29 18

Statistically significantly higher than PA Attribute at **p<0.05.

For both product categories, at p<0.05, 150 out of 180 cases (83%) find FCB response statistically significantly higher than PA response when attribute prompted. At p<0.1, 160 out of 180 cases (89%) find statistically significant results, while the remaining brands show results trending in favour of FCB. Therefore, as 179 out of 180 cases have results statistically significantly different or trending in favour of FCB, hypothesis 7a is supported. Light buyer response is higher for FCB than PA when attribute prompted.

Type of attribute

Examining results for attribute type show fewer statistically significant results for two attributes – ‘good value for money’ for cereal and ‘helps control cholesterol’ for butter/margarine. The two attributes are a non-product related price attribute and functional benefit attribute, respectively. There are no systematic differences in light buyer response across product categories for a type(s) of attribute. Light buyer FCB response is higher than PA response when attribute prompted for all attribute types.

9.1.3.2 Brand Prompted

H7b: Light brand buyer response will be higher for FCB than for PA when

brand prompted.

To address hypothesis 7b, which compares light buyer response for FCB and PA measures when brand prompted, cross-tabulations of light buyer purchase frequency (two purchases in a 12-month timeframe) and response for the two image measures is calculated. Results are presented at attribute level in Appendix G for both product categories.

Page 111: Understanding and Measuring Light Buyer Brand Equity

102

Results are consistent across brands/attributes for both categories, with response to the FCB measure higher than the PA measure. Light buyer response for the attribute ‘good value for money’ is provided as an example of cereal and butter/margarine results in Tables 54 and 55, respectively.

Table 54: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Good value for money.

Good value for money

%Ave Pen

FCB Brand PA Brand

Brand A 15 76** 46 Brand B 15 80** 44 Brand C 14 67** 17 Brand D 13 82** 50 Brand E 13 72** 28 Brand F 13 77** 31 Brand G 11 63** 21 Brand H 11 62** 32 Average 72 34

Statistically significantly higher than PA Brand at **p<0.05.

Table 55: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Good value for money.

Good value for money

%Ave Pen

FCB Brand PA Brand

Brand B 11 79** 34 Brand C 10 77** 48 Brand D 10 72** 30 Brand F 10 75** 23 Average 76 34

Statistically significantly higher than PA Brand at **p<0.05.

For cereal, all brands for all attributes have statistically significant differences at p<0.05, where light buyer response to FCB is higher than response to PA when brand prompted. For butter/margarine, 59 out of 60 cases show FCB response statistically

significantly higher than PA response. Though this difference is not statistically significant, results for Brand C for the attribute ‘spreads easily’ trend in favour of FCB (91%) than PA (80%).

Overall, 179 out of 180 cases (99%) find statistical differences in results at p<0.05. Hypothesis 7b that light buyer response is higher for FCB than for PA when brand prompted is supported. Findings are consistent across all attribute types.

Page 112: Understanding and Measuring Light Buyer Brand Equity

103

9.1.4 Summary To summarise results for light brand buyer response, the ratio between FCB and PA

response is calculated and compared for attribute and brand prompted methods. Results for cereal and butter/margarine are shown in Table 56 and 57. Results are presented as the average light buyer response across all attributes for each brand.

Table 56: Proportion of light brand buyer FCB and PA image response, attribute and brand prompted- Cereal.

Cereal %FCB Attribute

%PA Attribute Ratio %FCB

Brand %PA Brand Ratio

Brand A 69 47 1.5 73 32 2.3 Brand B 68 45 1.5 73 33 2.2 Brand C 72 47 1.5 77 34 2.3 Brand D 66 42 1.6 75 30 2.4 Brand E 70 46 1.5 78 36 2.1 Brand F 72 35 2.1 79 33 2.3 Brand G 64 40 1.6 67 30 2.2 Brand H 69 39 1.8 68 32 2.1 Average 69 43 1.6 74 33 2.3

Table 57: Proportion of light brand buyer FCB and PA image response, attribute and brand prompted- Butter/Margarine.

Butter/Margarine %FCB Attribute

%PA Attribute Ratio %FCB

Brand %PA Brand Ratio

Brand B 65 36 1.8 70 26 2.7 Brand C 59 34 1.7 67 28 2.4 Brand D 60 44 1.4 70 30 2.3 Brand F 62 41 1.5 79 25 3.1 Average 61 39 1.6 71 27 2.6

When attribute prompted, light buyer response for FCB is 1.6 times higher than for PA for cereal and butter/margarine categories. In comparison, when brand prompted, response for FCB is 2.3 and 2.6 times higher than for PA across the two categories. The larger ratio for brand prompted measures can be explained by a higher response for FCB and lower response for PA when compared to attribute prompted results.

Results indicate that the FCB measure when brand prompted captures the highest proportion of light brand buyer response, while light brand buyer response is lowest for the PA measure when brand prompted.

Page 113: Understanding and Measuring Light Buyer Brand Equity

104

To address the second criteria for what makes a good image measure for light brand buyers (as detailed in Chapter 7), the average number of attributes selected per brand and average number of brands selected per attribute is examined. Analysis will provide an indication of whether results reflect attributes linked to the brand in memory or whether attributes are typical of the category and/or there is a response bias present where light buyers are responding out of agreement (as noted in Joyce 1963).

9.1.4.1 Average Number of Attributes

As an extension to hypothesis 7, which found that light brand buyer response to FCB is statistically significantly higher compared to the PA approach, the average number of attributes selected per brand is analysed. For each image approach, respondents could select a maximum of 15 attributes per brand cue. Tables 58 to 61 detail results for both product categories, presented according to prompting method.

Table 58: Average number of attributes selected per brand, FCB and PA attribute prompted - Cereal.

Cereal Ave Number of Attributes (n=15)

FCB Attribute

PA Attribute Ratio

Brand A 10 7 1.5 Brand B 10 7 1.5 Brand C 11 7 1.5 Brand D 10 6 1.6 Brand E 10 7 1.5 Brand F 11 5 2.1 Brand G 10 6 1.6 Brand H 10 6 1.8 Average 10 6 1.6

Table 59: Average number of attributes selected per brand, FCB and PA brand prompted - Cereal.

Cereal Ave Number of Attributes (n=15)

FCB Brand PA Brand Ratio

Brand A 11 5 2.3 Brand B 11 5 2.2 Brand C 12 5 2.2 Brand D 11 5 2.4 Brand E 12 5 2.1 Brand F 12 5 2.3 Brand G 10 5 2.2 Brand H 10 5 2.1 Average 11 5 2.2

Page 114: Understanding and Measuring Light Buyer Brand Equity

105

Table 60: Average number of attributes selected per brand, FCB and PA attribute prompted – Butter/margarine.

Butter/Margarine Ave Number of Attributes (n=15)

FCB Attribute

PA Attribute Ratio

Brand B 10 5 1.8 Brand C 9 5 1.7 Brand D 9 7 1.4 Brand F 9 6 1.5 Average 9 6 1.6

Table 61: Average number of attributes selected per brand, FCB and PA brand prompted – Butter/margarine.

Butter/Margarine Ave Number of Attributes (n=15)

FCB Brand PA Brand Ratio

Brand B 11 4 2.7 Brand C 10 4 2.4 Brand D 10 4 2.3 Brand F 12 4 3.1 Average 11 4 2.6

A trend appears in the data, where across all brands, the number of attributes selected per brand for FCB is higher than for PA. Of a possible 15 associations, for FCB, light brand buyers select an average of 10 and 11 attributes for each brand when attribute and brand prompted, respectively. For PA, light brand buyers select an average of six and five attributes for each brand when attribute and brand prompted. Overall, light brand buyers select over 65% of all attributes for each brand in the FCB measure, compared to approximately 35% of all attributes for each brand in the PA measure.

9.1.4.2 Average Number of Brands

In addition to the average number of attributes selected per brand, the average number of brands linked to each attribute is calculated. Respondents could select a maximum of nine brands for each attribute (see Table 62 and 63).

It is important to note that results contain brands excluded from earlier analyses. This may result in a larger difference between prompting methods for each measure than described in previous results. The overall trend in light buyer response, however, remains the same. Results for PA trend to attribute prompted, while results for FCB trend to brand prompted. A ratio column is included to help identify differences between the two measures and across the two methods, which may be lost due to rounding of figures during data reduction.

Page 115: Understanding and Measuring Light Buyer Brand Equity

106

Table 62: Average number of brands selected per attribute for light buyers - Cereal.

Cereal

Average Number of Brands (n=9) FCB

Attribute (n=267)

PA Attribute (n=242)

Ratio FCB

Brand (n=248)

PA Brand

(n=250) Ratio

Stays crispy in milk 4 2 1.9 5 2 2.4 Natural 5 2 2.1 6 2 2.3 Good value for money 5 4 1.4 6 2 2.4 Good for the whole family 6 4 1.6 7 4 1.8 Kids would like it 7 4 1.5 7 4 1.6 Good for a treat 5 2 2.0 6 3 2.2 Would give me energy for the day 5 3 1.9 6 2 2.5 A healthy Option 4 2 1.9 5 3 1.9 Would taste great 6 4 1.5 7 4 1.9 Innovative 4 2 2.4 5 1 5.2 A modern brand 5 3 1.9 5 2 2.7 A brand I feel positively about 6 3 1.7 6 3 2.1 Better quality than other brands 6 3 2.0 6 2 3.0 Appeals to you more 4 2 1.7 5 2 2.9 Offers something other brands do not 4 2 2.2 5 1 3.9

Average 5 3 1.8 6 2 2.6

Table 63: Average number of brands selected per attribute for light buyers - Butter/Margarine.

Butter/Margarine

Average Number of Brands (n=9) FCB

Attribute (n=216)

PA Attribute (n=211)

Ratio FCB

Brand (n=208)

PA Brand

(n=185) Ratio

Spreads easily 7 5 1.6 7 4 1.8 Natural 4 2 2.4 5 1 3.5 Good value for money 5 2 2.5 6 2 2.6 Good for the whole family 6 3 1.7 6 2 2.6 Kids would like it 6 3 2.0 6 1 4.2 Good for cooking / baking 4 3 1.8 6 2 2.6 Helps control cholesterol 2 1 1.9 4 1 3.5 A healthy Option 3 2 1.9 4 2 2.2 Would taste great 5 3 1.6 6 3 2.2 Innovative 4 2 2.3 5 1 4.9 A modern brand 5 3 1.9 6 2 2.9 A brand I feel positively about 5 3 1.8 6 2 2.9 Better quality than other brands 4 2 1.9 5 2 3.1 Appeals to you more 4 2 1.9 4 1 3.8 Offers something other brands do not 3 1 2.4 4 1 5.8

Average 5 2 1.9 5 2 2.8

Page 116: Understanding and Measuring Light Buyer Brand Equity

107

For FCB, light buyers link, on average, 55% (five out of nine brands) of all brands

presented to each attribute. In comparison, for PA, light buyers link only 22% (two out of nine brands) of all brands to each attribute. The number of brands selected for each attribute is consistently higher for FCB than PA for both attribute and brand prompted methods.

Finally, FCB and PA measures are examined for their ability to distinguish between light and non/heavier buyer response.

9.1.4.3 Distinguishing between light and non/heavier buyer response

Given light buyers have less exposure to the brand than heavier buyers, the amount of

information stored in memory and accessible for retrieval is less than those who purchase the brand more frequently (Nedungadi 1990). An image measure that can differentiate between light and non/heavier buyer response is beneficial to marketers, where results can be used to assess different aspects of the brand. For example, they can assess how well messages contained in advertising efforts are being captured and remembered by non and lighter buyers, along with associations developed from more frequent purchase/consumption, i.e. taste.

If there is little difference between light and heavier buyer response levels for a measure, there is little to no room for light buyer response to increase over time. This is potentially problematic for the FCB measure, where light buyer response is already, on average, 70 per cent. Thus, further testing is employed to compare FCB and PA ability to discriminate between response for non-buyers, light and heavier buyer groups.

To assess response levels across the three buyer groups, cross-tabulations of buyer purchase frequency and response to each image measure is calculated for the four image measurement conditions. Light brand buyers are again defined as those who purchase the brand once or twice, with heavier buyers classified as purchasing three or more times and non-buyers not at all in the timeframe. Statistical testing is performed for light buyer response in comparison to non and to heavier buyer response. Results are presented for each brand in Appendix H.

Across the four image measures, response levels increase from non-buyer to light to heavier buyers, in line with respondent purchase frequency. Table 64 provides an example from the cereal category of non, light and heavier buyer average response across all attributes for each brand for the PA measure, attribute prompted.

Page 117: Understanding and Measuring Light Buyer Brand Equity

108

Table 64: Average proportion of non, light and heavier buyer cereal response, PA attribute prompted.

Cereal % Non-Buyer

% Light Buyer

% Heavier Buyer

Brand A 27 47 59 Brand B 21 45 54 Brand C 23 47 55 Brand D 19 42 49 Brand E 19 46 51 Brand F 17 35 49 Brand G 17 40 41 Brand H 18 39 47 Average 20 43 51

Light versus Non-Buyer Response

For cereal, there are statistical differences between light and non-buyer response for over 70% of brands (at p<0.1) for all four image measurement conditions. For FCB, when attribute prompted 108 out of 120 cases (90%) show light buyer response is higher than non-buyer response. The number of cases decreases for FCB attribute prompted to 106 out of 120 cases (88%), 104 cases (87%) for PA attribute prompted and finally, 88 cases (73%) for PA brand prompted.

Similarly, for butter/margarine, there are statistically significant differences across a majority of brands and attributes, where light buyer response is higher than non-buyer response. The number of cases with statistical differences is greatest for PA attribute prompted (58 out of 60 – 97%), decreasing to FCB brand prompted (48 out of 60 – 80%), FCB attribute prompted (47 out of 60 – 78%), and finally to PA brand prompted (40 out of 60 – 67%).

The lower incidence of statistical differences for PA brand prompted can be explained by the lower response levels for non and light buyers, which makes it difficult to distinguish between responses for the two buyer groups. Regardless, results highlight that the four image measures are equally able to differentiate between response for non and light buyers.

Page 118: Understanding and Measuring Light Buyer Brand Equity

109

Light versus Heavier Buyer Response

Next, for each brand, statistical testing is used to compare light and heavier buyer responses for each attribute.

For cereal, two brands are removed from the results, Brand H and Brand G (however

are still provided in Appendix H). The two brands have lower sample sizes for light and heavier buyer groups than other brands tested. In addition, results for the two brands show no statistical differences for PA attribute or brand prompted and FCB attribute prompted. However, for FCB brand prompted, there are seven and 10 cases where heavier buyer response is higher than light buyer response for Brand H and Brand G, respectively. FCB brand prompted captures the highest proportion of light buyer response, which may explain significant differences found for the measure, given the small sample size for each buyer group.

For the remaining six cereal brands, for both PA measures, 68 out of 90 cases (76%) show response is statistically significantly higher or trending with a five or more percentage point difference, to heavier buyer response. In comparison, FCB brand prompted show results statistically significantly different or trending to heavier buyer response for 61 out of 90 cases (68%), while only 51 cases (57%) with differences between light and heavier buyers when attribute prompted.

For butter/margarine, as with the cereal results, there are more cases with statistical differences in light and heavier buyer response for PA than for FCB. When attribute prompted, for PA, 46 out of 60 cases (77%) have results statistically significantly different or trending with a five or higher percentage point difference in response to heavier buyers. For FCB, the number of cases decreases to 43 out of 60 cases (72%) where heavier buyers response is statistically significantly higher than light buyer response. When brand prompted, similar patterns emerge again, with results statistically significantly higher or trending in favour of heavier buyer response for 41 cases (68%) for PA and 39 cases (65%) for FCB.

Overall, for the two product categories, the PA attribute prompted measure has the largest number of cases (76%) with statistical difference or a five or higher percentage point difference in response for light and heavier buyers. In contrast, FCB attribute prompted is the measure that shows the least differentiation between light and heavier buyer response (63% stat. sig. difference or 5+ percentage point difference trending).

Page 119: Understanding and Measuring Light Buyer Brand Equity

110

9.2 Image Discussion

Two image measures have been analysed from a light buyer perspective using two

methods of prompting for response. Discussion of the findings is presented according to the two criteria for image measures to capture light buyer response, as detailed in Chapter Seven. FCB and PA measure ability to distinguish between light and other brand buyer response is then discussed, followed by a summary and recommendations for use in CBBE surveys.

9.2.1 Proportion of Light Buyer Response FCB captures a higher proportion of light buyer responses than PA across both categories. Results are in line with Dolnicar, Rossiter and Grun (2012), who find a 35 percentage point difference in responses for FCB (M=71%) and PA (M=36%). The present study expands on Dolnicar, Rossiter and Grun’s (2012) research, which analysed the two measures using a brand prompted approach for all buyers and non-buyers. Here, the same patterns are observed for light brand buyers, where response to FCB is higher than PA, for brand and attribute prompting methods. For the FCB measure, the proportion of light buyer response trends in favour of the brand prompted method. However, only 35 out of 180 cases showed statistically significant differences in results. Similarly, Joyce (1963) notes that response does not differ to a great extent for attribute (M=87%) and brand (M=85%) prompted FCB methods when analysed for all buyers and non-buyers.

Despite lower light buyer response levels, PA results from the present study are in line with information retrieval from light buyer memory structures. Retrieval is cue dependent, where the cue given to light buyers determines the ability of information retrieval from memory networks (Collins & Loftus 1975). Prompting with an attribute cue provides multiple points in memory for activation and retrieval (Ratneshwar et al. 2001). Each attribute activates different parts of consumer memory, evoking a different number and composition of brands in each instance (Holden & Lutz 1992). This is reflected in a higher number of light buyer associations captured using an attribute prompted rather than brand prompted approach - 147 out of 180 cases with statistical differences or trending in favour of the PA measure when attribute prompted.

Page 120: Understanding and Measuring Light Buyer Brand Equity

111

9.2.2 Average Number of Brands/Attributes Selected Light buyers have less knowledge of the brand in memory from infrequent purchase

(Anderson 1983; Nedungadi & Hutchinson 1985). Thus, it is expected that linkages from brand to attribute are weak and difficult to retrieve (Collins & Loftus 1975). Yet, results for FCB show that light brand buyers select over 65% (10 out of 15) of all attributes for each brand and 55% (five out of nine) of all brands for each attribute. In contrast, for PA, light buyers select only around 33% (five out of 15) of all attributes for each brand and 22% (two out of nine) of all brands for each attribute.

The higher number of brands (and attributes) selected for each attribute (and brand) could suggest that FCB measures are better able to retrieve information from light buyer memory structures compared to PA measures. However, given the high number of attributes and brands selected for FCB than PA measures, this could also indicate that a response bias is present. Light buyers, who have little knowledge of the brand, may be selecting attributes out of ‘politeness’, resulting in an acquiesce bias in the FCB

measure (as suggested by Joyce 1963). Alternatively, when forced to select a response for a brand they know little about, light buyers may also be selecting attributes that are linked to the category cue in memory, but not necessarily the brand cue itself (Barsalou 1983).

Both explanations for the higher FCB response suggest that light buyers may be selecting responses that are not reflective of brand to attribute linkages held in memory structures. In addition, if light buyers select almost all attributes and brands, researcher ability to distinguish between attributes for a specific brand is limited. Therefore, the PA measure, despite capturing a lower proportion of light buyer response, provides a better representation of brand associations in light buyer memory.

9.2.3 Distinguishing between Light and Non/Heavier Buyer Response Overall, results find that the four image measures are equally able to differentiate between response for non and light buyers. For each measure, in over 80 per cent of cases, light buyer response is statistically significantly higher than non-buyer response.

Page 121: Understanding and Measuring Light Buyer Brand Equity

112

Examining the response levels of different brand buyers, findings indicate that the PA attribute prompted measure is better able to distinguish between light and heavier buyer responses across the two categories tested. In almost 80 per cent of cases, there were statistically significant differences or large differences with results trending in favour of heavier buyer response. The ability to distinguish between response for light and heavier buyers will allow marketers to assess the different associations held in memory for the two types of brand buyers. Furthermore, a larger difference between light and heavier buyer response indicates that there is room for light buyer response to potentially increase over time, in line with changes in brand equity. Further research is encouraged to test light buyer image response for the PA measure over time.

9.2.4 Summary The PA attribute prompted measure captures responses from approximately 40 per

cent of light buyers. However, the measure allows light buyers to indicate only what is evoked from the cue, capturing attributes linked to the brand rather than the category in memory. Therefore, results provide a more realistic indication of the associations light buyers consider in purchase/consumption situations. Furthermore, the PA attribute prompted measure is better able to differentiate between response for light buyers and non/heavier brand buyers, where attributes linked in memory can be assessed separately for these different types of brand buyers. Finally, significant differences between light and heavy buyer response allows the potential for light buyer response to increase over time.

In contract, FCB consistently captures a higher proportion of light buyer response than PA, regardless of the prompting method used. However, forcing responses requires that light buyers search memory structures for brand to attribute linkages that are weakly established. Light buyers select over half of the brands presented for each attribute and over half the attributes presented for each brand. Results suggest that light buyers may be selecting attributes that are typical of the category in addition to associations linked to the brand in memory (see Barsalou 1983) and/or that there is a response bias present, where light buyers feel they need to make a selection, responding out of ‘politeness’ (as detailed in Joyce 1963).

Overall, to assess the network of associations linked to the brand in light buyer memory, the PA attribute prompted measure is recommended for use in brand equity surveys.

Page 122: Understanding and Measuring Light Buyer Brand Equity

113

9.3 Chapter Summary

This chapter has analysed light buyer response to PA and FCB measures using two

methods of prompting for response. Key points from this chapter are:

• Light brand buyer response is statistically significantly higher for FCB than for PA, regardless of the prompting method used.

• For PA, light buyer response is higher for attribute prompted than for brand prompted. For FCB, light buyer response trends in favour of brand prompting.

• Light buyers associate, on average, over 65% of all attributes to each brand and 55% of all brands to each attribute for FCB. In contrast, around 35% of all attributes are linked to each brand and only 22% of all brands are linked to each attribute for PA.

• The more attributes and brands selected for FCB could suggest that light buyers are responding out of acquiesce and/or that results reflect associations linked to the category as well as the brand in memory.

• To assess associations linked to the brand in light buyer memory, the PA attribute prompted measure is recommended for use in CBBE surveys.

The follow chapter concludes this thesis.

Page 123: Understanding and Measuring Light Buyer Brand Equity

114

Chapter 10 CONCLUSION Implications for marketing theory and practice are discussed, before outlining the strengths and limitations of the study. This chapter concludes by stating avenues for

future research.

10.1 Contribution to Marketing Knowledge and Theory

The primary outcome of this research in terms of theory is the knowledge generated

about light buyer retrieval of information from memory structures. The results from this research help understand light brand buyers, the extent of their brand knowledge and how to better capture their responses in brand equity surveys. Specific contributions for awareness and image are detailed.

10.1.1 Brand Awareness This thesis determines that light buyers are better able to retrieve information from

memory networks than non-buyers in both recall and recognition awareness measures. Results find that buying the brand infrequently, such as once a year, is sufficient to create different awareness structures from non-buyers. Light buyers have stronger linkages between the brand and category in memory and therefore a higher response than non-buyers who are only exposed via marketing efforts (Hutchinson, Raman & Mantrala 1994). Findings show the imprint that using a brand makes on the brain and supports calls to control for past usage of the brand when interpreting brand equity scores (e.g., Christodoulides & De Chernatony 2010; Knox & Walker 2001; Romaniuk 2013).

For TOM awareness, where response is limited to one brand, light buyers are likely to

respond with a brand purchased more frequently and/or recently in their repertoire. Findings are in line with knowledge that buyers purchase from a repertoire of brands, where light buyers of one brand can be heavier buyers of other brands (Ehrenberg 2000a) and that brands are retrieved in line with linkage strength, as per frequency and recency effects (Anderson 1983). Light buyer response levels are therefore significantly lower than heavier buyers for TOM awareness.

Page 124: Understanding and Measuring Light Buyer Brand Equity

115

Light and heavier buyers are equally able to respond to unprompted and prompted awareness measures, where response levels for the two types of brand buyers show no significant difference. Findings indicate that light buyer ability to retrieve the brand name from memory increases when multiple responses are allowed and the brand name is provided as the activation cue for retrieval.

Light buyers are better able to respond to prompted awareness than to unprompted awareness. This thesis draws on the ANT of memory (Anderson & Bower 1979) to demonstrate that information is retrieved in a spreading activation, with retrieval in line with linkage strength (Collins & Loftus 1975). In unprompted awareness, which asks respondents to recall brands from memory networks, brands with stronger linkages are more accessible and are retrieved before brands purchased infrequently that have weaker linkages in memory (Alba & Chattopadhyay 1986; Anderson 1983). Therefore, competitor linkages to the category limit light buyer response. In prompted awareness, respondents indicate whether the brand name node is present in memory networks, allowing brands purchased frequently and infrequently to be selected. Limitations of accessibility and competitor linkages are removed, demonstrated by a higher light buyer response level to prompted awareness than to TOM and unprompted awareness.

The results arising from this thesis contribute to empirical data concerning brand awareness and buyer behaviour (e.g., Assael & Day 1968; Bird & Ehrenberg 1966; Hutchinson, Raman & Mantrala 1994; Nedungadi & Hutchinson 1985). Previous research has considered user and non-user response, however often in an advertising awareness context (e.g., Romaniuk & Wight 2009; Sharp, Beal & Romaniuk 2001). This study builds on Wight’s (2010) user and non-user brand awareness study, extending analysis to three brand buyer groups: non, light and heavier buyers. The present study and Wight’s (2010) research both examine consumer packaged goods categories, finding response levels higher for brand buyers than non-buyers and response for all buyer groups increasing from TOM to prompted awareness.

10.1.2 Brand Image As with findings for brand awareness, this research notes that purchasing the brand,

even infrequently, has a substantive impact on the storage and retrieval of information in consumers’ minds. Light buyer response is significantly greater than non-buyer response for both PA and FCB image measures and when using two different methods of prompting for response. Results highlight and confirm the need to control for past usage when utilising and analysing results from CBBE measures.

Page 125: Understanding and Measuring Light Buyer Brand Equity

116

This research provides an understanding of light buyer memory structures and associations retrieved from these structures for different image measures. Light buyers have less knowledge of the brand in memory from infrequent purchase (Nedungadi & Hutchinson 1985), and have linkages that are weakly established, making it harder for associations to be retrieved from the network (in line with recency and frequency effects, Anderson 1983). However, light brand buyers respond more easily to the FCB measure than to the PA measure. The higher response level across all brands and attributes does not coincide with knowledge of light buyer memory structures and the difficulty of retrieval from these structures. Instead, the FCB measure may capture light buyer associations of the brand, along with associations typical of the category. Attributes strongly linked to the category are readily retrieved from memory structures (Barsalou 1983). Light buyers are forced to select a response for each attribute and brand. A tendency to give polite response in FCB (as suggested by Joyce 1963) may result in light buyers selecting attributes linked to the category as well as those linked to the brand. The PA measure, in contrast, allows light buyers to indicate only what is evoked in memory from the cue provided (Holbrook, Moore & Winer 1982), capturing attributes linked only to the brand in memory networks.

Drawing from results for the PA measure, light buyer associations are more easily retrieved from memory structures when attribute prompted than when brand prompted. Findings fit with the notion that retrieval is cue dependent, where the cue given to light buyers determines the ability of information retrieval from memory networks (Collins & Loftus 1975). Prompting with an attribute cue provides multiple points in memory for activation and retrieval (Ratneshwar et al. 2001). Each attribute activates different parts of consumer memory, evoking a different number and composition of brands in each instance (Holden & Lutz 1992). Therefore, there is a higher probability of light buyers selecting more attributes than when a single node is activated for information retrieval from memory, i.e. when brand prompted.

The results from this thesis add to comparative image measurement studies in marketing literature (e.g., Barnard & Ehrenberg 1990; Dolnicar, Rossiter & Grun 2012; Driesener & Romaniuk 2006; Joyce 1963). This research, which analyses breakfast cereal and butter/margarine, contributes to results for the two established image measures, which also test consumer packaged goods categories (e.g., Barnard & Ehrenberg 1990; Bird, Channon & Ehrenberg 1970; Dolnicar, Rossiter & Grun 2012).

Page 126: Understanding and Measuring Light Buyer Brand Equity

117

10.2 Contribution to Marketing Practice

This research aids marketers by providing guidance on where response for the largest

portion of a brand’s customer base, light brand buyers, is better captured for two brand equity measures. This thesis also highlights best practice for analysing results for awareness and image measures. Recommendations for marketing practice are stated for brand awareness and brand image.

10.2.1 Brand Awareness From a practitioner perspective, results from this thesis demonstrate that light buyer response increases in accordance with the difficulty of the measure. Findings from the present study that response increases from TOM to unprompted to prompted awareness is consistent with results from prior studies (e.g., Laurent, Kapferer & Roussel 1995; Romaniuk et al. 2004; Romaniuk & Wight 2009; Wight 2010). To assess light brand buyer awareness, marketers should examine the prompted awareness measure, which captures the highest proportion of light brand response. The prompted awareness measure will provide marketers with results that more accurately reflect the number of light buyers who have the brand name linked to the category cue in memory.

Secondly, this thesis provides a greater understanding of analysis and interpretation of awareness results for light buyers. For all awareness measures, light buyers are better able to retrieve the brand name from memory than non-buyers. This suggests that analysis should occur separately for non and light buyer groups. In comparison to heavier buyers, light buyer response is significantly lower for TOM awareness. However, for unprompted and prompted awareness measures, light and heavier buyers respond in a similar manner. Therefore, while analysis should occur separately for non, light and heavier buyers for TOM awareness, there is no need to separate light and heavier buyer groups for unprompted and prompted awareness analysis. Rather, analysis should occur separately for buyers and non-buyers. Recommendations from this thesis are in line with previous findings by Wight (2010), who suggests separating awareness analysis for users and non-users. This thesis extends Wight’s work to examine brand awareness from a light buyer perspective.

Page 127: Understanding and Measuring Light Buyer Brand Equity

118

10.2.2 Brand Image Four image measurement conditions were selected for comparison that differed based

on their response format (free/forced) and prompting method (attribute/brand). Previously, Joyce (1963) incorporates PA and FCB measures and the two methods of prompting for response into his research design. Drawing from a sample of over 2,000 respondents, results from the present study update those conducted over 50 years ago via face-to-face interviews, extending analysis to a modern, online context and for a different type of brand buyer. Replication of the image measures confirms results found by Joyce (1963), that a PA attribute prompted technique is the justifiably preferred method for marketer use. Furthermore, similarity in response across the two studies supports the use of online technologies to capture consumer brand associations.

A key contribution for practice is providing marketers with guidance for when different types of image measures should be used in survey design. The PA measure captures light buyer associations linked to the brand in memory. Response for the measure can

be differentiated for light and non/heavier brand buyers, where significant differences between light and heavy buyer response allows the potential for light buyer response to increase over time. Findings fit with requirements for CBBE surveys, where measures must reflect brand knowledge contained in memory structures and be sensitive to change over time (Aaker 1996a). In comparison, the FCB measure consistently returns a higher response for light buyers, regardless of the prompting method used. As light buyers select almost all brands and attributes in the FCB measure, this could suggest response is reflecting associations other than those linked to the brand (as discussed in Section 10.1.2). More importantly, if light buyer response is high, the measure may not be sensitive to changes in buyer associations over time. Marketers may find the FCB measure useful when a stable and higher light buyer response is required (in line with Dolnicar, Rossiter & Grun 2012, findings), while the PA measure is recommended for use in CBBE surveys.

This research also provides guidance on the implementation of FCB and PA measures, whether brand or attribute prompted, to capture a higher proportion of light buyer response. Within the FCB measure, light buyer response trends in favour of brand prompting. However few brands show statistically significant differences in light buyer response for the two prompting methods. Therefore, marketers are free to select either approach when employing the FCB measure. For PA, response is significantly higher when attribute prompted than when brand prompted. Hence, brand equity surveys should incorporate a PA attribute prompted approach.

Page 128: Understanding and Measuring Light Buyer Brand Equity

119

10.3 Strengths of the Present Study

A notable strength of this thesis is the scope of research, where two brand equity

measures have been analysed for light brand buyers.

Part One compared three brand awareness measures for over 10,000 respondents, 52 brands, across four product categories and three countries. Awareness was analysed for light brand buyers in comparison with non-buyers and heavier buyers. A key strength of part one was analysing matched panel and brand equity survey data, where purchase data and response to awareness questions are available for the same individual.

In addition to analysing secondary data for four product categories, primary data was collected for comparison of four image measurement conditions. Data was collected in the UK for over 2,000 respondents, allowing a sample size of 500 for comparison between image measures and prompting method. A split sample approach allowed data to be collected for two product categories, each testing nine brands and 15 attributes. The use of multiple data sources for both brand equity measures contributes to the generalisability of the results.

10.4 Limitations of the Present Study

Despite the strengths of this research, it also has limits of scope, as analysis for brand awareness and brand image measures was for consumer packaged goods categories. Further research to extend beyond packaged goods categories is encouraged, such as durables or services (as described in Sharp, Wright & Goodhardt 2002).

Previous research (e.g., Dolnicar & Grün 2007; Driesener & Romaniuk 2006) determines that scaling and sorting image measures produce similar results, yet sorting techniques are considered to be faster and the preferred approach for marketer use. This research has examined two sorting measures for light brand buyer response. To add to knowledge of light buyer brand equity, future research should test the PA measure in comparison to scaling measures, such as ranking and rating.

Finally, this research focuses on light brand buyer response to brand image measures, only briefly addressing response for non-buyer and heavier buyer groups. Future research to further examine differences between light and non/heavier buyer response to the different measures and methods of prompting is encouraged.

Page 129: Understanding and Measuring Light Buyer Brand Equity

120

10.5 Avenues for Future Research

This study has highlighted four potential areas for future research, with discussion of

each provided below.

This research is focused on examining light brand buyer response to brand equity measures. Analysis for the present study draws from Anschuetz’s (2002) research, classifying light brand buyers via purchase frequency as the ‘once only’ buyer. Though the importance of light buyers is often stated, there is no clear definition of how to classify a light brand buyer. To account for inconsistencies in buyer classification, multiple methods of classification can be used to test if results are generalisable. For instance, Romaniuk and Wight (2014) analyse heavy buyer stability, classifying heavy buyers according to purchase weight relative to other buyers (highest 20% and highest 10%), and purchase frequency (buying more than five times in a given time period). Future research should test the generalisability of results presented in this thesis for different methods of light buyer classification. Further classification may look to Goodhardt’s 50:30:20 conceptualisation (Sharp 2010), which has been previously been used to classify buyers into light, medium and heavy buyer categories (Ludwichowska 2013). Alternatively, others have defined light buyers as ‘purchasing below average’ (Morrison 1966; Sharp 2013).

Future research should examine differences in light buyer response for different types of attributes. The current research design incorporated a mix of attribute types, consistent with Keller (1993). Previous research on brand image response has examined usage and response for evaluative and descriptive attributes (e.g., Barwise & Ehrenberg 1985). Analysis, however, has not been performed at the light brand buyer level. A brief search within the literature returns a myriad of attribute types for use in future research. For example, characteristic (those that directly relate to the physical features of the brand and/or product), functional (those that directly refer to what the product will do for the user) and imagery benefits (perceptions of what the brand is doing or providing for the user) (Hirschman 1980; Lefkoff-Hagius & Mason 1990, 1993; Steenkamp & Van Trijp 1997). Other attribute types include search, experience and credence (Srinivasan & Till 2002), concrete/abstract (Holbrook & Hirschman 1982) and hard/soft (Biel 1991). Understanding how light brand buyers respond to different types of attributes and incorporating these into a PA attribute prompted measure will ensure a more accurate assessment of the brand’s performance and effectiveness of past marketing activities.

Page 130: Understanding and Measuring Light Buyer Brand Equity

121

In addition to examining the types of attributes, further research should test light buyer response for differences in attribute wording. Romaniuk (2011) raises concern over strong question wording and light buyer ability to respond, where asking a question using strong adjectives, such as ‘best’ or ‘favourite’, implies that only one response is needed. Light buyers are occasional buyers of the brand and are also buyers of other brands in the category. As such, light buyers may respond with the brand they have the most associations with in memory, and which are readily accessible, i.e. the brand purchased more frequently. This may mean that statements such as ‘one of my favourite brands’ will reflect a heavier buyer response. Other authors note that attributes should also be worded to avoid ambiguity, in a way that all responses will be favourable, or with a fixed attribute level (i.e. cleans very well or being extremely convenient) (Joyce 1963; Rossiter, Dolnicar & Grun 2015). Identifying the ways in which attributes should and should not be worded will increase the effectiveness of image measures, in particular for capturing light brand buyer responses.

A fourth area of future research is to examine light brand buyer response over time. Past

studies (e.g., Romaniuk & Wight 2009; Sharp, Beal & Romaniuk 2001) have suggested separating analysis according to users and non-users to indicate brand reach and whether marketing strategies are achieving retention and/or acquisition of buyers. Light buyers make up the largest proportion of buyers in a brand’s customer base (Ehrenberg 1959) and consequently account for the largest portion of buyers a brand gains as it grows (Ehrenberg, Goodhardt & Barwise 1990). Therefore, changes in the brand are likely to be seen more prominently as changes in light buyer responses. Research should analyse light brand buyer equity during times of growth and decline, drawing from current research findings of where light brand buyer response is better captured.

Page 131: Understanding and Measuring Light Buyer Brand Equity

122

List of References Aaker, DA 1972, 'A Measure of Brand Acceptance', Journal of Marketing Research, vol. 9, May, pp. 160-167. Aaker, DA 1991, Managing Brand Equity: Capitalizing on the Value of a Brand Name, The Free Press, New York. Aaker, DA 1992, 'The Value of Brand Equity', Journal of Business Strategy, vol. 13, no. 4, pp. 27-32. Aaker, DA 1996a, Building Strong Brands, Free Press, New York. Aaker, DA 1996b, 'Measuring Brand Equity across Products and Markets', California Management Review, vol. 38, no. 3, Spring, pp. 102-120. Aaker, JL 1997, 'Dimensions of Brand Personality', Journal of Marketing Research, vol. 34, no. August, pp. 347-356. Agarwal, MK & Rao, VR 1996, 'An Empirical Comparison of Consumer-Based Measures of Brand Equity', Marketing Letters, vol. 7, no. 3, pp. 237-247. Alba, JW & Chattopadhyay, A 1986, 'Salience Effects in Brand Recall', Journal of Marketing Research, vol. 23, no. 4, November, pp. 363-369. Anderson, JR 1974, 'Retrieval of Propositional Information from Long-Term Memory', Cognitive psychology, vol. 6, no. 4, pp. 451-474. Anderson, JR & Bower, GH 1979, Human Associative Memory, Lawrence Erlbaum, Hillsdale, NJ. Anderson, JR 1983, 'A Spreading Activation Theory of Memory', Journal of Verbal Learning and Verbal Behavior, vol. 22, pp. 261-295. Anschuetz, N 2002, 'Why a Brand's Most Valuable Consumer Is the Next One It Adds', Journal of Advertising Research, vol. 42, no. 1, pp. 15-21. Assael, H & Day, GS 1968, 'Attitudes and Awareness as Predictors of Market Share', Journal of Advertising Research, vol. 8, no. 4, pp. 3-10. Baldinger, AL, Blair, E & Echambadi, R 2002, 'Why Brands Grow', Journal of Advertising Research, vol. 1, pp. 7-14. Barnard, NR & Ehrenberg, A 1990, 'Robust Measures of Consumer Brand Beliefs', Journal of Marketing Research, vol. 27, no. 4, pp. 477-484. Barsalou, L 2003, 'Situated Simulation in the Human Conceptual System', Language and cognitive processes, vol. 18, no. 5-6, pp. 513-562. Barsalou, LW 1983, 'Ad Hoc Categories', Memory & Cognition, vol. 11, no. No. 3, pp. 211-227.

Page 132: Understanding and Measuring Light Buyer Brand Equity

123

Barwise, TP & Ehrenberg, A 1985, 'Consumer Beliefs and Brand Usage', Journal of the Market Research Society, vol. 27, no. 2, pp. 81-93. Baumgartner, H & Steenkamp, J 2001, 'Response Styles in Marketing Research: A Cross National Investigation', Journal of Marketing Research, vol. 38, no. May, pp. 143-156. Biel, AL 1991, 'The Brandscape: Converting Brand Image into Equity', Admap, no. October, pp. 41-46. Bird, M & Ehrenberg, A 1966, 'Non-Awareness and Non-Usage', Journal of Advertising Research, vol. 6, no. December, pp. 4-8. Bird, M, Channon, C & Ehrenberg, A 1970, 'Brand Image and Brand Usage', Journal of Marketing Research, vol. 7, no. 3, pp. 307-314. Bird, M & Ehrenberg, A 1970, 'Consumer Attitudes and Brand Usage', Journal of the Market Research Society, vol. 12, no. 4, pp. 233-247. Castleberry, SB & Ehrenberg, A 1990, 'Brand Usage: A Factor in Consumer Beliefs', Marketing Research, vol. June, no. 2, pp. 14-21. Christodoulides, G & De Chernatony, L 2010, 'Consumer-Based Brand Equity Conceptualization and Measurement: A Literature Review', International Journal of Market Research, vol. 52, no. 1, pp. 43-66. Cobb-Walgren, CJ, Ruble, CA & Donthu, N 1995, 'Brand Equity, Brand Preference, and Purchase Intent', Journal of Advertising, vol. 24, no. 3, pp. 25-40. Collins, AM & Loftus, EF 1975, 'A Spreading Activation Theory of Semantic Processing', Psychological Review, vol. 82, no. 6, pp. 407-428. Crimmins, JC 1992, 'Better Measurement and Management of Brand Value', Journal of Advertising Research, vol. 32, no. 4, July/August, pp. 11-19. Dall'Olmo Riley, F 2012, Changes in Attitudes and Behaviour, Kingston Business School. Desai, KK & Hoyer, WD 2000, 'Descriptive Characteristics of Memory-Based Consideration Sets: Influence of Usage Occasion Frequency and Usage Location Frequency', Journal of Consumer Research, vol. 27, no. December, pp. 309-323. Diamantopoulos, A, Reynolds, NL & Simintiras, AC 2006, 'The Impact of Response Styles on the Stability of Cross-National Comparisons', Journal of Business Research, vol. 59, no. 8, pp. 925-935. Dolnicar, S & Grün, B 2007, 'How Constrained a Response: A Comparison of Binary, Ordinal and Metric Answer Formats', Journal of Retailing and Consumer Services, vol. 14, no. 2, pp. 108-122. Dolnicar, S & Rossiter, JR 2008, 'The Low Stability of Brand-Attribute Associations Is Partly Due to Market Research Methodology', International Journal of Research in Marketing, vol. 25, pp. 104-108.

Page 133: Understanding and Measuring Light Buyer Brand Equity

124

Dolnicar, S, Grün, B & Leisch, F 2011, 'Quick, Simple and Reliable: Forced Binary Survey Questions', International Journal of Market Research, vol. 53, no. 2, pp. 233-254. Dolnicar, S, Rossiter, JR & Grun, B 2012, '"Pick-Any" Measures Contaminate Brand Image Studies', International Journal of Market Research, vol. 54, no. 6, pp. 821-834. Driesener, C & Romaniuk, J 2006, 'Comparing Methods of Brand Image Measurement', International Journal of Market Research, vol. 48, no. 6, pp. 681-698. Ehrenberg, A 1959, 'The Pattern of Consumer Purchases', Applied Statistics, vol. 8, no. 1, pp. 26-41. Ehrenberg, A 1968, 'The Practical Meaning and Usefulness of the Nbd/Lsd Theory of Repeat-Buying', Applied Statistics, vol. 17, no. 1, pp. 17-32. Ehrenberg, A 1988, Repeat-Buying: Facts, Theory and Applications, Oxford University Press, London. Ehrenberg, A 1990, 'A Hope for the Future of Statistics: Msod', The American Statistician, vol. 44, no. 3, pp. 195-196. Ehrenberg, A, Goodhardt, G & Barwise, TP 1990, 'Double Jeopardy Revisited', Journal of Marketing, vol. 54, no. 3, pp. 82-91. Ehrenberg, A 1995, 'Empirical Generalisations, Theory, and Method', Marketing science, vol. 14, no. 3, pp. G20-G28. Ehrenberg, A 2000a, 'Repeat-Buying: Facts, Theory and Applications', Journal of Empirical Generalisations in Marketing Science, vol. 5, pp. 392-770. Ehrenberg, A 2000b, 'Data Reduction - Analysing and Interpreting Statistical Data', Journal of Empirical Generalisations in Marketing Science, vol. 5. Ehrenberg, A, Uncles, MD & Goodhardt, GG 2004, 'Understanding Brand Performance Measures: Using Dirichlet Benchmarks', Journal of Business Research, vol. 57, no. 12, pp. 1307-1325. Farquhar, PH 1989, 'Managing Brand Equity', Marketing Research, vol. 1, no. 1, September 1989, pp. 24-33. Farquhar, PH 1990, 'Managing Brand Equity', Journal of Advertising Research, vol. 30, no. 4, August/September, pp. RC7-RC12. Galesic, M & Bosnjak, M 2009, 'Effects of Questionnaire Length on Participation and Indicators of Response Quality in a Web Survey', Public Opinion Quarterly, vol. 73, no. 2, pp. 349-360. Goldsmith, RE, Flynn, LR & Bonn, M 1994, 'An Empirical Stud of Heavy Users of Travel Agencies', Journal of Travel Research, vol. 33, no. 1, pp. 38-43. Goodhardt, G & Ehrenberg, A 1967, 'Conditional Trend Analysis: A Breakdown by Initial Purchasing Level', Journal of Marketing Research, vol. 4, no. May, pp. 155-161.

Page 134: Understanding and Measuring Light Buyer Brand Equity

125

Goodhardt, GJ, Ehrenberg, A & Chatfield, C 1984, 'The Dirichlet: A Comprehensive Model of Buying Behaviour', Journal of the Royal Statistical Society, vol. 147, no. 5, pp. 621-655. Gruber, A 1969, 'Top-of-Mind Awareness and Share of Families: An Observation', Journal of Marketing Research, vol. 6, no. May, pp. 227-231. Haley, RI & Case, PB 1979, 'Testing Thirteen Attitude Scales for Agreement and Brand Discrimination', Journal of Marketing, vol. 43, no. Fall, pp. 20-32. Heil, M, Rösler, F & Hennighausen, E 1994, 'Dynamics of Activation in Long-Term Memory: The Retrieval of Verbal, Pictorial, Spatial, and Color Information', Journal of Experimental Psychology: Learning, Memory and Cognition, vol. 20, no. 1, pp. 185-200. Hirschman, EC 1980, 'Attributes of Attributes and Layers of Meaning', Advances in Consumer Research, vol. 7, pp. 7-12. Hogan, S 2012, 'Attribute Elicitation Procedures: A Comparison of Four Techniques', University of South Australia. Holbrook, MB & Hirschman, EC 1982, 'The Experiential Aspects of Consumption: Consumer Fantasies, Feelings, and Fun', Journal of Consumer Research, vol. 9, no. September, pp. 132-140. Holbrook, MB, Moore, WL & Winer, RS 1982, 'Constructing Joint Spaces from Pick-Any Data: A New Tool for Consumer Analysis', Journal of Consumer Research, vol. 9, no. June, pp. 99-105. Holden, SJ & Lutz, RJ 1992, 'Ask Not What the Brand Can Evoke; Ask What Can Evoke the Brand?', Advances in Consumer Research, vol. 19, no. 1, pp. 101-107. Holden, SJS 1993, 'Understanding Brand Awareness: Let Me Give You a C(L)Ue!', Advances in Consumer Research, vol. 20, no. 1, pp. 383-388. Howard, JA & Sheth, JN 1969, The Theory of Buyer Behavior, John & Wiley Sons, Inc, New York. Hoyer, WD & Brown, SP 1990, 'Effects of Brand Awareness on Choice for a Common, Repeat-Purchase Product', Journal of Consumer Research, vol. 17, no. 2, pp. 141-148. Huang, R & Sarigöllü, E 2012, 'How Brand Awareness Relates to Market Outcome, Brand Equity, and the Marketing Mix', Journal of Business Research, vol. 65, no. 1, pp. 92-99. Hutchinson, JW 1983, 'Expertise and the Structure of Free Recall', Advances in Consumer Research, vol. 10, pp. 585-589. Hutchinson, JW, Raman, K & Mantrala, MK 1994, 'Finding Choice Alternatives in Memory: Probability Models of Brand Name Recall', Journal of Marketing Research, vol. 31, no. November, pp. 441-461.

Page 135: Understanding and Measuring Light Buyer Brand Equity

126

Joyce, T 1963, 'Techniques of Brand Image Measurement', New Developments in Research, Market Research Society, London, pp. 45-63. Kapferer, J 2008, The New Strategic Brand Management: Creating and Sustaining Brand Equity Long Term, Kogan Page Ltd, Kapferer, J-N 1995, Strategic Brand Management - New Approaches to Creating and Evaluating Brand Equity, Kogan Page Limited, London. Keller, K 2001, Building Customer-Based Brand Equity: A Blueprint for Creating Strong Brands, Marketing Science Institute, Cambridge, Mass. Keller, KL 1993, 'Conceptualizing, Measuring, and Managing Customer-Based Brand Equity', The Journal of Marketing, vol. 57, no. 1, pp. 1-22. Keller, KL & Davey, KK 2001, 'Building Customer-Based Brand Equity', Advertising Research Foundation workshop, New York. Keller, KL 2005, 'Measuring Brand Equity', Dartmouth College, To appear in Handbook of Marketing Research -Do's and Dont's. Keller, KL & Lehmann, DR 2009, 'Assessing Long-Term Brand Potential', Journal of Brand Management, vol. 17, no. 1, pp. 6-17. Knox, S & Walker, D 2001, 'Measuring and Managing Brand Loyalty', Journal of Strategic Marketing, vol. 9, no. 2, pp. 111-128. Koch, R 1999, The 80/20 Principle: The Secret to Success by Achieving More with Less, Doubleday, New York. Krosnick, JA 1999, 'Survey Research', Annual review of psychology, vol. 50, no. 1, pp. 537-567. Laurent, G, Kapferer, J-N & Roussel, F 1995, 'The Underlying Structure of Brand Awareness Scores', Marketing science, vol. 14, no. No. 3, Part 2, pp. G170-G179. Lefkoff-Hagius, R & Mason, CH 1990, 'The Role of Tangible and Intangible Attributes in Similarity and Preference Judgement', Advances in Consumer Research, vol. 17, no. 1, pp. 135-143. Lefkoff-Hagius, R & Mason, CH 1993, 'Characteristic, Beneficial, and Image Attributes in Consumer Judgements of Similarity and Preference', Journal of Consumer Research, vol. 20, June 1993, pp. 100 - 109. Light, L 1994, 'Brand Loyalty Marketing: Today's Marketing Mandate', Editor & Publisher, no. December 10, pp. 20T-24T. Likert, R 1932, 'A Technique for the Measurement of Attitudes', Archives of Psychology, vol. 140, pp. 44-53. Ludwichowska, GM 2013, 'Can We Fix Errors in Self-Reported Buying Frequencies', Submission for the award of Master of Business (Research), no. forthcoming.

Page 136: Understanding and Measuring Light Buyer Brand Equity

127

Lynch, JG, Jr & Srull, TK 1982, 'Memory and Attentional Factors in Consumer Choice: Concepts and Research Methods', Journal of Consumer Research, vol. 9, no. 1, pp. 18-37. Macdonald, E & Sharp, B 1996, 'Management Perceptions of the Importance of Brand Awareness as an Indication of Advertising Effectiveness', Marketing Research On-Line, vol. 1, pp. 1-15. Macdonald, E & Sharp, B 2000, 'Brand Awareness Effects on Consumer Decision Making for a Common, Repeat Purchase Product: A Replication', Journal of Business Research, vol. 48, no. 1, pp. 5-15. McDonald, C & Ehrenberg, A 2003, What Happens When Brands Gain or Lose Share?: Customer Acquisition or Increased Loyalty?, Ehrenberg-Bass Institute for Marketing Science, Adelaide. Morrison, DG 1966, 'Interpurchase Time and Brand Loyalty', Journal of Marketing Research, vol. 3, no. August, pp. 289-291. Myers, CA 2003, 'Managing Brand Equity: A Look at the Impact of Attributes', Journal of Product & Brand Management, vol. 12, no. 1, pp. 39-51. Neal, W & Strauss, R 2008, 'A Framework for Measuring and Managing Brand Equity What Gets Measured Gets Managed', Marketing Research, vol. 20, no. 2, p. 6. Nedungadi, P & Hutchinson, JW 1985, 'The Prototypicality of Brands: Relationships with Brand Awareness, Preference and Usage', Advances in Consumer Research, vol. 12, no. 1, pp. 498-503. Nedungadi, P 1990, 'Recall and Consumer Consideration Sets: Influencing Choice without Altering Brand Evaluations', Journal of Consumer Research, vol. 17, no. 3, pp. 263-276. Nenycz-Thiel, M, Beal, V, Ludwichowska, G & Romaniuk, J 2012, 'Investigating the Accuracy of Self-Reports of Brand Usage Behavior', Journal of Business Research, vol. 66, no. 2, pp. 224-232. Neuman, W 2011, Social Research Methods: Qualitative and Quantitative Approaches, International Edition, 7th edn, Allyn & Bacon, Boston. Percy, L & Rossiter, JR 1992, 'A Model of Brand Awareness and Brand Attitude Advertising Strategies', Psychology and Marketing, vol. 9, no. 4, July/August, pp. 263-274. Persky, J 1992, 'Pareto's Law', Journal of Economic Perspectives, vol. 6, no. 2, p. 181. Ratneshwar, S & Shocker, AD 1991, 'Substitution in Use and the Role of Usage Context in Product Category Structures', Journal of Marketing Research, vol. 28, no. 3, pp. 281-295. Ratneshwar, S, Barsalou, LW, Pechmann, C & Moore, M 2001, 'Goal-Derived Categories: The Role of Personal and Situational Goals in Category Representations', Journal of Consumer Psychology, vol. 10, no. 3, pp. 147-157.

Page 137: Understanding and Measuring Light Buyer Brand Equity

128

Romaniuk, J & Sharp, B 2004, 'Conceptualizing and Measuring Brand Salience', Marketing Theory, vol. 4, no. 4, pp. 327-342. Romaniuk, J, Sharp, B, Paech, S & Driesener, C 2004, 'Brand and Advertising Awareness: A Replication and Extension of a Known Empirical Generalisation', Australasian Marketing Journal, vol. 12, no. 3, pp. 70-80. Romaniuk, J & Sharp, B 2008, Where Knowledge of Your Brand Resides: The Pareto Share of Brand Knowledge, Ehrenberg-Bass Institute for Marketing Science, Adelaide. Romaniuk, J & Wight, S 2009, 'The Influence of Brand Usage on Responses to Advertising Awareness Measures', International Journal of Market Research, vol. 51, no. 2, pp. 203-218. Romaniuk, J & Wight, S 2010, Do Your Heavy Buyers Stay Heavy, and What Are They Worth?, Ehrenberg-Bass Institute, Adelaide. Romaniuk, J 2011, 'Are You Blinded by the Heavy (Buyers)...Or Are You Seeing the Light?', Journal of Advertising Research, vol. 51, no. 4, pp. 561-563. Romaniuk, J, Bogomolova, S & Dall'Olmo Riley, F 2012, 'Brand Image and Brand Usage: Is a Forty-Year-Old Empirical Generalization Still Useful?', Journal of Advertising Research, vol. 52, no. 2. Romaniuk, J 2013, 'How Healthy Is Your Brand-Health Tracker? A Five-Point Checklist to Build Returns on a Critical Research Investment', Journal of Advertising Research, vol. 53, no. 1, pp. 11-13. Romaniuk, J, Dawes, J & Nenycz-Thiel, M 2014, 'Generalizations Regarding the Growth and Decline of Manufacturer and Store Brands', Journal of Retailing and Consumer Services, vol. 21, no. 5, pp. 725-734. Romaniuk, J & Wight, S 2014, 'The Stability and Sales Contribution of Heavy Buying Households', Journal of Consumer Behaviour, vol. forthcoming. Rossiter, JR, Dolnicar, S & Grun, B 2015, 'Why the Level-Free Forced-Choice Binary Measure of Brand Benefit Beliefs Works So Well', International Journal of Market Research, vol. 57, no. 2, pp. 239-256. Schmittlein, DC, Bemmaor, AC & Morrison, DG 1985, 'Why Does the Nbd Model Work? Robustness in Representing Product Purchases, Brand Purchases and Imperfectly Recorded Purchases', Marketing science, vol. 4, no. No. 3, Summer, pp. 255-266. Schmittlein, DC, Cooper, LG & Morrison, DG 1993, 'Truth in Concentration in the Land of (80/20) Laws', Marketing science, vol. 12, no. 2, Spring, pp. 167-183. Sharp, A & Romaniuk, J 2002, 'Brand to Attribute or Attribute to Brand - Which Is the Path to Stability?', European Marketing Academy 31st annual conference, University of Minho, Portugal, 28 - 31 May.

Page 138: Understanding and Measuring Light Buyer Brand Equity

129

Sharp, B, Beal, V & Romaniuk, J 2001, 'First Steps Towards a Marketing Empirical Generalisation: Brand Usage and Subsequent Advertising Recall', ANZMAC, Auckland, 3-5 December. Sharp, B, Wright, M & Goodhardt, G 2002, 'Purchase Loyalty Is Polarised into Either Repertoire or Subscription Patterns', Australasian Marketing Journal, vol. 10, no. 3, pp. 7-20. Sharp, B & Romaniuk, J 2007, There Is a Pareto Law - but Not as You Know It, Ehrenberg-Bass Institute for Marketing Science, Adelaide. Sharp, B 2010, How Brands Grow, Oxford University Press, Melbourne. Sharp, B 2013, Marketing: Theory, Evidence, Practice, Oxford University Press, Melbourne. Shulman, A 1973, 'A Comparison of Two Scales on Extremity Response Bias', Public Opinion Quarterly, vol. 37, no. 3, Fall, pp. 407-412. Simon, CJ & Sullivan, MW 1993, 'The Measurement and Determinants of Brand Equity: A Financial Approach', Marketing science, vol. 12, no. 1, pp. 28-52. Srinivasan, S & Till, BD 2002, 'Evaluation of Search, Experience and Credence Attributes: Role of Brand Name and Product Trial', Journal of Product and Brand Management, vol. 11, no. 7, pp. 417-432. Steenkamp, J-B & Van Trijp, H 1997, 'Attribute Elicitation in Marketing Research: A Comparison of Three Procedures', Marketing Letters, vol. 8, no. 2, April, pp. 153-165. Stern, P & Ehrenberg, A 2003, 'Expectations Vs. Reality', Marketing Insights, Marketing Research, vol. Spring, pp. 40-43. Sudman, S & Wansink, B 2002, Consumer Panels, 2 edn, American Marketing Association, Chicago, Ill. Sylvester, AK, McQueen, J & Moore, SD 1994, 'Brand Growth and 'Phase 4' Marketing', Admap, September. Tulving, E & Pearlstone, Z 1966, 'Availability Versus Accessibility of Information in Memory for Words', Journal of Verbal Learning and Verbal Behavior, vol. 5, no. 4, pp. 381-391. Tulving, E 1972, 'Episodic and Semantic Memory', Organization of memory, pp. 381-402. Tulving, E & Thomson, D 1973, 'Encoding Specificity and Retrieval Processes in Episodic Memory', Psychological Review, vol. 80, no. 5, pp. 352-373. Twedt, DW 1964, 'How Important to Marketing Strategy Is the "Heavy User"?', Journal of Marketing, vol. 28, no. 1, pp. 71-72.

Page 139: Understanding and Measuring Light Buyer Brand Equity

130

Uncles, MD, Hammond, KA, Ehrenberg, A & Davies, RE 1994, 'A Replication Study of Two Brand-Loyalty Measures', European journal of operational research, vol. 76, no. 2, pp. 375-385. Uncles, MD, Ehrenberg, A & Hammond, K 1995, 'Patterns of Buyer Behavior: Regularities, Models, and Extensions', Marketing science, vol. 14, no. 3, pp. G61-G70. Uncles, MD & Wright, M 2004, 'Editorial: Empirical Generalisation in Marketing', Australasian Marketing Journal, vol. 12, no. 3, pp. 5-12. Van Osselaer, SMJ & Janiszewski, C 2001, 'Two Ways of Learning Brand Associations', Journal of Consumer Research, vol. 28, no. September, pp. 202-223. Wight, ST 2010, 'Brand Awareness Metrics: The Underlying Awareness of Brand Users and Non-Users.', Ehrenberg-Bass Institute for Marketing Science, University of South Australia. Zaichkowsky, JL 2010, 'Strategies for Distinctive Brands', Journal of Brand Management, vol. 17, no. 8, pp. 548-560. Zinkhan, GM, Locander, WB & Leight, JH 1986, 'Dimensional Relationships of Aided Recall and Recognition', Journal of Advertising, vol. 15, no. 1, pp. 38-46.

Page 140: Understanding and Measuring Light Buyer Brand Equity

Appendix A: Image Questionnaire INTRODUCTION: Thank you for agreeing to take part in this study. It will take approximately 10-15 minutes. All of your responses are kept confidential. This survey is about your knowledge and experience of brands. There are no right or wrong answers, we are just interested in your honest opinions and answers. The first questions are about you and are for classification purposes only. [ASK ALL/SR] [TERMINATE CODE 1] QS1. Which of the following categories captures your age? Please select one answer below.

CODE

<18 years 1 18 – 28 years 2 29 – 34 years 3 36 – 44 years 4 45 – 54 years 5 55 – 64 years 6

65+ years 7 [REPRESENTATIVE QUOTA ACROSS CODES 2-7] [ASK ALL/SR] QS2. Are you… Please select one answer below.

CODE

Male 1 Female 2

[QUOTA – EVEN SPLIT FOR EACH GROUP] [ASK ALL/SR] [TERMINATE IF QS4=10] QS4. Which of the following best describes the region of the UK you live in? Please select one answer below.

CODE

London 1 Midlands 2

North East 3 Yorkshire and The Humber 4

Lancashire / North West 5 South 6

Scotland 7 East England 8

Wales / West England 9 None of these [EXC] 10

[SAMPLE TO ENSURE NATIONALLY REPRESENTATIVE

131

[ASK ALL/MR – RANDOMISE 1-7] [TERMINATE IF CU1 DOES NOT = 1 & 3] QCU1. Which of the following categories have you bought in the past 12 months? Please select all that apply.

CODE

Butter or Margarine 1 Biscuits 2

Breakfast Cereal 3 Instant Coffee 4

Packet Tea 5 Energy Drinks 6

Soft Drinks 7 None of these [EXC] 8

[ASK ALL/ ASK CATEGORIES SELECTED AT QCU1 /MR] QCU2. How many times have you bought from each of these categories in the past 12 months? Please use the pull-down menu to choose the appropriate number.

CATEGORY SELECTED AT QCU1 PULL DOWN BOX WITH OPTIONS OF 1 TO 10+

RESPONDENTS RANDOMLY ALLOCATED IN ONE OF FOUR GROUPS. EACH RESPONDENT ANSWERS TWO METHODS FOR TWO DIFFERENT PRODUCT CATEGORIES. BREAKFAST CEREAL GROUP A – PA- ATTRIBUTE GROUP B – FCB- ATTRIBUTE GROUP C – PA- BRAND GROUP D – FCB- BRAND

BUTTER AND MARGARINE GROUP A – FCB- BRAND GROUP B – PA- BRAND GROUP C – FCB- ATTRIBUTE GROUP D – PA- ATTRIBUTE

PICK ANY (ATTRIBUTE PROMPTED) APPROACH – FOR GROUP A Q2A. We would like to know how you regard different brands of breakfast cereal. We will give you a series of statements and for each we would like to know which brands you think have this quality. You can select as many or as few brands as you like, it doesn’t matter if you have used the brand or not, it’s just your opinion we are after. ONE ATTRIBUTE ON EACH PAGE EACH WITH A BRAND LIST UNDERNEATH IT WITH MULTIPLE RESPONSES POSSIBLE, AND THEN THE ORDER OF PAGES IS ROTATED BRAND LIST RANDOMISED FOR EACH RESPONDENT, EXCEPT NONE OF THESE ALWAYS AT THE END.

Page 141: Understanding and Measuring Light Buyer Brand Equity

Q2 <insert attribute> Please select all that apply

1. Brand A 2. Brand B 3. Brand C 4. Brand D 5. Brand E 6. Brand F 7. Brand G 8. Brand H 9. Brand I

97. None of these / Don’t know [EXC]

ATTRIBUTE LIST – BREAKFAST CEREAL 1 Good value for money 9 A healthy option 2 Good for the whole family 10 Natural 3 Would taste great 11 A modern brand 4 Stays crispy in milk 12 A brand I feel positively about 5 Would give me energy for the day 13 Better quality than other brands 6 Kids would like it 14 Appeals to you more than other brands 7 Innovative 15 Offers something other brands do not 8 Good for a treat FORCED CHOICE BINARY (ATTRIBUTE PROMPTED) APPROACH – FOR GROUP B Q2B. We would like to know how you regard different brands of breakfast cereal. We will give you a series of statements. For each brands, we’d like you to tell us if you think it has this quality or not. It doesn’t matter if you have bought/tried the brand or not, it’s just your opinion we are after. BRAND AND ATTRIBUTE LIST FROM Q2A. ONE ATTRIBUTE ON EACH PAGE EACH WITH A BRAND LIST UNDERNEATH IT WITH ONE RESPONSE POSSIBLE, AND THEN ORDER OF PAGES IS ROTATED. ATTRIBUTE LIST RANDOMISED FOR EACH RESPONDENT.

Q2 <insert attribute> Please select one answer for each brand.

1. Brand 1 � Yes � No 2. Brand 2 � Yes � No 3. Brand 3 � Yes � No 4. Brand 4 � Yes � No 5. Brand 5 � Yes � No

132

PICK ANY (BRAND PROMPTED) APPROACH – FOR GROUP C Q2C. We will give you a series of brands and for each brand we would like to know which qualities you think this brand has. We’d now like you to think about the breakfast cereal, Brand A. You can select as many or as few qualities as you like, it doesn’t matter if you have bought/tried the brand or not, it’s your opinion we are after. BRAND AND ATTRIBUTE LIST FROM Q2A. ONE BRAND ON EACH PAGE EACH WITH AN ATTRIBUTE LIST UNDERNEATH IT WITH MULTIPLE RESPONSES POSSIBLE, AND ORDER OF PAGES IS ROTATED. ATTRIBUTE LIST RANDOMISED FOR EACH RESPONDENT, EXCEPT NONE OF THESE ALWAYS AT THE END.

Q2 <insert brand> Please select all that apply

1. Attribute 1 2. Attribute 2 3. Attribute 3 4. Attribute 4 5. Attribute 5

97. None of these / Don’t know [EXC] FORCED CHOICE BINARY (BRAND PROMPTED) APPROACH – FOR GROUP D Q2D. We’d not like you to think about the breakfast cereal, Brand A. We will give you a series of statements and for each we would like to know if you think the brand has that quality or not. It doesn’t matter if you have bought/tried the brand or not, it’s your opinion we are after. BRAND AND ATTRIBUTE LIST FROM Q2A. ONE ATTRIBUTE ON EACH PAGE EACH WITH A BRAND LIST UNDERNEATH IT WITH ONE RESPONSE POSSIBLE, AND THEN ORDER OF PAGES ROTATED. ATTRIBUTES LIST RANDOMISED FOR EACH RESPONDENT.

Q2 <insert brand> Please select one answer for each brand.

1. Attribute 1 � Yes � No 2. Attribute 2 � Yes � No 3. Attribute 3 � Yes � No 4. Attribute 4 � Yes � No 5. Attribute … � Yes � No

Page 142: Understanding and Measuring Light Buyer Brand Equity

CATEGORY ONE (BREAKFAST CEREAL) USAGE [ASK ALL/MR - RANDOMISE] Q3A. Which brands of Breakfast Cereal have you bought in the past 12 months? Please select all that apply.

CODE

Brand A 1 Brand B 2 Brand C 3 Brand D 4 Brand E 5 Brand F 6 Brand G 7 Brand H 8 Brand I 9

Don’t know / can’t remember 97 Other (please write) 98

[ASK ALL/ FOR ALL BRANDS SELECTED AT Q3A /MR] Q3B. How many times have you bought each of these brands in the past 12 months? Please use the pull-down menu to choose the appropriate number.

THOSE SELECTED AT Q3A PULL DOWN BOX WITH OPTIONS OF 1 TO 10+

FORCED CHOICE BINARY (BRAND PROMPTED) APPROACH – FOR GROUP A Q4A. We’d now like you to think about the butter/margarine, Brand A. We will give you a series of statements and for each we would like to know if you think the brand has the quality of not. It doesn’t matter if you have used the brand or not, its just your opinion we are after. ONE ATTRIBUTE ON EACH PAGE EACH WITH A BRAND LIST UNDERNEATH IT WITH ONE RESPONSE POSSIBLE, AND THEN ORDER OF PAGES ROTATED. ATTRIBUTES LIST RANDOMISED FOR EACH RESPONDENT.

133

Q4 <insert brand> Please select one answer for each brand.

1. Good value for money � Yes � No 2. Good for the whole family � Yes � No 3. Would taste great � Yes � No 4. Spreads easily � Yes � No 5. Help control cholesterol � Yes � No 6. Kids would like it � Yes � No 7. Innovative � Yes � No 8. Good for cooking / baking � Yes � No 9. A healthy option � Yes � No 10. Natural � Yes � No 11. A modern brand � Yes � No 12. A brand I feel positively about � Yes � No 13. Better quality than other brands � Yes � No 14. Appeals to you more than other brands � Yes � No 15. Offers something other brands do not � Yes � No

BRAND LIST – BUTTER AND MARGARINE

1 Brand A 7 Brand G 2 Brand B 8 Brand H 3 Brand C 9 Brand I 4 Brand D 5 Brand E 6 Brand F PICK ANY (BRAND PROMPTED) APPROACH – FOR GROUP B Q4B. We will give you a series of brands and for each brand we would like to know which qualities you think this brand has. We’d not like you to think about the butter/margarine, Brand A. You can select as many of as few qualities as you like, it doesn’t matter if you have bought/tried the brand or not, it’s your opinion we are after. BRAND AND ATTRIBUTE LIST FROM Q4A. ONE ATTRIBUTE ON EACH PAGE EACH WITH A BRAND LIST UNDERNEATH IT WITH MULTIPLE RESPONSES POSSIBLE, AND THEN ORDER OF PAGES ROTATED. BRAND LIST RANDOMISED FOR EACH RESPONDENT, EXCEPT NONE OF THESE ALWAYS AT THE END.

Page 143: Understanding and Measuring Light Buyer Brand Equity

Q4 <insert brand> Please select all that apply

1. Attribute 1 2. Attribute 2 3. Attribute 3 4. Attribute 4 5. Attribute 5

97. None of these / Don’t know [EXC] FORCED CHOICE BINARY (ATTRIBUTE PROMPTED) APPROACH – FOR GROUP C Q4C. We would like to know how you regard different brands of butter/margarine. We will give you a series of statements. For each brands, we’d like you to tell us if you think it has this quality or not. It doesn’t matter if you have bought/tried the brand or not, it’s just your opinion we are after. BRAND AND ATTRIBUTE LIST FROM Q4A. ONE ATTRIBUTE ON EACH PAGE EACH WITH A BRAND LIST UNDERNEATH IT WITH ONE RESPONSE POSSIBLE, AND THEN ORDER OF PAGES ROTATED. ATTRIBUTES LIST RANDOMISED FOR EACH RESPONDENT.

Q4 <insert attribute> Please select one answer for each brand.

1. Brand 1 � Yes � No 2. Brand 2 � Yes � No 3. Brand 3 � Yes � No 4. Brand 4 � Yes � No 5. Brand 5 � Yes � No

PICK ANY APPROACH (ATTRIBUTE PROMPTED) – FOR GROUP D Q4D. We would like to know how you regard different brands of butter/margarine. We will give you a series of statements and for each we would like to know which brands you think have this quality. You can select as many or as few brands as you like, it doesn’t matter if you have bought/tried the brand or not, it’s your opinion we are after. BRAND AND ATTRIBUTE LIST FROM Q4A. ONE ATTRIBUTE ON EACH PAGE EACH WITH A BRAND LIST UNDERNEATH IT WITH MULTIPLE RESPONSES POSSIBLE, AND THEN ORDER OF PAGES ROTATED. BRAND LIST RANDOMISED FOR EACH RESPONDENT, EXCEPT NONE OF THESE ALWAYS AT THE END.

134

Q4 <insert attribute> Please select all that apply

1. Brand 1 2. Brand 2 3. Brand 3 4. Brand 4 5. … 6. Brand 9

97. None of these [EXC] CATEGORY TWO USAGE (BUTTER AND MARGARINE) [ASK ALL/MR - RANDOMISE] Q5A. Which brands of Butter and Margarine have you bought in the past 12 months? Please select all that apply.

CODE

Brand A 1 Brand B 2 Brand C 3 Brand D 4 Brand E 5 Brand F 6 Brand G 7 Brand H 8 Brand I 9

Don’t know / can’t remember 97 Other (please write) 98

[ASK ALL/ FOR ALL BRANDS SELECTED AT Q5A /MR] Q5B. How many times have you bought each of these brands in the past 12 months? Please use the pull-down menu to choose the appropriate number.

THOSE SELECTED AT Q6A PULL DOWN BOX WITH OPTIONS OF 1 TO 10+

Page 144: Understanding and Measuring Light Buyer Brand Equity

[ASK ALL/ SR] Q6A. Which of the following categories best represents your total annual income? Please select an answer below.

CODE

Less than £10,000 1 £10,000 to £20,000 2 £20,000 to £30,000 3 £30,000 to £50,000 4

£50,000 to £100,000 5 £100,000 or more 6

Don’t know / Prefer not to say 97 [ASK ALL/ SR] Q6B. Which best describes your highest level of education? Please select an answer below.

CODE

Did not complete Secondary School 1 Secondary Education (GCSE, or similar) 2

Post-Secondary education (A-Levels, or similar) 3 Undergraduate / Bachelor Degree 4

Postgraduate Degree 5 Doctorate (Ph.D) 6

Not Sure 97

THANK YOU FOR TAKING THIS SURVEY

135

Page 145: Understanding and Measuring Light Buyer Brand Equity

Appendix B: Income & Education Demographics HM Revenue and Customs, as part of the Government of United Kingdom National1 Statistics,

determines that the median income before tax for the tax period 2012-2013 is £21,000. A

further breakdown of income by percentile points shows that 80% of taxpayers earn £10,000 to

£50,000 per financial year. This breakdown of UK income is used to formulate appropriate

categories to capture respondent income. In line with this distribution, respondents are asked to

identify out of five categories, which best represents their total annual income.

Table 65: Income categories tested and UK taxpayer distribution.

Survey Income Categories

% 2012-13 Taxpayer

distribution Less than £10,000 5% £10,000 to £20,000 40% £20,000 to £30,000 25% £30,000 to £50,000 15%

£50,000 to £100,000 8% £100,000 or more 2%

In line with qualifications across all regions of the United Kingdom2 (including, England, Wales

and Scotland), six categories are provided to respondents to assess their highest level of education. Categories range from ‘did not complete secondary school’ to ‘Doctorate (Ph.D)’. Table X and X detail the breakdown of income and education levels for the four groups surveyed.

1 https://www.gov.uk/government/statistics/distribution-of-total-income-before-and-after-tax-by-gender-2010-to-2011 https://www.gov.uk/government/statistics/percentile-points-for-total-income-before-and-after-tax-1992-to-2011

2 https://www.gov.uk/what-different-qualification-levels-mean/compare-different-qualification-levels

136

Table 66: Income categories tested and UK taxpayer distribution.

Income Categories % Income Group A Group B Group C Group D

Less than £10,000 15 11 10 13 £10,000 to £20,000 20 30 25 27 £20,000 to £30,000 19 20 21 18 £30,000 to £50,000 24 22 25 24 £50,000 to £100,000 13 11 10 10 £100,000 or more 1 1 1 2 Don’t know 9 6 8 7

Table 67: Income categories tested and UK taxpayer distribution.

Education Categories % Education Group A Group B Group C Group D

Did not complete Secondary School 2 2 1 2 Secondary (GCSE, or similar) 33 33 31 28 Post-Secondary (A-levels, or similar) 28 27 28 30 Undergraduate / Bachelor degree 26 25 26 27 Postgraduate degree / study 9 9 13 10 Doctorate (Ph.D) 1 2 1 2 Not sure / prefer not to say 2 7 1 1

Page 146: Understanding and Measuring Light Buyer Brand Equity

Appendix C: Buyer Purchase Frequency

Distribution Figure 20: Distribution of buyer purchase frequency, Cereal.

137

Figure 11: Distribution of buyer purchase frequency, Butter/margarine.

Page 147: Understanding and Measuring Light Buyer Brand Equity

Appendix D: PA: Attribute versus Brand

Prompted Table 68: Proportion of light brand buyer cereal PA response – Stays crispy in milk.

Stays crispy in milk

%Ave Pen3

PA Attribute

PA Brand

Brand A 15 10 6 Brand B 15 38* 24 Brand C 14 52** 34 Brand D 13 33 25 Brand E 13 23 24 Brand F 13 30* 46 Brand G 11 25 37 Brand H 11 36 35 Average 31 29

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05.

Table 69: Proportion of light brand buyer cereal PA response – Natural.

Natural %Ave Pen

PA Attribute

PA Brand

Brand A 15 73** 56 Brand B 15 47 43 Brand C 14 9 17 Brand D 13 37* 24 Brand E 13 46 43 Brand F 13 11 21 Brand G 11 6 6 Brand H 11 7 10 Average 29 27

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05.

3 % Average light brand buyer penetration across the four survey groups.

138

Table 70: Proportion of light brand buyer cereal PA response – Good value for money.

Good value for money

%Ave Pen

PA Attribute

PA Brand

Brand A 15 74** 46 Brand B 15 70** 44 Brand C 14 52** 17 Brand D 13 68** 50 Brand E 13 71** 28 Brand F 13 46* 31 Brand G 11 47** 21 Brand H 11 52** 32 Average 60 34

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05

Table 11: Proportion of light brand buyer cereal PA response – Good for the whole family.

Good for the whole family

%Ave Pen

PA Attribute

PA Brand

Brand A 15 73** 54 Brand B 15 71 59 Brand C 14 44 34 Brand D 13 61 54 Brand E 13 63 57 Brand F 13 48 59 Brand G 11 33 35 Brand H 11 46 42 Average 55 49

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05

Table 72: Proportion of light brand buyer cereal PA response – Kids would like it.

Kids would like it %Ave Pen

PA Attribute

PA Brand

Brand A 15 23 28 Brand B 15 25 20 Brand C 14 31 34 Brand D 13 71* 57 Brand E 13 40 43 Brand F 13 74* 57 Brand G 11 92** 75 Brand H 11 64 60 Average 52 47

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05

Page 148: Understanding and Measuring Light Buyer Brand Equity

Table 73: Proportion of light brand buyer cereal PA response – Good for a treat.

Good for a treat %Ave Pen

PA Attribute

PA Brand

Brand A 15 9 10 Brand B 15 22 26 Brand C 14 53 51 Brand D 13 26 26 Brand E 13 17 31* Brand F 13 23 26 Brand G 11 69 63 Brand H 11 54 50 Average 34 36

Statistically significantly higher than PA Attribute at *p<0.1.

Table 74: Proportion of light brand buyer cereal PA response – Would give me energy for the day.

Would give me energy for day

%Ave Pen

PA Attribute

PA Brand

Brand A 15 77** 53 Brand B 15 48** 31 Brand C 14 47 38 Brand D 13 26 21 Brand E 13 60* 42 Brand F 13 34 25 Brand G 11 22 15 Brand H 11 32 27 Average 43 32

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05.

Table 75: Proportion of light brand buyer cereal PA response – A healthy option.

A healthy option %Ave Pen

PA Attribute

PA Brand

Brand A 15 77 71 Brand B 15 53 47 Brand C 14 17 22 Brand D 13 41 36 Brand E 13 58 60 Brand F 13 20 39** Brand G 11 10 10 Brand H 11 4 10 Average 35 37

Statistically significantly higher than PA Attribute at **p<0.05

139

Table 76: Proportion of light brand buyer cereal PA response – Would taste great.

Would taste great %Ave Pen

PA Attribute

PA Brand

Brand A 15 53** 34 Brand B 15 70** 53 Brand C 14 81 74 Brand D 13 68** 44 Brand E 13 65* 49 Brand F 13 57 49 Brand G 11 75* 58 Brand H 11 71* 55 Average 68 52

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05

Table 77: Proportion of light brand buyer cereal PA response – Innovative.

Innovative %Ave Pen

PA Attribute

PA Brand

Brand A 15 19** 4 Brand B 15 18** 6 Brand C 14 28 18 Brand D 13 17 11 Brand E 13 33** 13 Brand F 13 23* 11 Brand G 11 25 15 Brand H 11 23 13 Average 23 12

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05.

Table 78: Proportion of light brand buyer cereal PA response – A modern brand.

A modern brand %Ave Pen

PA Attribute

PA Brand

Brand A 15 19 10 Brand B 15 32** 16 Brand C 14 58** 28 Brand D 13 26* 15 Brand E 13 31 27 Brand F 13 36 46 Brand G 11 45* 29 Brand H 11 39 32 Average 36 25

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05.

Page 149: Understanding and Measuring Light Buyer Brand Equity

Table 79: Proportion of light brand buyer cereal PA response – A brand I feel positively about.

Feel positively about

%Ave Pen

PA Attribute

PA Brand

Brand A 15 63** 44 Brand B 15 70** 51 Brand C 14 61* 45 Brand D 13 59** 43 Brand E 13 65** 40 Brand F 13 44 41 Brand G 11 43 31 Brand H 11 46 38 Average 56 42

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05

Table 80: Proportion of light brand buyer cereal PA response – Better quality than other brands.

Better quality %Ave Pen

PA Attribute

PA Brand

Brand A 15 53** 31 Brand B 15 56** 34 Brand C 14 69** 35 Brand D 13 39** 19 Brand E 13 44 31 Brand F 13 21 21 Brand G 11 39** 19 Brand H 11 45* 28 Average 46 28

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05.

Table 81: Proportion of light brand buyer cereal PA response – Appeals to you more than other brands.

Appeals to you more

%Ave Pen

PA Attribute

PA Brand

Brand A 15 43** 26 Brand B 15 42** 27 Brand C 14 64** 38 Brand D 13 32* 18 Brand E 13 44 34 Brand F 13 36** 15 Brand G 11 43** 15 Brand H 11 46 33 Average 44 26

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05.

140

Table 82: Proportion of light brand buyer cereal PA response – Offers something other brands do not.

Offers something others don’t

%Ave Pen

PA Attribute

PA Brand

Brand A 15 40** 9 Brand B 15 19 10 Brand C 14 36 29 Brand D 13 24* 13 Brand E 13 33 21 Brand F 13 23 13 Brand G 11 29 27 Brand H 11 21 12 Average 28 17

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05

Table 83: Proportion of light brand buyer butter/margarine PA response – Spreads easily.

Spreads easily %Ave Pen

PA Attribute

PA Brand

Brand B 11 61 57 Brand C 10 72 80 Brand D 10 35 32 Brand F 10 80** 53 Average 62 56

Statistically significantly higher than PA Brand at **p<0.05.

Table 84: Proportion of light brand buyer butter/margarine PA response – Natural.

Natural %Ave Pen

PA Attribute

PA Brand

Brand B 11 28 21 Brand C 10 16 15 Brand D 10 70** 43 Brand F 10 16 14 Average 32 23

Statistically significantly higher than PA Brand at **p<0.05.

Table 85: Proportion of light brand buyer butter/margarine PA response – Good value for money.

Good value for money

%Ave Pen

PA Attribute

PA Brand

Brand B 11 26 34 Brand C 10 45 48 Brand D 10 30 30 Brand F 10 30 23 Average 33 34

Page 150: Understanding and Measuring Light Buyer Brand Equity

Table 86: Proportion of light brand buyer butter/margarine PA response – Good for the whole family.

Good for the whole family

%Ave Pen

PA Attribute

PA Brand

Brand B 11 54* 36 Brand C 10 53 46 Brand D 10 48 36 Brand F 10 68** 26 Average 56 36

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05.

Table 87: Proportion of light brand buyer butter/margarine PA response – Kids would like it.

Kids would like it %Ave Pen

PA Attribute

PA Brand

Brand B 11 40** 15 Brand C 10 47** 26 Brand D 10 50** 23 Brand F 10 58** 19 Average 49 21

Statistically significantly higher than PA Brand at **p<0.05.

Table 88: Proportion of light brand buyer butter/margarine PA response – Good for cooking/baking.

Good for cooking/baking

%Ave Pen

PA Attribute

PA Brand

Brand B 11 30 26 Brand C 10 22 17 Brand D 10 55 43 Brand F 10 42* 23 Average 37 27 Statistically significantly higher than PA Brand at *p<0.1.

Table 89: Proportion of light brand buyer butter/margarine PA response – Helps control cholesterol.

Helps control cholesterol

%Ave Pen

PA Attribute

PA Brand

Brand B 11 28 17 Brand C 10 12 4 Brand D 10 13 9 Brand F 10 18 21 Average 18 13

141

Table 90: Proportion of light brand buyer butter/margarine PA response – A healthy option.

A healthy option %Ave Pen

PA Attribute

PA Brand

Brand B 11 30 36 Brand C 10 16 22 Brand D 10 20 13 Brand F 10 16 40** Average 20 28

Statistically significantly higher than PA Attribute at **p<0.05.

Table 91: Proportion of light brand buyer butter/margarine PA response – Would taste great.

Would taste great %Ave Pen

PA Attribute

PA Brand

Brand B 11 47 32 Brand C 10 52 52 Brand D 10 80 68 Brand F 10 66** 42 Average 61 49

Statistically significantly higher than PA Brand at **p<0.05.

Table 92: Proportion of light brand buyer butter/margarine PA response – Innovative.

Innovative %Ave Pen

PA Attribute

PA Brand

Brand B 11 19** 6 Brand C 10 28 15 Brand D 10 20 9 Brand F 10 42** 16 Average 27 12

Statistically significantly higher than PA Brand at **p<0.05.

Table 93: Proportion of light brand buyer butter/margarine PA response – A modern brand.

A modern brand %Ave Pen

PA Attribute

PA Brand

Brand B 11 39 30 Brand C 10 50** 28 Brand D 10 25 26 Brand F 10 46* 28 Average 40 28

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05.

Page 151: Understanding and Measuring Light Buyer Brand Equity

Table 94: Proportion of light brand buyer butter/margarine PA response – A brand I feel positively about.

Feel positively about

%Ave Pen

PA Attribute

PA Brand

Brand B 11 51** 30 Brand C 10 40 26 Brand D 10 68** 45 Brand F 10 44* 26 Average 51 32

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05.

Table 95: Proportion of light brand buyer butter/margarine PA response – Better quality than other brands.

Better quality %Ave Pen

PA Attribute

PA Brand

Brand B 11 37* 21 Brand C 10 21 13 Brand D 10 60** 38 Brand F 10 40** 21 Average 39 23

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05.

Table 96: Proportion of light brand buyer butter/margarine PA response – Appeals to you more than other brands.

Appeals to you more

%Ave Pen

PA Attribute

PA Brand

Brand B 11 26* 13 Brand C 10 24 15 Brand D 10 58** 26 Brand F 10 32 21 Average 35 19

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05.

Table 97: Proportion of light brand buyer butter/margarine PA response – Offers something other brands do not.

Offers something others don’t

%Ave Pen

PA Attribute

PA Brand

Brand B 11 18 9 Brand C 10 16 15 Brand D 10 35** 11 Brand F 10 24* 9 Average 23 11

Statistically significantly higher than PA Brand at *p<0.1 **p<0.05

142

Page 152: Understanding and Measuring Light Buyer Brand Equity

Appendix E: FCB: Attribute versus Brand

Prompted Table 98: Proportion of light brand buyer cereal FCB response – Stays crispy in milk.

Stays crispy in milk

%Ave Pen

FCB Brand

FCB Attribute

Brand A 15 31 26 Brand B 15 63 56 Brand C 14 75 71 Brand D 13 67 54 Brand E 13 62 49 Brand F 13 71 63 Brand G 11 63** 44 Brand H 11 63 68 Average 62 54

Statistically significantly higher than FCB Attribute at **p<0.05.

Table 99: Proportion of light brand buyer cereal FCB response – Natural.

Natural %Ave Pen

FCB Brand

FCB Attribute

Brand A 15 94 95 Brand B 15 81 73 Brand C 14 61** 40 Brand D 13 78* 64 Brand E 13 92** 75 Brand F 13 67* 52 Brand G 11 39 27 Brand H 11 44 34 Average 70 57

Statistically significantly higher than FCB Attribute at *p<0.1 **p<0.05.

Table 100: Proportion of light brand buyer cereal FCB response – Good value for money.

Good value for money

%Ave Pen

FCB Brand

FCB Attribute

Brand A 15 76 78 Brand B 15 80 81 Brand C 14 67 65 Brand D 13 82 77 Brand E 13 72 71 Brand F 13 77 68 Brand G 11 63 67 Brand H 11 62 70 Average 72 72

143

Table 101: Proportion of light brand buyer cereal FCB response – Good for the whole family.

Good for the whole family

%Ave Pen

FCB Brand

FCB Attribute

Brand A 15 88 93 Brand B 15 90 90 Brand C 14 80 81 Brand D 13 90 84 Brand E 13 90 88 Brand F 13 92 85 Brand G 11 75 62 Brand H 11 79 80 Average 86 83

Table 102: Proportion of light brand buyer cereal FCB response – Kids would like it.

Kids would like it %Ave Pen

FCB Brand

FCB Attribute

Brand A 15 57 46 Brand B 15 67** 48 Brand C 14 78 75 Brand D 13 87** 96 Brand E 13 72 63 Brand F 13 89 94 Brand G 11 100 97 Brand H 11 97 96 Average 81 77

Statistically significantly higher than FCB Attribute at **p<0.05.

Table 103: Proportion of light brand buyer cereal FCB response – Good for a treat.

Good for a treat %Ave Pen

FCB Brand

FCB Attribute

Brand A 15 54** 37 Brand B 15 60 55 Brand C 14 92** 76 Brand D 13 70 59 Brand E 13 72** 52 Brand F 13 86** 65 Brand G 11 91 87 Brand H 11 90 86 Average 77 65

Statistically significantly higher than FCB Attribute at **p<0.05.

Page 153: Understanding and Measuring Light Buyer Brand Equity

Table 104: Proportion of light brand buyer cereal FCB response – Would give me energy for the day.

Would give me energy for day

%Ave Pen

FCB Brand

FCB Attribute

Brand A 15 89 86 Brand B 15 76 79 Brand C 14 77 78 Brand D 13 73* 58 Brand E 13 86 80 Brand F 13 77 75 Brand G 11 52 52 Brand H 11 67 68 Average 75 72

Statistically significantly higher than FCB Attribute at *p<0.1.

Table 105: Proportion of light brand buyer cereal FCB response – A healthy option.

A healthy option %Ave Pen

FCB Brand

FCB Attribute

Brand A 15 96 94 Brand B 15 87** 70 Brand C 14 56 52 Brand D 13 82** 65 Brand E 13 90 83 Brand F 13 67 58 Brand G 11 29 22 Brand H 11 24 30 Average 66 59

Statistically significantly higher than FCB Attribute at **p<0.05.

Table 106: Proportion of light brand buyer cereal FCB response – Would taste great.

Would taste great %Ave Pen

FCB Brand

FCB Attribute

Brand A 15 86 88 Brand B 15 89 86 Brand C 14 94 98 Brand D 13 87 86 Brand E 13 86 89 Brand F 13 92 89 Brand G 11 91 89 Brand H 11 92 98 Average 89 90

144

Table 107: Proportion of light brand buyer cereal FCB response – Innovative.

Innovative %Ave Pen

FCB Brand

FCB Attribute

Brand A 15 60 48 Brand B 15 50 43 Brand C 14 70 59 Brand D 13 63** 46 Brand E 13 58 49 Brand F 13 73 60 Brand G 11 50 57 Brand H 11 52 52 Average 59 52

Statistically significantly higher than FCB Attribute at **p<0.05.

Table 108: Proportion of light brand buyer cereal FCB response – A modern brand.

A modern brand %Ave Pen

FCB Brand

FCB Attribute

Brand A 15 53 49 Brand B 15 53 51 Brand C 14 75 76 Brand D 13 54 49 Brand E 13 78** 57 Brand F 13 80 83 Brand G 11 75 71 Brand H 11 62 64 Average 66 63

Statistically significantly higher than FCB Attribute at **p<0.05.

Table 109: Proportion of light brand buyer cereal FCB response – A brand I feel positively about.

Feel positively about

%Ave Pen

FCB Brand

FCB Attribute

Brand A 15 89 89 Brand B 15 81 88 Brand C 14 84 84 Brand D 13 81 82 Brand E 13 88 83 Brand F 13 85 86 Brand G 11 77 78 Brand H 11 78 86 Average 83 84

Page 154: Understanding and Measuring Light Buyer Brand Equity

Table 110: Proportion of light brand buyer cereal FCB response – Better quality than other brands.

Better quality %Ave Pen

FCB Brand

FCB Attribute

Brand A 15 80 81 Brand B 15 84 83 Brand C 14 84 88 Brand D 13 78 69 Brand E 13 76 77 Brand F 13 77 77 Brand G 11 75 78 Brand H 11 78 79 Average 79 79

Table 111: Proportion of light brand buyer cereal FCB response – Appeals to you more than other brands.

Appeals to you more

%Ave Pen

FCB Brand

FCB Attribute

Brand A 15 67 70 Brand B 15 71 66 Brand C 14 81 78 Brand D 13 63 62 Brand E 13 68 58 Brand F 13 61 60 Brand G 11 61 71 Brand H 11 71 68 Average 68 67

Table 112: Proportion of light brand buyer cereal FCB response – Offers something other brands do not.

Offers something others don’t

%Ave Pen

FCB Brand

FCB Attribute

Brand A 15 77** 60 Brand B 15 57 53 Brand C 14 81** 65 Brand D 13 67** 43 Brand E 13 74 68 Brand F 13 83* 71 Brand G 11 59 56 Brand H 11 62 61 Average 70 60

Statistically significantly higher than FCB Attribute at *p<0.1 **p<0.05.

145

Table 113: Proportion of light brand buyer butter/margarine FCB response – Spreads easily.

Spreads easily %Ave Pen

FCB Brand

FCB Attribute

Brand B 11 85 95*** Brand C 10 91 94 Brand D 10 53 39 Brand F 10 98* 91 Average 82 80

Statistically significantly higher than FCB Attribute at *p<0.1. Statistically significantly higher than FCB Brand at ***p<0.1.

Table 114: Proportion of light brand buyer butter/margarine FCB response – Natural.

Natural %Ave Pen

FCB Brand

FCB Attribute

Brand B 11 60 65 Brand C 10 53 42 Brand D 10 93 89 Brand F 10 64 49 Average 68 61

Table 115: Proportion of light brand buyer butter/margarine FCB response – Good value for money.

Good value for money

%Ave Pen

FCB Brand

FCB Attribute

Brand B 11 79 73 Brand C 10 77 75 Brand D 10 72 60 Brand F 10 75** 56 Average 76 66

Statistically significantly higher than FCB Attribute at **p<0.05.

Table 116: Proportion of light brand buyer butter/margarine FCB response – Good for the whole family.

Good for the whole family

%Ave Pen

FCB Brand

FCB Attribute

Brand B 11 75 87 Brand C 10 72 81 Brand D 10 93** 76 Brand F 10 87 76 Average 82 80

Statistically significantly higher than FCB Attribute at **p<0.05.

Page 155: Understanding and Measuring Light Buyer Brand Equity

Table 117: Proportion of light brand buyer butter/margarine FCB response – Kids would like it.

Kids would like it %Ave Pen

FCB Brand

FCB Attribute

Brand B 11 68 62 Brand C 10 74 79 Brand D 10 86 79 Brand F 10 84 73 Average 78 73

Table 118: Proportion of light brand buyer butter/margarine FCB response – Good for cooking/baking.

Good for cooking/baking

%Ave Pen

FCB Brand

FCB Attribute

Brand B 11 72 58 Brand C 10 74** 44 Brand D 10 95** 79 Brand F 10 80** 53 Average 80 59

Statistically significantly higher than FCB Attribute at **p<0.05.

Table 119: Proportion of light brand buyer butter/margarine FCB response – Helps control cholesterol.

Helps control cholesterol

%Ave Pen

FCB Brand

FCB Attribute

Brand B 11 55 40 Brand C 10 40 31 Brand D 10 23 15 Brand F 10 69** 29 Average 47 29

Statistically significantly higher than FCB Attribute at **p<0.05.

Table 120: Proportion of light brand buyer butter/margarine FCB response – A healthy option.

A healthy option %Ave Pen

FCB Brand

FCB Attribute

Brand B 11 60* 45 Brand C 10 49 42 Brand D 10 44 40 Brand F 10 66** 31 Average 55 40

Statistically significantly higher than FCB Attribute at *p<0.1 **p<0.05.

146

Table 121: Proportion of light brand buyer butter/margarine FCB response – Would taste great.

Would taste great %Ave Pen

FCB Brand

FCB Attribute

Brand B 11 91 80 Brand C 10 77 77 Brand D 10 95 87 Brand F 10 97** 78 Average 90 80

Statistically significantly higher than FCB Attribute at **p<0.05.

Table 122: Proportion of light brand buyer butter/margarine FCB response – Innovative.

Innovative %Ave Pen

FCB Brand

FCB Attribute

Brand B 11 60 48 Brand C 10 66 54 Brand D 10 42 32 Brand F 10 70 60 Average 60 49

Table 123: Proportion of light brand buyer butter/margarine FCB response – A modern brand.

A modern brand %Ave Pen

FCB Brand

FCB Attribute

Brand B 11 74 60 Brand C 10 85 75 Brand D 10 63** 44 Brand F 10 80 78 Average 75 64

Statistically significantly higher than FCB Attribute at **p<0.05.

Table 124: Proportion of light brand buyer butter/margarine FCB response – A brand I feel positively about.

Feel positively about

%Ave Pen

FCB Brand

FCB Attribute

Brand B 11 83 80 Brand C 10 70 65 Brand D 10 88 82 Brand F 10 92** 78 Average 83 76

Statistically significantly higher than FCB Attribute at **p<0.05.

Page 156: Understanding and Measuring Light Buyer Brand Equity

Table 125: Proportion of light brand buyer butter/margarine FCB response – Better quality than other brands.

Better quality %Ave Pen

FCB Brand

FCB Attribute

Brand B 11 72 73 Brand C 10 55 46 Brand D 10 74 71 Brand F 10 79** 60 Average 70 63

Statistically significantly higher than FCB Attribute at **p<0.05.

Table 126: Proportion of light brand buyer butter/margarine FCB response – Appeals to you more than other brands.

Appeals to you more

%Ave Pen

FCB Brand

FCB Attribute

Brand B 11 60 63 Brand C 10 62 48 Brand D 10 77 69 Brand F 10 74 64 Average 68 61

Table 127: Proportion of light brand buyer butter/margarine FCB response – Offers something other brands do not.

Offers something others don’t

%Ave Pen

FCB Brand

FCB Attribute

Brand B 11 57** 38 Brand C 10 64** 37 Brand D 10 49 40 Brand F 10 64 49 Average 58 41

Statistically significantly higher than FCB Attribute at **p<0.05

147

Page 157: Understanding and Measuring Light Buyer Brand Equity

Appendix F: PA versus FCB Attribute Prompted Table 128: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Stays crispy in milk.

Stays crispy in milk

%Ave Pen

FCB Attribute

PA Attribute

Brand A 15 26** 10 Brand B 15 56** 38 Brand C 14 71** 52 Brand D 13 54** 33 Brand E 13 49** 23 Brand F 13 63** 30 Brand G 11 44** 25 Brand H 11 68** 36 Average 54 31

Statistically significantly higher than PA Attribute at **p<0.05.

Table 129: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Natural.

Natural %Ave Pen

FCB Attribute

PA Attribute

Brand A 15 95** 73 Brand B 15 73** 47 Brand C 14 40** 9 Brand D 13 64** 37 Brand E 13 75** 46 Brand F 13 52** 11 Brand G 11 27** 6 Brand H 11 34** 7 Average 57 29

Statistically significantly higher than PA Attribute at **p<0.05.

148

Table 130: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Good value for money.

Good value for money

%Ave Pen

FCB Attribute

PA Attribute

Brand A 15 78 74 Brand B 15 81* 70 Brand C 14 65* 52 Brand D 13 77 68 Brand E 13 71 71 Brand F 13 68** 46 Brand G 11 67** 47 Brand H 11 70** 52 Average 72 60

Statistically significantly higher than PA Attribute at *p<0.1 **p<0.05.

Table 131: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Good for the whole family.

Good for the whole family

%Ave Pen

FCB Attribute

PA Attribute

Brand A 15 93** 73 Brand B 15 90** 71 Brand C 14 81** 44 Brand D 13 84** 61 Brand E 13 88** 63 Brand F 13 85** 48 Brand G 11 62** 33 Brand H 11 80** 46 Average 83 55

Statistically significantly higher than PA Attribute at **p<0.05.

Table 132: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Kids would like it.

Kids would like it %Ave Pen

FCB Attribute

PA Attribute

Brand A 15 46** 23 Brand B 15 48** 25 Brand C 14 75** 31 Brand D 13 96** 71 Brand E 13 63** 40 Brand F 13 94** 74 Brand G 11 97 92 Brand H 11 96** 64 Average 77 52

Statistically significantly higher than PA Attribute at **p<0.05.

Page 158: Understanding and Measuring Light Buyer Brand Equity

Table 133: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Good for a treat.

Good for a treat %Ave Pen

FCB Attribute

PA Attribute

Brand A 15 37** 9 Brand B 15 55** 22 Brand C 14 76** 53 Brand D 13 59** 26 Brand E 13 52** 17 Brand F 13 65** 23 Brand G 11 87** 69 Brand H 11 86** 54 Average 65 34

Statistically significantly higher than PA Attribute at **p<0.05.

Table 134: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Would give me energy for the day.

Would give me energy for day

%Ave Pen

FCB Attribute

PA Attribute

Brand A 15 86 77 Brand B 15 79** 48 Brand C 14 78** 47 Brand D 13 58** 26 Brand E 13 80** 60 Brand F 13 75** 34 Brand G 11 52** 22 Brand H 11 68** 32 Average 72 43

Statistically significantly higher than PA Attribute at **p<0.05.

Table 135: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – A healthy option.

A healthy option %Ave Pen

FCB Attribute

PA Attribute

Brand A 15 94** 77 Brand B 15 70** 53 Brand C 14 52** 17 Brand D 13 65** 41 Brand E 13 83** 58 Brand F 13 58** 20 Brand G 11 22* 10 Brand H 11 30** 4 Average 59 35

Statistically significantly higher than PA Attribute at *p<0.1 **p<0.05.

149

Table 136: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Would taste great.

Would taste great %Ave Pen

FCB Attribute

PA Attribute

Brand A 15 88** 53 Brand B 15 86** 70 Brand C 14 98** 81 Brand D 13 86** 68 Brand E 13 89** 65 Brand F 13 89** 57 Brand G 11 89** 75 Brand H 11 98** 71 Average 90 68

Statistically significantly higher than PA Attribute at **p<0.05.

Table 137: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Innovative.

Innovative %Ave Pen

FCB Attribute

PA Attribute

Brand A 15 48** 19 Brand B 15 43** 18 Brand C 14 59** 28 Brand D 13 46** 17 Brand E 13 49* 33 Brand F 13 60** 23 Brand G 11 57** 25 Brand H 11 52** 23 Average 52 23

Statistically significantly higher than PA Attribute at *p<0.1 **p<0.05.

Table 138: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – A modern brand.

A modern brand %Ave Pen

FCB Attribute

PA Attribute

Brand A 15 49** 19 Brand B 15 51** 32 Brand C 14 76** 58 Brand D 13 49** 26 Brand E 13 57** 31 Brand F 13 83** 36 Brand G 11 71** 45 Brand H 11 64** 39 Average 63 36

Statistically significantly higher than PA Attribute at **p<0.05.

Page 159: Understanding and Measuring Light Buyer Brand Equity

Table 139: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – A brand I feel positively about.

Feel positively about

%Ave Pen

FCB Attribute

PA Attribute

Brand A 15 89** 63 Brand B 15 88** 70 Brand C 14 84** 61 Brand D 13 82** 59 Brand E 13 83** 65 Brand F 13 86** 44 Brand G 11 78** 43 Brand H 11 86** 46 Average 84 56

Statistically significantly higher than PA Attribute at **p<0.05.

Table 140: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Better quality than other brands.

Better quality %Ave Pen

FCB Attribute

PA Attribute

Brand A 15 81** 53 Brand B 15 83** 56 Brand C 14 88** 69 Brand D 13 69** 39 Brand E 13 77** 44 Brand F 13 77** 21 Brand G 11 78** 39 Brand H 11 79** 45 Average 79 46

Statistically significantly higher than PA Attribute at **p<0.05.

Table 141: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Appeals to you more than other brands.

Appeals to you more

%Ave Pen

FCB Attribute

PA Attribute

Brand A 15 70** 43 Brand B 15 66** 42 Brand C 14 78* 64 Brand D 13 62** 32 Brand E 13 58 44 Brand F 13 60** 36 Brand G 11 71** 43 Brand H 11 68** 46 Average 67 44

Statistically significantly higher than PA Attribute at *p<0.1 **p<0.05.

150

Table 142: Proportion of light brand buyer cereal response, FCB and PA attribute prompted – Offers something other brands do not.

Offers something others don’t

%Ave Pen

FCB Attribute

PA Attribute

Brand A 15 60** 40 Brand B 15 53** 19 Brand C 14 65** 36 Brand D 13 43** 24 Brand E 13 68** 33 Brand F 13 71** 23 Brand G 11 56** 29 Brand H 11 61** 21 Average 60 28

Statistically significantly higher than PA Attribute at **p<0.05.

Table 143: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Spreads easily.

Spreads easily %Ave Pen

FCB Attribute

PA Attribute

Brand B 11 95** 61 Brand C 10 94** 72 Brand D 10 39 35 Brand F 10 91 80 Average 80 62

Statistically significantly higher than PA Attribute at **p<0.05.

Table 144: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Natural.

Natural %Ave Pen

FCB Attribute

PA Attribute

Brand B 11 65** 28 Brand C 10 42** 16 Brand D 10 89** 70 Brand F 10 49** 16 Average 61 32

Statistically significantly higher than PA Attribute at **p<0.05.

Page 160: Understanding and Measuring Light Buyer Brand Equity

Table 145: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Good value for money.

Good value for money

%Ave Pen

FCB Attribute

PA Attribute

Brand B 11 73** 26 Brand C 10 75** 45 Brand D 10 60** 30 Brand F 10 56** 30 Average 66 33

Statistically significantly higher than PA Attribute at **p<0.05.

Table 146: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Good for the whole family.

Good for the whole family

%Ave Pen

FCB Attribute

PA Attribute

Brand B 11 87** 54 Brand C 10 81** 53 Brand D 10 76** 48 Brand F 10 76 68 Average 80 56

Statistically significantly higher than PA Attribute at **p<0.05.

Table 147: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Kids would like it.

Kids would like it %Ave Pen

FCB Attribute

PA Attribute

Brand B 11 62** 40 Brand C 10 79** 47 Brand D 10 79** 50 Brand F 10 73 58 Average 73 49

Statistically significantly higher than PA Attribute at **p<0.05.

Table 148: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Good for cooking/baking.

Good for cooking/baking

%Ave Pen

FCB Attribute

PA Attribute

Brand B 11 58** 30 Brand C 10 44** 22 Brand D 10 79** 55 Brand F 10 53 42 Average 59 37

Statistically significantly higher than PA Attribute at **p<0.05.

151

Table 149: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Helps control cholesterol.

Helps control cholesterol

%Ave Pen

FCB Attribute

PA Attribute

Brand B 11 40 28 Brand C 10 31** 12 Brand D 10 15 13 Brand F 10 29 18 Average 29 18

Statistically significantly higher than PA Attribute at **p<0.05.

Table 150: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – A healthy option.

A healthy option %Ave Pen

FCB Attribute

PA Attribute

Brand B 11 45* 30 Brand C 10 42** 16 Brand D 10 40** 20 Brand F 10 31* 16 Average 40 20

Statistically significantly higher than PA Attribute at *p<0.1 **p<0.05.

Table 151: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Would taste great.

Would taste great %Ave Pen

FCB Attribute

PA Attribute

Brand B 11 80** 47 Brand C 10 77** 52 Brand D 10 87 80 Brand F 10 78 66 Average 80 61

Statistically significantly higher than PA Attribute at **p<0.05.

Table 152: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Innovative.

Innovative %Ave Pen

FCB Attribute

PA Attribute

Brand B 11 48** 19 Brand C 10 54** 28 Brand D 10 32 20 Brand F 10 60* 42 Average 49 27

Statistically significantly higher than PA Attribute at *p<0.1 **p<0.05.

Page 161: Understanding and Measuring Light Buyer Brand Equity

Table 153: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – A modern brand.

A modern brand %Ave Pen

FCB Attribute

PA Attribute

Brand B 11 60** 39 Brand C 10 75** 50 Brand D 10 44* 25 Brand F 10 78** 46 Average 64 40

Statistically significantly higher than PA Attribute at *p<0.1 **p<0.05.

Table 154: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – A brand I feel positively about.

Feel positively about

%Ave Pen

FCB Attribute

PA Attribute

Brand B 11 80** 51 Brand C 10 65** 40 Brand D 10 82* 68 Brand F 10 78** 44 Average 76 51

Statistically significantly higher than PA Attribute at *p<0.1 **p<0.05.

Table 155: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Better quality than other brands.

Better quality %Ave Pen

FCB Attribute

PA Attribute

Brand B 11 73** 37 Brand C 10 46** 21 Brand D 10 71 60 Brand F 10 60** 40 Average 63 39

Statistically significantly higher than PA Attribute at **p<0.05.

Table 156: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Appeals to you more than other brands.

Appeals to you more

%Ave Pen

FCB Attribute

PA Attribute

Brand B 11 63** 26 Brand C 10 48** 24 Brand D 10 69 58 Brand F 10 64** 32 Average 61 35

Statistically significantly higher than PA Attribute at **p<0.05.

152

Table 157: Proportion of light brand buyer butter/margarine response, FCB and PA attribute prompted – Offers something other brands do not.

Offers something others don’t

%Ave Pen

FCB Attribute

PA Attribute

Brand B 11 38** 18 Brand C 10 37** 16 Brand D 10 40 35 Brand F 10 49** 24 Average 41 23

Statistically significantly higher than PA Attribute at **p<0.05

Page 162: Understanding and Measuring Light Buyer Brand Equity

Appendix G: PA versus FCB Brand Prompted Table 158: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Stays crispy in milk.

Stays crispy in milk

%Ave Pen FCB Brand PA Brand

Brand A 15 31** 6 Brand B 15 63** 24 Brand C 14 75** 34 Brand D 13 67** 25 Brand E 13 62** 24 Brand F 13 71** 46 Brand G 11 63** 37 Brand H 11 63** 35 Average 62 29

Statistically significantly higher than PA Brand at **p<0.05.

Table 159: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Natural.

Natural %Ave Pen FCB Brand PA Brand

Brand A 15 94** 56 Brand B 15 81** 43 Brand C 14 61** 17 Brand D 13 78** 24 Brand E 13 92** 43 Brand F 13 67** 21 Brand G 11 39** 6 Brand H 11 44** 10 Average 70 27

Statistically significantly higher than PA Brand at **p<0.05.

153

Table 160: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Good value for money.

Good value for money

%Ave Pen FCB Brand PA Brand

Brand A 15 76** 46 Brand B 15 80** 44 Brand C 14 67** 17 Brand D 13 82** 50 Brand E 13 72** 28 Brand F 13 77** 31 Brand G 11 63** 21 Brand H 11 62** 32 Average 72 34

Statistically significantly higher than PA Brand at **p<0.05.

Table 161: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Good for the whole family.

Good for the whole family

%Ave Pen FCB Brand PA Brand

Brand A 15 88** 54 Brand B 15 90** 59 Brand C 14 80** 34 Brand D 13 90** 54 Brand E 13 90** 57 Brand F 13 92** 59 Brand G 11 75** 35 Brand H 11 79** 42 Average 86 49

Statistically significantly higher than PA Brand at **p<0.05.

Table 162: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Kids would like it.

Kids would like it %Ave Pen FCB Brand PA Brand

Brand A 15 57** 28 Brand B 15 67** 20 Brand C 14 78** 34 Brand D 13 87** 57 Brand E 13 72** 43 Brand F 13 89** 57 Brand G 11 100** 75 Brand H 11 97** 60 Average 81 47

Statistically significantly higher than PA Brand at **p<0.05

Page 163: Understanding and Measuring Light Buyer Brand Equity

Table 163: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Good for a treat.

Good for a treat %Ave Pen FCB Brand PA Brand

Brand A 15 54** 10 Brand B 15 60** 26 Brand C 14 92** 51 Brand D 13 70** 26 Brand E 13 72** 31 Brand F 13 86** 26 Brand G 11 91** 63 Brand H 11 90** 50 Average 77 36

Statistically significantly higher than PA Brand at **p<0.05.

Table 164: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Would give me energy for the day.

Would give me energy for day

%Ave Pen FCB Brand PA Brand

Brand A 15 89** 53 Brand B 15 76** 31 Brand C 14 77** 38 Brand D 13 73** 21 Brand E 13 86** 42 Brand F 13 77** 25 Brand G 11 52** 15 Brand H 11 67** 27 Average 75 32

Statistically significantly higher than PA Brand at **p<0.05.

Table 165: Proportion of light brand buyer cereal response, FCB and PA brand prompted – A healthy option.

A healthy option %Ave Pen FCB Brand PA Brand

Brand A 15 96** 71 Brand B 15 87** 47 Brand C 14 56** 22 Brand D 13 82** 36 Brand E 13 90** 60 Brand F 13 67** 39 Brand G 11 29** 10 Brand H 11 24** 10 Average 66 37

Statistically significantly higher than PA Brand at **p<0.05.

154

Table 166: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Would taste great.

Would taste great %Ave Pen FCB Brand PA Brand

Brand A 15 86** 34 Brand B 15 89** 53 Brand C 14 94** 74 Brand D 13 87** 44 Brand E 13 86** 49 Brand F 13 92** 49 Brand G 11 91** 58 Brand H 11 92** 55 Average 89 52

Statistically significantly higher than PA Brand at **p<0.05.

Table 167: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Innovative.

Innovative %Ave Pen FCB Brand PA Brand

Brand A 15 60** 4 Brand B 15 50** 6 Brand C 14 70** 18 Brand D 13 63** 11 Brand E 13 58** 13 Brand F 13 73** 11 Brand G 11 50** 15 Brand H 11 52** 13 Average 59 12

Statistically significantly higher than PA Brand at **p<0.05.

Table 168: Proportion of light brand buyer cereal response, FCB and PA brand prompted – A modern brand.

A modern brand %Ave Pen FCB Brand PA Brand

Brand A 15 53** 10 Brand B 15 53** 16 Brand C 14 75** 28 Brand D 13 54** 15 Brand E 13 78** 27 Brand F 13 80** 46 Brand G 11 75** 29 Brand H 11 62** 32 Average 66 25

Statistically significantly higher than PA Brand at **p<0.05.

Page 164: Understanding and Measuring Light Buyer Brand Equity

Table 169: Proportion of light brand buyer cereal response, FCB and PA brand prompted – A brand I feel positively about.

Feel positively about

%Ave Pen FCB Brand PA Brand

Brand A 15 89** 44 Brand B 15 81** 51 Brand C 14 84** 45 Brand D 13 81** 43 Brand E 13 88** 40 Brand F 13 85** 41 Brand G 11 77** 31 Brand H 11 78** 38 Average 83 42

Statistically significantly higher than PA Brand at **p<0.05.

Table 170: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Better quality than other brands.

Better quality %Ave Pen FCB Brand PA Brand

Brand A 15 80** 31 Brand B 15 84** 34 Brand C 14 84** 35 Brand D 13 78** 19 Brand E 13 76** 31 Brand F 13 77** 21 Brand G 11 75** 19 Brand H 11 78** 28 Average 79 28

Statistically significantly higher than PA Brand at **p<0.05.

Table 171: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Appeals to you more than other brands.

Appeals to you more

%Ave Pen FCB Brand PA Brand

Brand A 15 67** 26 Brand B 15 71** 27 Brand C 14 81** 38 Brand D 13 63** 18 Brand E 13 68** 34 Brand F 13 61** 15 Brand G 11 61** 15 Brand H 11 71** 33 Average 68 26

Statistically significantly higher than PA Brand at **p<0.05.

155

Table 172: Proportion of light brand buyer cereal response, FCB and PA brand prompted – Offers something other brands do not.

Offers something others don’t

%Ave Pen FCB Brand PA Brand

Brand A 15 77** 9 Brand B 15 57** 10 Brand C 14 81** 29 Brand D 13 67** 13 Brand E 13 74** 21 Brand F 13 83** 13 Brand G 11 59** 27 Brand H 11 62** 12 Average 70 17

Statistically significantly higher than PA Brand at **p<0.05.

Table 173: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Spreads easily.

Spreads easily %Ave Pen FCB Brand PA Brand

Brand B 11 85** 57 Brand C 10 91 80 Brand D 10 53** 32 Brand F 10 98** 53 Average 82 56

Statistically significantly higher than PA Brand at **p<0.05.

Table 174: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Natural.

Natural %Ave Pen FCB Brand PA Brand

Brand B 11 60** 21 Brand C 10 53** 15 Brand D 10 93** 43 Brand F 10 64** 14 Average 68 23

Statistically significantly higher than PA Brand at **p<0.05.

Page 165: Understanding and Measuring Light Buyer Brand Equity

Table 175: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Good value for money.

Good value for money

%Ave Pen FCB Brand PA Brand

Brand B 11 79** 34 Brand C 10 77** 48 Brand D 10 72** 30 Brand F 10 75** 23 Average 76 34

Statistically significantly higher than PA Brand at **p<0.05.

Table 176: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Good for the whole family.

Good for the whole family

%Ave Pen FCB Brand PA Brand

Brand B 11 75** 36 Brand C 10 72** 46 Brand D 10 93** 36 Brand F 10 87** 26 Average 82 36

Statistically significantly higher than PA Brand at **p<0.05.

Table 177: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Kids would like it.

Kids would like it %Ave Pen FCB Brand PA Brand

Brand B 11 68** 15 Brand C 10 74** 26 Brand D 10 86** 23 Brand F 10 84** 19 Average 78 21

Statistically significantly higher than PA Brand at **p<0.05.

Table 178: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Good for cooking/baking.

Good for cooking/baking

%Ave Pen

FCB Brand PA Brand

Brand B 11 72** 26 Brand C 10 74** 17 Brand D 10 95** 43 Brand F 10 80** 23 Average 80 27

Statistically significantly higher than PA Brand at **p<0.05.

156

Table 179: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Helps control cholesterol.

Helps control cholesterol

%Ave Pen FCB Brand PA Brand

Brand B 11 55** 17 Brand C 10 40** 4 Brand D 10 23** 9 Brand F 10 69** 21 Average 47 13

Statistically significantly higher than PA Brand at **p<0.05.

Table 180: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – A healthy option.

A healthy option %Ave Pen FCB Brand PA Brand

Brand B 11 60** 36 Brand C 10 49** 22 Brand D 10 44** 13 Brand F 10 66** 40 Average 55 28

Statistically significantly higher than PA Brand at **p<0.05.

Table 181: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Would taste great.

Would taste great %Ave Pen FCB Brand PA Brand

Brand B 11 91** 32 Brand C 10 77** 52 Brand D 10 95** 68 Brand F 10 97** 42 Average 90 49

Statistically significantly higher than PA Brand at **p<0.05.

Table 182: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Innovative.

Innovative %Ave Pen FCB Brand PA Brand

Brand B 11 60** 6 Brand C 10 66** 15 Brand D 10 42** 9 Brand F 10 70** 16 Average 60 12

Statistically significantly higher than PA Brand at **p<0.05.

Page 166: Understanding and Measuring Light Buyer Brand Equity

Table 183: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – A modern brand.

A modern brand %Ave Pen FCB Brand PA Brand

Brand B 11 74** 30 Brand C 10 85** 28 Brand D 10 63** 26 Brand F 10 80** 28 Average 75 28

Statistically significantly higher than PA Brand at **p<0.05.

Table 184: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – A brand I feel positively about.

Feel positively about

%Ave Pen FCB Brand PA Brand

Brand B 11 83** 30 Brand C 10 70** 26 Brand D 10 88** 45 Brand F 10 92** 26 Average 83 32

Statistically significantly higher than PA Brand at **p<0.05.

Table 185: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Better quality than other brands.

Better quality %Ave Pen FCB Brand PA Brand

Brand B 11 72** 21 Brand C 10 55** 13 Brand D 10 74** 38 Brand F 10 79** 21 Average 70 23

Statistically significantly higher than PA Brand at **p<0.05.

Table 186: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Appeals to you more than other brands.

Appeals to you more

%Ave Pen FCB Brand PA Brand

Brand B 11 60** 13 Brand C 10 62** 15 Brand D 10 77** 26 Brand F 10 74** 21 Average 68 19

Statistically significantly higher than PA Brand at **p<0.05.

157

Table 187: Proportion of light brand buyer butter/margarine response, FCB and PA brand prompted – Offers something other brands do not.

Offers something others don’t

%Ave Pen FCB Brand PA Brand

Brand B 11 57** 9 Brand C 10 64** 15 Brand D 10 49** 11 Brand F 10 64** 9 Average 58 11

Statistically significantly higher than PA Brand at **p<0.05

Page 167: Understanding and Measuring Light Buyer Brand Equity

Appendix H: Non, Light and Heavier Buyer Image Response Table 188: Proportion of non, light and heavier buyer cereal response, PA attribute prompted – Brand A.

Brand A (PA Attribute) % Non-Buyer

(n=202)

% Light Buyer (n=70)

% Heavier Buyer

(n=232) Stays crispy in milk 7 10 13 Natural 55 73** 85**** Good value for money 32 74** 79 Good for the whole family 54 73** 89**** Kids would like it 13 23** 28 Good for a treat 7 9 25**** Would give me energy for the day 50 77** 81 A healthy option 60 77** 91**** Would taste great 20 53** 64*** Innovative 10 19* 29*** A modern brand 12 19 32**** A brand I feel positively about 28 63** 82**** Better quality than other brands 24 53** 71**** Appeals to you more 12 43** 61**** Offers something other brands do not 25 40** 49

Average 27 47** 59**** Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

158

Table 189: Proportion of non, light and heavier buyer cereal response, PA brand prompted – Brand A.

Brand A (PA Brand) % Non-Buyer

(n=194)

% Light Buyer (n=68)

% Heavier Buyer

(n=242) Stays crispy in milk 5 6 11 Natural 48 56 67*** Good value for money 23 46** 50 Good for the whole family 35 54** 73**** Kids would like it 14 28** 31 Good for a treat 9 10 20*** Would give me energy for the day 33 53** 56 A healthy option 56 71** 79 Would taste great 18 34** 48 Innovative 5 4 13**** A modern brand 6 10 21**** A brand I feel positively about 21 44** 59**** Better quality than other brands 14 31** 44*** Appeals to you more 6 26** 39**** Offers something other brands do not 15 9 24****

Average 21 32** 42 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 190: Proportion of non, light and heavier buyer cereal response, FCB attribute prompted – Brand A.

Brand A (FCB Attribute) % Non-Buyer

(n=171)

% Light Buyer (n=81)

% Heavier Buyer

(n=250) Stays crispy in milk 20 26 20 Natural 81 95** 91 Good value for money 63 78** 84 Good for the whole family 84 93** 96 Kids would like it 37 46 58**** Good for a treat 29 37 47 Would give me energy for the day 76 86* 92 A healthy option 89 94 96 Would taste great 50 88** 93 Innovative 30 48** 54 A modern brand 42 49 49 A brand I feel positively about 57 89** 93 Better quality than other brands 62 81** 88 Appeals to you more 28 70** 82**** Offers something other brands do not 48 60* 72****

Average 53 69** 74 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Page 168: Understanding and Measuring Light Buyer Brand Equity

Table 191: Proportion of non, light and heavier buyer cereal response, FCB brand prompted – Brand A.

Brand A (FCB Brand) % Non-Buyer

(n=199)

% Light Buyer (n=83)

% Heavier Buyer

(n=221) Stays crispy in milk 24 31 38 Natural 83 94** 95 Good value for money 69 76 87**** Good for the whole family 77 88** 96**** Kids would like it 38 57** 71**** Good for a treat 28 54** 67**** Would give me energy for the day 75 89** 91 A healthy option 86 96** 96 Would taste great 46 86** 89 Innovative 37 60** 63 A modern brand 39 53** 59 A brand I feel positively about 60 89** 96**** Better quality than other brands 60 80** 88*** Appeals to you more 39 67** 81**** Offers something other brands do not 62 77** 80

Average 55 73** 80 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 192: Proportion of non, light and heavier buyer cereal response, PA Attribute prompted – Brand B.

Brand B (PA Attribute) % Non-Buyer

(n=264)

% Light Buyer (n=73)

% Heavier Buyer

(n=167) Stays crispy in milk 22 38** 41 Natural 33 47** 57 Good value for money 35 70** 79 Good for the whole family 48 71** 84**** Kids would like it 12 25** 29 Good for a treat 7 22** 28 Would give me energy for the day 20 48** 52 A healthy option 25 53** 56 Would taste great 28 70** 75 Innovative 7 18** 26 A modern brand 9 32** 38 A brand I feel positively about 27 70** 74 Better quality than other brands 27 56** 76**** Appeals to you more 13 42** 60**** Offers something other brands do not 6 19** 35****

Average 21 45** 54 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

159

Table 193: Proportion of non, light and heavier buyer cereal response, PA Brand prompted – Brand B.

Brand B (PA Brand) % Non-Buyer

(n=240)

% Light Buyer (n=70)

% Heavier Buyer

(n=194) Stays crispy in milk 16 24* 37*** Natural 32 43* 48 Good value for money 23 44** 56*** Good for the whole family 45 59** 73**** Kids would like it 16 20 36**** Good for a treat 9 26** 29 Would give me energy for the day 18 31** 42 A healthy option 37 47 57 Would taste great 23 53** 56 Innovative 4 6 13*** A modern brand 11 16 23 A brand I feel positively about 29 51** 62 Better quality than other brands 23 34* 51**** Appeals to you more 12 27** 42**** Offers something other brands do not 5 10 21****

Average 20 33** 43 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 194: Proportion of non, light and heavier buyer cereal response, FCB Attribute prompted – Brand B.

Brand B (FCB Attribute) % Non-Buyer

(n=238)

% Light Buyer (n=80)

% Heavier Buyer

(n=184) Stays crispy in milk 37 56** 56 Natural 70 73 85**** Good value for money 57 81** 85 Good for the whole family 81 90* 98**** Kids would like it 45 48 68**** Good for a treat 27 55** 54 Would give me energy for the day 56 79** 77 A healthy option 69 70 88**** Would taste great 69 86** 95**** Innovative 28 43** 48 A modern brand 35 51** 53 A brand I feel positively about 65 88** 92 Better quality than other brands 60 83** 90*** Appeals to you more 38 66** 82**** Offers something other brands do not 32 53** 57

Average 51 68** 75 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Page 169: Understanding and Measuring Light Buyer Brand Equity

Table 195: Proportion of non, light and heavier buyer cereal response, FCB Brand prompted – Brand B.

Brand B (FCB Brand) % Non-Buyer

(n=246)

% Light Buyer (n=70)

% Heavier Buyer

(n=187) Stays crispy in milk 42 63** 59 Natural 76 81 92**** Good value for money 60 80** 88*** Good for the whole family 80 90** 97**** Kids would like it 52 67** 76 Good for a treat 46 60** 72*** Would give me energy for the day 59 76** 84 A healthy option 73 87** 93 Would taste great 63 89** 93 Innovative 33 50** 61*** A modern brand 34 53** 65*** A brand I feel positively about 62 81** 95**** Better quality than other brands 63 84** 93**** Appeals to you more 39 71** 87**** Offers something other brands do not 38 57** 75****

Average 55 73** 82*** Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 196: Proportion of non, light and heavier buyer cereal response, PA attribute prompted – Brand C.

Brand C (PA Attribute) % Non-Buyer

(n=292)

% Light Buyer (n=62)

% Heavier Buyer

(n=148) Stays crispy in milk 32 52** 57 Natural 8 9 16 Good value for money 15 52** 67**** Good for the whole family 23 44** 63**** Kids would like it 22 31* 35 Good for a treat 34 53** 60 Would give me energy for the day 21 47** 56 A healthy option 12 17 24 Would taste great 47 81** 83 Innovative 21 28 44**** A modern brand 35 58** 57 A brand I feel positively about 22 61** 74**** Better quality than other brands 22 69** 67 Appeals to you more 19 64** 69 Offers something other brands do not 17 36** 46

Average 23 47** 55 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

160

Table 197: Proportion of non, light and heavier buyer cereal response, PA brand prompted – Brand C.

Brand C (PA Brand) % Non-Buyer

(n=301)

% Light Buyer (n=65)

% Heavier Buyer

(n=138) Stays crispy in milk 21 34** 46*** Natural 11 17 31**** Good value for money 10 17* 43**** Good for the whole family 29 34 59**** Kids would like it 19 34** 40 Good for a treat 24 51** 49 Would give me energy for the day 17 38** 41 A healthy option 16 22 36**** Would taste great 35 74** 74 Innovative 9 18** 22 A modern brand 25 28 46**** A brand I feel positively about 16 45** 60**** Better quality than other brands 15 35** 51**** Appeals to you more 9 38** 53**** Offers something other brands do not 14 29** 34

Average 18 34** 46*** Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 198: Proportion of non, light and heavier buyer cereal response, FCB attribute prompted – Brand C.

Brand C (FCB Attribute) % Non-Buyer

(n=290)

% Light Buyer (n=83)

% Heavier Buyer

(n=129) Stays crispy in milk 58 71** 72 Natural 34 40 52*** Good value for money 39 65** 75 Good for the whole family 52 81** 81 Kids would like it 53 75** 78 Good for a treat 56 76** 89**** Would give me energy for the day 54 78** 77 A healthy option 31 52** 53 Would taste great 67 98** 95 Innovative 47 59** 66 A modern brand 66 76* 81 A brand I feel positively about 53 84** 91 Better quality than other brands 56 88** 92 Appeals to you more 37 78** 85 Offers something other brands do not 48 65** 70

Average 50 72** 77 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Page 170: Understanding and Measuring Light Buyer Brand Equity

Table 199: Proportion of non, light and heavier buyer cereal response, FCB brand prompted – Brand C.

Brand C (FCB Brand) % Non-Buyer

(n=276)

% Light Buyer (n=79)

% Heavier Buyer

(n=148) Stays crispy in milk 67 75 79 Natural 52 61 74**** Good value for money 46 67** 82**** Good for the whole family 61 80** 87 Kids would like it 63 78** 84 Good for a treat 63 92** 93 Would give me energy for the day 57 77** 86*** A healthy option 45 56* 68*** Would taste great 70 94** 97 Innovative 57 70** 80*** A modern brand 67 75 86**** A brand I feel positively about 55 84** 91 Better quality than other brands 61 84** 91*** Appeals to you more 43 81** 89*** Offers something other brands do not 59 81** 84

Average 58 77** 85 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 200: Proportion of non, light and heavier buyer cereal response, PA Attribute prompted – Brand D.

Brand D (PA Attribute) % Non-Buyer

(n=342)

% Light Buyer (n=76)

% Heavier Buyer (n=86)

Stays crispy in milk 19 33** 36 Natural 19 37** 41 Good value for money 26 68** 77 Good for the whole family 26 61** 70 Kids would like it 61 71 67 Good for a treat 10 26** 40*** Would give me energy for the day 13 26** 29 A healthy option 14 41** 44 Would taste great 25 68** 76 Innovative 9 17** 33**** A modern brand 14 26** 27 A brand I feel positively about 15 59** 65 Better quality than other brands 12 39** 49 Appeals to you more 8 32** 48**** Offers something other brands do not 8 24** 30

Average 19 42** 49 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

161

Table 201: Proportion of non, light and heavier buyer cereal response, PA Brand prompted – Brand D.

Brand D (PA Brand) % Non-Buyer

(n=309)

% Light Buyer (n=72)

% Heavier Buyer

(n=123) Stays crispy in milk 16 25* 45**** Natural 22 24 39**** Good value for money 20 50** 46 Good for the whole family 29 54** 61 Kids would like it 62 57 66 Good for a treat 13 26** 33 Would give me energy for the day 9 21** 32*** A healthy option 18 36** 46 Would taste great 21 44** 54 Innovative 5 11** 16 A modern brand 10 15 29**** A brand I feel positively about 18 43** 55*** Better quality than other brands 10 19** 33**** Appeals to you more 9 18** 30 Offers something other brands do not 6 13* 20

Average 18 30** 40 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 202: Proportion of non, light and heavier buyer cereal response, FCB Attribute prompted – Brand D.

Brand D (FCB Attribute) % Non-Buyer

(n=306)

% Light Buyer (n=74)

% Heavier Buyer

(n=122) Stays crispy in milk 39 54** 61 Natural 50 64** 77**** Good value for money 49 77** 81 Good for the whole family 65 84** 95**** Kids would like it 87 96** 93 Good for a treat 42 59** 66 Would give me energy for the day 39 58** 60 A healthy option 45 65** 80**** Would taste great 65 86** 93 Innovative 30 46** 56 A modern brand 37 49* 66**** A brand I feel positively about 43 82** 87 Better quality than other brands 45 69** 79 Appeals to you more 26 62** 70 Offers something other brands do not 32 43* 59****

Average 46 66** 75 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Page 171: Understanding and Measuring Light Buyer Brand Equity

Table 203: Proportion of non, light and heavier buyer cereal response, FCB Brand prompted – Brand D.

Brand D (FCB Brand) % Non-Buyer

(n=312)

% Light Buyer (n=67)

% Heavier Buyer

(n=124) Stays crispy in milk 51 67** 58 Natural 66 78* 87*** Good value for money 60 82** 85 Good for the whole family 71 90** 94 Kids would like it 89 87 94*** Good for a treat 58 70* 79 Would give me energy for the day 50 73** 70 A healthy option 61 82** 86 Would taste great 67 87** 95**** Innovative 47 63** 60 A modern brand 45 54 65 A brand I feel positively about 58 81** 92**** Better quality than other brands 50 78** 83 Appeals to you more 38 63** 85**** Offers something other brands do not 48 67** 69

Average 57 75** 80 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 204: Proportion of non, light and heavier buyer cereal response, PA Attribute prompted – Brand E.

Brand E (PA Attribute) % Non-Buyer

(n=339)

% Light Buyer (n=48)

% Heavier Buyer

(n=117) Stays crispy in milk 13 23* 31 Natural 34 46 52 Good value for money 19 71** 70 Good for the whole family 37 63** 76*** Kids would like it 19 40** 32 Good for a treat 6 17** 21 Would give me energy for the day 31 60** 66 A healthy option 36 58** 62 Would taste great 18 65** 67 Innovative 13 33** 38 A modern brand 15 31** 32 A brand I feel positively about 18 65** 74 Better quality than other brands 12 44** 53 Appeals to you more 6 44** 59*** Offers something other brands do not 13 33** 38

Average 19 46** 51 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

162

Table 205: Proportion of non, light and heavier buyer cereal response, PA Brand prompted – Brand E.

Brand E (PA Brand % Non-Buyer

(n=326)

% Light Buyer (n=67)

% Heavier Buyer

(n=111) Stays crispy in milk 13 24** 23 Natural 39 43 55*** Good value for money 13 28** 50**** Good for the whole family 33 57** 67 Kids would like it 23 43** 34 Good for a treat 9 31** 21 Would give me energy for the day 27 42** 51 A healthy option 42 60** 64 Would taste great 18 49** 53 Innovative 4 13** 16 A modern brand 13 27** 21 A brand I feel positively about 18 40** 48 Better quality than other brands 14 31** 36 Appeals to you more 8 34** 36 Offers something other brands do not 9 21** 23

Average 19 36** 40 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 206: Proportion of non, light and heavier buyer cereal response, FCB Attribute prompted – Brand E.

Brand E (FCB Attribute) % Non-Buyer

(n=317)

% Light Buyer (n=65)

% Heavier Buyer

(n=120) Stays crispy in milk 35 49** 57 Natural 69 75 85 Good value for money 55 71** 83*** Good for the whole family 74 88** 96**** Kids would like it 47 63** 65 Good for a treat 31 52** 53 Would give me energy for the day 70 80* 90*** A healthy option 71 83** 90 Would taste great 49 89** 93 Innovative 34 49** 63 A modern brand 41 57** 58 A brand I feel positively about 46 83** 93**** Better quality than other brands 53 77** 88**** Appeals to you more 25 58** 81**** Offers something other brands do not 45 68** 60

Average 50 70** 77 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Page 172: Understanding and Measuring Light Buyer Brand Equity

Table 207: Proportion of non, light and heavier buyer cereal response, FCB Brand prompted – Brand E.

Brand E (FCB Brand) % Non-Buyer

(n=336)

% Light Buyer (n=50)

% Heavier Buyer

(n=117) Stays crispy in milk 49 62* 58 Natural 73 92** 91 Good value for money 56 72** 88**** Good for the whole family 71 90** 92 Kids would like it 49 72** 81 Good for a treat 38 72** 74 Would give me energy for the day 69 86** 89 A healthy option 74 90** 91 Would taste great 56 86** 92 Innovative 41 58** 73*** A modern brand 46 78** 65 A brand I feel positively about 53 88** 93 Better quality than other brands 52 76** 87*** Appeals to you more 35 68** 83**** Offers something other brands do not 55 74** 83

Average 55 78** 83 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 208: Proportion of non, light and heavier buyer cereal response, PA Attribute prompted – Brand F.

Brand F (PA Attribute) % Non-Buyer

(n=356)

% Light Buyer (n=61)

% Heavier Buyer (n=87)

Stays crispy in milk 16 30** 37 Natural 6 11 28**** Good value for money 16 46** 74**** Good for the whole family 19 48** 67**** Kids would like it 53 74** 59 Good for a treat 14 23* 31 Would give me energy for the day 16 34** 52**** A healthy option 10 20** 36**** Would taste great 22 57** 67 Innovative 14 23* 34 A modern brand 27 36 56**** A brand I feel positively about 12 44** 60*** Better quality than other brands 10 21** 46**** Appeals to you more 8 36** 54**** Offers something other brands do not 11 23** 32

Average 17 35** 49*** Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

163

Table 209: Proportion of non, light and heavier buyer cereal response, PA Brand prompted – Brand F.

Brand F (PA Brand) % Non-Buyer

(n=344)

% Light Buyer (n=61)

% Heavier Buyer (n=99)

Stays crispy in milk 23 46** 58 Natural 14 21 34*** Good value for money 15 31** 56**** Good for the whole family 31 59** 62 Kids would like it 59 57 75**** Good for a treat 19 26 37 Would give me energy for the day 12 25** 33 A healthy option 20 39** 48 Would taste great 22 49** 59 Innovative 7 11 13 A modern brand 28 46** 53 A brand I feel positively about 15 41** 55*** Better quality than other brands 10 21** 34*** Appeals to you more 5 15** 33**** Offers something other brands do not 9 13 16

Average 19 33** 44 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 210: Proportion of non, light and heavier buyer cereal response, FCB Attribute prompted – Brand F.

Brand F (FCB Attribute) % Non-Buyer

(n=320)

% Light Buyer (n=65)

% Heavier Buyer

(n=117) Stays crispy in milk 41 63** 62 Natural 28 52** 61 Good value for money 35 68** 70 Good for the whole family 46 85** 89 Kids would like it 86 94* 87 Good for a treat 39 65** 62 Would give me energy for the day 39 75** 73 A healthy option 29 58** 71*** Would taste great 37 89** 91 Innovative 38 60** 62 A modern brand 61 83** 74 A brand I feel positively about 29 86** 82 Better quality than other brands 38 77** 70 Appeals to you more 16 60** 66 Offers something other brands do not 39 71** 53

Average 40 72** 71 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Page 173: Understanding and Measuring Light Buyer Brand Equity

Table 211: Proportion of non, light and heavier buyer cereal response, FCB Brand prompted – Brand F.

Brand F (FCB Brand) % Non-Buyer

(n=324)

% Light Buyer (n=66)

% Heavier Buyer

(n=113) Stays crispy in milk 57 71** 67 Natural 49 67** 81**** Good value for money 47 77** 82 Good for the whole family 57 92** 88 Kids would like it 79 89* 94 Good for a treat 54 86** 84 Would give me energy for the day 52 77** 80 A healthy option 48 67** 79*** Would taste great 55 92** 92 Innovative 51 73** 68 A modern brand 62 80** 84 A brand I feel positively about 39 85** 88 Better quality than other brands 39 77** 83 Appeals to you more 29 61** 80**** Offers something other brands do not 46 83** 72

Average 51 79** 81 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 212: Proportion of non, light and heavier buyer cereal response, PA Attribute prompted – Brand G.

Brand G (PA Attribute) % Non-Buyer

(n=370)

% Light Buyer (n=51)

% Heavier Buyer (n=83)

Stays crispy in milk 11 25** 28 Natural 3 6 7 Good value for money 12 47** 55 Good for the whole family 11 33** 35 Kids would like it 81 92** 90 Good for a treat 30 69** 64 Would give me energy for the day 11 22** 18 A healthy option 2 10 5 Would taste great 25 75** 75 Innovative 9 25** 37 A modern brand 21 45** 51 A brand I feel positively about 8 43** 47 Better quality than other brands 7 39** 33 Appeals to you more 6 43** 37 Offers something other brands do not 11 29** 34 Average 17 40** 41

Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05. Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

164

Table 213: Proportion of non, light and heavier buyer cereal response, PA Brand prompted – Brand G.

Brand G (PA Brand) % Non-Buyer

(n=370)

% Light Buyer (n=51)

% Heavier Buyer (n=83)

Stays crispy in milk 9 37** 24 Natural 4 6 12 Good value for money 8 21** 28 Good for the whole family 10 35** 38 Kids would like it 69 75 75 Good for a treat 29 63** 61 Would give me energy for the day 5 15** 24 A healthy option 3 10 15 Would taste great 20 58** 60 Innovative 6 15** 21 A modern brand 15 29** 32 A brand I feel positively about 9 31** 40 Better quality than other brands 6 19** 29 Appeals to you more 4 15** 26 Offers something other brands do not 10 27** 23 Average 14 30** 34

Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05. Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 214: Proportion of non, light and heavier buyer cereal response, FCB Attribute prompted – Brand G.

Brand G (FCB Attribute) % Non-Buyer

(n=339)

% Light Buyer (n=63)

% Heavier Buyer

(n=100) Stays crispy in milk 29 44** 56 Natural 14 27** 37 Good value for money 33 67** 67 Good for the whole family 32 62** 66 Kids would like it 96 97 95 Good for a treat 53 87** 83 Would give me energy for the day 29 52** 56 A healthy option 8 22** 32 Would taste great 41 89** 87 Innovative 33 57** 62 A modern brand 51 71** 76 A brand I feel positively about 25 78** 79 Better quality than other brands 30 78** 71 Appeals to you more 16 71** 59 Offers something other brands do not 34 56** 56 Average 35 64** 65

Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05. Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Page 174: Understanding and Measuring Light Buyer Brand Equity

Table 215: Proportion of non, light and heavier buyer cereal response, FCB Brand prompted – Brand G.

Brand G (FCB Brand) % Non-Buyer

(n=352)

% Light Buyer (n=56)

% Heavier Buyer (n=95)

Stays crispy in milk 57 63 71 Natural 26 39** 44 Good value for money 46 63** 77*** Good for the whole family 42 75** 86*** Kids would like it 93 100** 96 Good for a treat 70 91** 94 Would give me energy for the day 41 52 76**** A healthy option 19 29* 47**** Would taste great 61 91** 96 Innovative 47 50 71**** A modern brand 63 75* 89**** A brand I feel positively about 40 77** 89**** Better quality than other brands 38 75** 86*** Appeals to you more 21 61** 80**** Offers something other brands do not 48 59 73*** Average 47 67** 78

Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05. Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 216: Proportion of non, light and heavier buyer cereal response, PA Attribute prompted – Brand H.

Brand H (PA Attribute) % Non-Buyer

(n=365)

% Light Buyer (n=56)

% Heavier Buyer (n=83)

Stays crispy in milk 22 36** 48 Natural 2 7 16 Good value for money 19 52** 66*** Good for the whole family 16 46** 55 Kids would like it 65 64 64 Good for a treat 29 54** 55 Would give me energy for the day 16 32** 36 A healthy option 3 4 13**** Would taste great 37 71** 82 Innovative 8 23** 37*** A modern brand 16 39** 48 A brand I feel positively about 13 46** 59 Better quality than other brands 11 45** 47 Appeals to you more 8 46** 51 Offers something other brands do not 8 21** 34

Average 18 39** 47 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

165

Table 217: Proportion of non, light and heavier buyer cereal response, PA Brand prompted – Brand H.

Brand H (PA Brand) % Non-Buyer

(n=366)

% Light Buyer (n=60)

% Heavier Buyer (n=78)

Stays crispy in milk 20 35** 33 Natural 7 10 13 Good value for money 13 32** 41 Good for the whole family 18 42** 55 Kids would like it 63 60 72 Good for a treat 29 50** 50 Would give me energy for the day 11 27** 32 A healthy option 6 10 18 Would taste great 31 55** 68 Innovative 5 13** 9 A modern brand 13 32** 29 A brand I feel positively about 14 38** 45 Better quality than other brands 10 28** 26 Appeals to you more 7 33** 32 Offers something other brands do not 7 12 21

Average 17 32** 36 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 218: Proportion of non, light and heavier buyer cereal response, FCB Attribute prompted – Brand H.

Brand H (FCB Attribute) % Non-Buyer

(n=360)

% Light Buyer (n=56)

% Heavier Buyer (n=86)

Stays crispy in milk 48 68** 73 Natural 16 34** 44 Good value for money 41 70** 74 Good for the whole family 42 80** 78 Kids would like it 92 96 98 Good for a treat 55 86** 86 Would give me energy for the day 41 68** 67 A healthy option 11 30** 29 Would taste great 61 98** 98 Innovative 31 52** 57 A modern brand 43 64** 64 A brand I feel positively about 40 86** 88 Better quality than other brands 45 79** 84 Appeals to you more 29 68** 85**** Offers something other brands do not 34 61** 60

Average 42 69** 72 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Page 175: Understanding and Measuring Light Buyer Brand Equity

Table 219: Proportion of non, light and heavier buyer cereal response, FCB Brand prompted – Brand H.

Brand H (FCB Brand) % Non-Buyer

(n=346)

% Light Buyer (n=63)

% Heavier Buyer (n=94)

Stays crispy in milk 57 63 66 Natural 29 44** 52 Good value for money 47 62** 83**** Good for the whole family 50 79** 82 Kids would like it 90 97* 96 Good for a treat 66 90** 91 Would give me energy for the day 49 67** 83**** A healthy option 22 24 50**** Would taste great 69 92** 93 Innovative 41 52* 65 A modern brand 45 62** 73 A brand I feel positively about 46 78** 90**** Better quality than other brands 45 78** 88*** Appeals to you more 37 71** 84*** Offers something other brands do not 42 62** 81****

Average 49 68** 79 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 220: Proportion of non, light and heavier buyer butter/margarine response, PA Attribute prompted – Brand B.

Brand B (PA Attribute) % Non-Buyer

(n=343)

% Light Buyer (n=57)

% Heavier Buyer

(n=103) Spreads easily 44 61** 67 Natural 10 28** 37 Good value for money 12 26** 48**** Good for the whole family 26 54** 65 Kids would like it 15 40** 33 Good for cooking/baking 16 30** 33 Helps control cholesterol 18 28* 32 A healthy option 17 30** 39 Would taste great 13 47** 54 Innovative 10 19** 29 A modern brand 20 39** 37 A brand I feel positively about 15 51** 67**** Better quality than other brands 10 37** 51*** Appeals to you more 6 26** 50**** Offers something other brands do not 7 18** 25

Average 16 36** 44 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

166

Table 221: Proportion of non, light and heavier buyer butter/margarine response, PA Brand prompted – Brand B.

Brand B (PA Brand) % Non-Buyer

(n=361)

% Light Buyer (n=47)

% Heavier Buyer (n=94)

Spreads easily 42 57** 61 Natural 12 21* 32 Good value for money 15 34** 49*** Good for the whole family 24 36* 49 Kids would like it 8 15* 22 Good for cooking/baking 17 26 38*** Helps control cholesterol 16 17 36**** A healthy option 23 36** 47 Would taste great 18 32** 40 Innovative 6 6 24**** A modern brand 16 30** 38 A brand I feel positively about 15 30** 48**** Better quality than other brands 10 21** 33*** Appeals to you more 5 13** 32**** Offers something other brands do not 4 9 15

Average 15 26* 38 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 222: Proportion of non, light and heavier buyer butter/margarine response, FCB Attribute prompted – Brand B.

Brand B (FCB Attribute) % Non-Buyer

(n=340)

% Light Buyer (n=60)

% Heavier Buyer

(n=104) Spreads easily 85 95** 94 Natural 35 65** 66 Good value for money 52 73** 77 Good for the whole family 59 87** 82 Kids would like it 52 62 73 Good for cooking/baking 45 58** 58 Helps control cholesterol 34 40 44 A healthy option 41 45 61**** Would taste great 46 80** 84 Innovative 35 48** 49 A modern brand 49 60 69 A brand I feel positively about 41 80** 86 Better quality than other brands 36 73** 76 Appeals to you more 20 63** 65 Offers something other brands do not 26 38* 49

Average 44 65** 69 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Page 176: Understanding and Measuring Light Buyer Brand Equity

Table 223: Proportion of non, light and heavier buyer butter/margarine response, FCB Brand prompted – Brand B.

Brand B (FCB Brand) % Non-Buyer

(n=352)

% Light Buyer (n=53)

% Heavier Buyer (n=99)

Spreads easily 83 85 95**** Natural 45 60** 66 Good value for money 53 79** 85 Good for the whole family 64 75 89**** Kids would like it 57 68 88**** Good for cooking/baking 61 72 77 Helps control cholesterol 42 55* 67 A healthy option 49 60 74*** Would taste great 57 91** 87 Innovative 41 60** 68 A modern brand 58 74** 76 A brand I feel positively about 52 83** 85 Better quality than other brands 45 72** 77 Appeals to you more 31 60** 73 Offers something other brands do not 38 57** 67

Average 52 70** 78 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 224: Proportion of non, light and heavier buyer butter/margarine response, PA Attribute prompted – Brand C.

Brand C (PA Attribute) % Non-Buyer

(n=350)

% Light Buyer (n=58)

% Heavier Buyer (n=95)

Spreads easily 45 72** 86**** Natural 4 16** 20 Good value for money 11 45** 69**** Good for the whole family 16 53** 71**** Kids would like it 25 47** 71**** Good for cooking/baking 8 22** 44**** Helps control cholesterol 7 12** 16 A healthy option 8 16* 29**** Would taste great 15 52** 78**** Innovative 17 28* 45**** A modern brand 27 50** 67**** A brand I feel positively about 7 40** 63**** Better quality than other brands 4 21** 38**** Appeals to you more 5 24** 58**** Offers something other brands do not 7 16** 27***

Average 14 34** 52**** Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

167

Table 225: Proportion of non, light and heavier buyer butter/margarine response, PA Brand prompted – Brand C.

Brand C (PA Brand) % Non-Buyer

(n=346)

% Light Buyer (n=46)

% Heavier Buyer

(n=110) Spreads easily 37 80** 64 Natural 4 15** 16 Good value for money 16 48** 53 Good for the whole family 15 46** 43 Kids would like it 12 26** 29 Good for cooking/baking 9 17* 33**** Helps control cholesterol 5 4 11 A healthy option 11 22** 22 Would taste great 12 52** 57 Innovative 9 15 19 A modern brand 23 28 39 A brand I feel positively about 7 26** 34 Better quality than other brands 4 13** 25*** Appeals to you more 4 15** 27*** Offers something other brands do not 4 15** 12

Average 11 28** 32 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 226: Proportion of non, light and heavier buyer butter/margarine response, FCB Attribute prompted – Brand C.

Brand C (FCB Attribute) % Non-Buyer

(n=354)

% Light Buyer (n=52)

% Heavier Buyer (n=98)

Spreads easily 86 94* 96 Natural 22 42** 59**** Good value for money 45 75** 89**** Good for the whole family 47 81** 87 Kids would like it 64 79** 91**** Good for cooking/baking 28 44** 57 Helps control cholesterol 16 31** 31 A healthy option 27 42** 57*** Would taste great 41 77** 88*** Innovative 37 54** 70**** A modern brand 63 75* 92**** A brand I feel positively about 25 65** 91**** Better quality than other brands 21 46** 66**** Appeals to you more 15 48** 83**** Offers something other brands do not 22 37** 53****

Average 37 59** 74*** Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Page 177: Understanding and Measuring Light Buyer Brand Equity

Table 227: Proportion of non, light and heavier buyer butter/margarine response, FCB Brand prompted – Brand C.

Brand C (FCB Brand) % Non-Buyer

(n=349)

% Light Buyer (n=47)

% Heavier Buyer

(n=108) Spreads easily 76 91** 94 Natural 28 53** 54 Good value for money 46 77** 88*** Good for the whole family 50 72** 87**** Kids would like it 56 74** 88**** Good for cooking/baking 42 74** 76 Helps control cholesterol 25 40** 49 A healthy option 31 49** 62 Would taste great 49 77** 88*** Innovative 39 66** 63 A modern brand 60 85** 87 A brand I feel positively about 34 70** 78 Better quality than other brands 28 55** 70*** Appeals to you more 25 62** 70 Offers something other brands do not 33 64** 59

Average 41 67** 74 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 228: Proportion of non, light and heavier buyer butter/margarine response, PA Attribute prompted – Brand D.

Brand D (PA Attribute) % Non-Buyer

(n=356)

% Light Buyer (n=40)

% Heavier Buyer

(n=107) Spreads easily 11 35** 25 Natural 37 70** 77 Good value for money 8 30** 50**** Good for the whole family 24 48** 69**** Kids would like it 22 50** 63 Good for cooking/baking 24 55** 66 Helps control cholesterol 3 13 9 A healthy option 8 20** 37**** Would taste great 36 80** 89 Innovative 7 20** 17 A modern brand 13 25** 33 A brand I feel positively about 16 68** 79*** Better quality than other brands 20 60** 78**** Appeals to you more 11 58** 79**** Offers something other brands do not 9 35** 46 Average 17 44** 54

Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05. Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

168

Table 229: Proportion of non, light and heavier buyer butter/margarine response, PA Brand prompted – Brand D.

Brand D (PA Brand) % Non-Buyer

(n=369)

% Light Buyer (n=47)

% Heavier Buyer (n=86)

Spreads easily 12 32** 35 Natural 27 43** 50 Good value for money 12 30** 44*** Good for the whole family 16 36** 42 Kids would like it 8 23** 29 Good for cooking/baking 23 43** 41 Helps control cholesterol 3 9 6 A healthy option 9 13 26*** Would taste great 33 68** 66 Innovative 6 9 9 A modern brand 11 26** 29 A brand I feel positively about 13 45** 49 Better quality than other brands 12 38** 38 Appeals to you more 7 26** 40*** Offers something other brands do not 4 11 12 Average 13 30** 34

Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05. Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 230: Proportion of non, light and heavier buyer butter/margarine response, FCB Attribute prompted – Brand D.

Brand D (FCB Attribute) % Non-Buyer

(n=323)

% Light Buyer (n=62)

% Heavier Buyer

(n=119) Spreads easily 41 39 41 Natural 68 89** 89 Good value for money 40 60** 78**** Good for the whole family 60 76** 86*** Kids would like it 65 79** 83 Good for cooking/baking 65 79** 85 Helps control cholesterol 10 15 22 A healthy option 26 40** 50 Would taste great 72 87** 97**** Innovative 28 32 39 A modern brand 46 44 49 A brand I feel positively about 45 82** 92**** Better quality than other brands 46 71** 89**** Appeals to you more 27 69** 85**** Offers something other brands do not 28 40** 58**** Average 45 60** 70

Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05. Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Page 178: Understanding and Measuring Light Buyer Brand Equity

Table 231: Proportion of non, light and heavier buyer butter/margarine response, FCB Brand prompted – Brand D.

Brand D (FCB Brand) % Non-Buyer

(n=393)

% Light Buyer (n=43)

% Heavier Buyer (n=68)

Spreads easily 45 53 60 Natural 68 93** 90 Good value for money 53 72** 75 Good for the whole family 60 93** 85 Kids would like it 67 86** 82 Good for cooking/baking 70 95** 87 Helps control cholesterol 23 23 43**** A healthy option 38 44 53 Would taste great 77 95** 96 Innovative 36 42 51 A modern brand 45 63** 71 A brand I feel positively about 53 88** 93 Better quality than other brands 52 74** 81 Appeals to you more 41 77** 82 Offers something other brands do not 39 49 59 Average 51 70** 74

Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05. Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 232: Proportion of non, light and heavier buyer butter/margarine response, PA Attribute prompted – Brand F.

Brand F (PA Attribute) % Non-Buyer

(n=386)

% Light Buyer (n=50)

% Heavier Buyer (n=67)

Spreads easily 42 80** 79 Natural 9 16* 22 Good value for money 11 30** 40 Good for the whole family 23 68** 54*** Kids would like it 23 58** 45 Good for cooking/baking 17 42** 43 Helps control cholesterol 11 18 25 A healthy option 9 16* 28*** Would taste great 22 66** 63 Innovative 14 42** 33 A modern brand 30 46** 54 A brand I feel positively about 11 44** 67**** Better quality than other brands 11 40** 52 Appeals to you more 7 32** 51**** Offers something other brands do not 6 24** 30

Average 16 41** 46 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

169

Table 233: Proportion of non, light and heavier buyer butter/margarine response, PA Brand prompted – Brand F.

Brand F (PA Brand) % Non-Buyer

(n=392)

% Light Buyer (n=43)

% Heavier Buyer (n=67)

Spreads easily 41 53* 69 Natural 7 14 16 Good value for money 15 23 39*** Good for the whole family 22 26 43*** Kids would like it 12 19 34*** Good for cooking/baking 18 23 30 Helps control cholesterol 9 21** 27 A healthy option 16 40** 28 Would taste great 23 42** 60*** Innovative 10 16 15 A modern brand 20 28 43*** A brand I feel positively about 13 26** 37 Better quality than other brands 13 21 36*** Appeals to you more 7 21** 30 Offers something other brands do not 7 9 15

Average 15 25** 35 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Table 234: Proportion of non, light and heavier buyer butter/margarine response, FCB Attribute prompted – Brand F.

Brand F (FCB Attribute) % Non-Buyer

(n=380)

% Light Buyer (n=45)

% Heavier Buyer (n=79)

Spreads easily 85 91 95 Natural 34 49** 71**** Good value for money 45 56 76**** Good for the whole family 55 76** 86 Kids would like it 62 73 92**** Good for cooking/baking 38 53** 71**** Helps control cholesterol 21 29 38 A healthy option 30 31 57**** Would taste great 53 78** 92**** Innovative 39 60** 65 A modern brand 65 78* 78 A brand I feel positively about 36 78** 85 Better quality than other brands 33 60** 81**** Appeals to you more 21 64** 75 Offers something other brands do not 22 49** 63

Average 42 62** 75 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

Page 179: Understanding and Measuring Light Buyer Brand Equity

Table 235: Proportion of non, light and heavier buyer butter/margarine response, FCB Brand prompted – Brand F.

Brand F (FCB Brand) % Non-Buyer

(n=362)

% Light Buyer (n=61)

% Heavier Buyer (n=81)

Spreads easily 93 98* 94 Natural 48 64** 73 Good value for money 57 75** 83 Good for the whole family 70 87** 94 Kids would like it 65 84** 85 Good for cooking/baking 67 80** 85 Helps control cholesterol 54 69** 68 A healthy option 58 66 78 Would taste great 65 97** 94 Innovative 50 70** 77 A modern brand 78 80 94**** A brand I feel positively about 55 92** 93 Better quality than other brands 47 79** 85 Appeals to you more 32 74** 80 Offers something other brands do not 36 64** 64

Average 58 79** 83 Statistically significantly higher than Non-buyer at *p<0.1; **p<0.05.

Statistically significantly higher than Light buyer at ***p<0.1; ****p<0.05.

170