Factors influencing brand loyalty in the retail sphere Nekita Exner A research report submitted to the Faculty of Commerce, Law and Management, University of the Witwatersrand, in partial fulfilment of the requirements for the degree of Master of Management in the field of Strategic Marketing for the degree MMSM Wits Business School February 2021 i ABSTRACT A key success factor in gaining market share in today’s rapid ever-changing environment, relies on sustaining long-term relationships with all stakeholders. Changing apathetic customers into loyal customers and establishing long term relationships with customers is critical for organizational success. The purpose of this study is to gain insight into what drives consumers loyalty to a loyalty program and to a brand in the retail sphere in South Africa. A quantitative research method was employed to gain insight into these pertinent constructs to gather further understanding from the South African customer’s perspective. The questionnaires were administered online via Qualtrics, using a convenience sampling method. The study draws research insights from a sample of 203 educated, older, and high-income earning South African respondents. Structural equation modeling was used as an analysis technique. The researcher also developed key findings, such as that consumers are most loyal to Dischem and Woolworths. In terms of theory development, this study provides a cohesive framework and model on the subject of brand loyalty. In terms of practical implications, the results of this study produce some key implications for managers in the retail industry in South Africa. This study shows that there a distinct and positive link between loyalty to a loyalty programme and loyalty to the retail brand. Other notable findings include the positive drivers of loyalty towards a loyalty programme, which include loyalty programme value and programme social benefits. Furthermore, customisation was found to play a critical role in loyalty towards a loyalty programme as well as loyalty towards a retail brand. The most notable finding is that loyalty schemes do influence brand loyalty directly within the retail sphere in South Africa. ii DECLARATION I, Nekita Exner, declare that this research report is my own work except where it has been indicated in the various references and acknowledgements. The report is submitted in partial fulfilment of the requirements for the degree of Master of Management in the field of Strategic Marketing through Wits Business School at the University of the Witwatersrand, Johannesburg. This report has not been submitted previously for any degree or examination at this or any other university. _______________________ Nekita Exner Signed at Johannesburg Date: 27.02.2021 iii ACKNOWLEDGEMENTS I would like to extend my gratitude to the following people and acknowledge them for their contribution to this study. Firstly, to my supervisor, Neale Penman, without whom I could never have completed this study. I can’t thank you enough for all your wisdom, guidance, patience, and mentorship during this long process. I will forever be grateful. Secondly, the incredible lecturers, and administration staff at the Wits Business School. Lastly, to my incredible partner, and my family and friends for their guidance and support during my studies. iv TABLE OF CONTENTS ABSTRACT ............................................................................................................ I DECLARATION ..................................................................................................... II ACKNOWLEDGEMENTS..................................................................................... III LIST OF TABLES ............................................................................................... VIII LIST OF FIGURES ............................................................................................... IX CHAPTER 1: INTRODUCTION ........................................................................... 1 1.1 Purpose of the study ....................................................................................... 3 1.2 Context of the study ........................................................................................ 3 1.3 Problem statement .......................................................................................... 4 1.4 Significance of the study ................................................................................. 6 1.5 Delimitations of the study ................................................................................ 7 1.6 Definition of terms ........................................................................................... 7 1.7 Assumptions ................................................................................................... 8 CHAPTER 2: LITERATURE REVIEW ................................................................ 9 2.1 Introduction ..................................................................................................... 9 2.2 Theoretical Grounding ..................................................................................... 9 2.3 Drivers of programme loyalty......................................................................... 11 2.3.1 Loyalty programme value ......................................................................................... 11 2.3.2 Social Benefits of the programme ............................................................................ 11 2.4 Customization and loyalty schemes .............................................................. 12 2.5 Customization and brand loyalty ................................................................... 13 v 2.6 Loyalty Schemes and brand loyalty ............................................................... 14 2.7 Conceptual model ......................................................................................... 17 2.8 Conclusion of Literature Review .................................................................... 17 2.8.1 Research hypotheses............................................................................................... 17 CHAPTER 3: RESEARCH METHODOLOGY ................................................... 21 3.1 Research methodology / paradigm ................................................................ 21 3.2 Research Design........................................................................................... 21 3.3 Population and sample .................................................................................. 22 3.3.1 Population ................................................................................................................ 22 3.3.2 Sample and sampling method .................................................................................. 23 3.4 The research instrument ............................................................................... 23 3.5 Procedure for data collection ......................................................................... 24 3.6 Data analysis and interpretation .................................................................... 25 3.7 Validity and reliability of research .................................................................. 26 3.7.1 Validity ...................................................................................................................... 26 3.7.2 Reliability .................................................................................................................. 26 3.7.3 Ethical Consideration ............................................................................................... 27 CHAPTER 4: PRESENTATION OF THE RESEARCH RESULTS ................... 28 4.1 Introduction ................................................................................................... 28 4.2 Data screening .............................................................................................. 28 4.3 Sample Characteristics / Demographic ......................................................... 28 4.3.1 Gender ..................................................................................................................... 28 4.3.2 Age ........................................................................................................................... 29 4.3.3 Level of education .................................................................................................... 29 4.3.4 Income Level ............................................................................................................ 30 4.4 Discussion regarding Sample Characteristics / Demographic ....................... 30 4.5 Model fit ........................................................................................................ 31 4.6 Validity .......................................................................................................... 32 vi 4.7 Reliability ...................................................................................................... 37 4.8 Correlation analysis ....................................................................................... 38 4.9 Hypothesis Testing ........................................................................................ 38 4.10 Summary of hypotheses ................................................................................ 43 CHAPTER 5: RESEARCH DISCUSSION ......................................................... 44 5.1 Introduction ................................................................................................... 44 5.2 Discussion of the research problem .............................................................. 44 5.3 Sub-Problems ............................................................................................... 44 5.4 Discussion regarding the Sample Characteristics / Demographic .................. 45 5.5 Evaluating the results based on the retailer and loyalty scheme chosen by the respondents ............................................................................................ 45 5.6 Discussion pertaining to hypotheses ............................................................. 48 5.6.1 Hypothesis 1 ............................................................................................................. 48 5.6.2 Hypothesis 2 ............................................................................................................. 49 5.6.3 Hypothesis 3 & 4 ...................................................................................................... 50 5.6.4 Hypothesis 5 ............................................................................................................. 51 5.7 Demographic data and share of wallet .......................................................... 53 5.7.1 Drivers’ Share of Wallet............................................................................................ 53 5.8 Conclusion .................................................................................................... 55 CHAPTER 6: CONCLUSIONS AND RECOMMENDATIONS ........................... 56 6.1 Introduction ................................................................................................... 56 6.2 Conclusions of the study ............................................................................... 56 6.3 Limitations of the study & Directions for future research ................................ 57 6.4 Recommendations ........................................................................................ 58 6.5 Suggestions for further research ................................................................... 59 REFERENCES .................................................................................................... 61 APPENDIX A ....................................................................................................... 72 Programme loyalty and brand/company loyalty (Evanschitzky et al., 2012) ................ 72 vii APPENDIX B ....................................................................................................... 73 Customization and Brand Loyalty (Coelho & Henseler, 2012) ..................................... 73 APPENDIX C ....................................................................................................... 74 Research Questionnaire ............................................................................................. 74 viii LIST OF TABLES Table 1: Summary of the research hypotheses and relevant literature ................ 18 Table 2: Respondent age distribution .................................................................. 29 Table 3: Absolute Fit Indexes ............................................................................... 31 Table 4: KMO and Bartlett's Test ......................................................................... 32 Table 5: Total Variance Explained ....................................................................... 33 Table 6: Pattern Matrix ......................................................................................... 35 Table 7: Reliability of scale .................................................................................. 37 Table 8: Descriptive Statistics and Pearson Correlation ...................................... 38 Table 9: SEM Model - Regression Weights ......................................................... 40 Table 10: Summary of all hypotheses .................................................................. 43 Table 11: Summary statistics ............................................................................... 46 Table 12: Model summary.................................................................................... 54 Table 13: ANOVA ................................................................................................ 54 Table 14: Coefficients .......................................................................................... 54 ix LIST OF FIGURES Figure 1: Conceptual model ................................................................................. 17 Figure 2: Respondent gender .............................................................................. 28 Figure 3: Highest level of education ..................................................................... 29 Figure 4: Income levels ........................................................................................ 30 Figure 5: Scree plot .............................................................................................. 34 Figure 6: CFA Model of New model ..................................................................... 36 Figure 7: Detailed SEM analysis .......................................................................... 39 Figure 8: Simplified SEM analysis ........................................................................ 40 1 CHAPTER 1: INTRODUCTION Brand loyalty is a central theme of marketing theory and practice as a result of the many strategic benefits that it confers (Alhaddad, 2015). Marketers have long been interested in this concept, which measures the attachment that a customer has to a brand (Chinomona & Maziriri, 2017) as well as the customer’s intention to repurchase a particular product or service (Kim, Park, Park, Kim, & Kim, 2018). In terms of traditional consumer marketing, a firm that has cultivated strong customer loyalty will enjoy the following advantages: the ability to maintain premium pricing; greater bargaining power with respect to distribution channels; reduced selling costs; a strong barrier to potential new entries into the product/service category; synergistic advantages of brand extensions to related product/service categories; repeat purchases; and recommendations by customers of the firm to friends and relatives (Sharma, 2017). Loyal customers are more profitable than non-loyal customers. Global recognition of this fact has led companies to focus on retaining loyal customers (Cengiz & Akdemir-Cengiz, 2016). To be successful, retailers must persuade customers to repurchase their products. One way to do this is by building brand loyalty. Today, strategies to enhance and maintain brand loyalty are so commonly deployed by retailers that increasing its brand loyalty has become a strategic imperative for any firm that wishes to compete in the retail sphere (Moretta, Cavacece, Russo, & Granata, 2019). Establishing customer loyalty is regarded as one of the largest challenges faced by marketers today. Alrubaiee and Al-Nazer (2010) argue that creating loyal customers is the single most significant driver of an organization’s long-term financial performance, leading to increased sales and customer share, lower costs, and higher prices. It is therefore crucial for firms to understand the factors that influence brand loyalty (Srivastava, 2018). Retailers in South Africa are under threat; building brand loyalty is no longer optional if they are to survive (Kasai & Chauke, 2017). The global economic climate, together with South Africa’s declining economy, increasing unemployment rate and currency 2 depreciation, have resulted in a significant decrease in customers’ disposable income. Now more than ever, retailers need all the tools available to encourage repeat purchases (Beneke, Hayworth, Hobson, & Mia, 2012). According to The Sunday Times Top Brands Survey, South African’s have clear favourites when it comes to which loyalty programmes are their most preferred programmes. The survey revealed that amongst the loyalty programmes, the top are: Pick n Pay Smart Shopper, Clicks Clubcard, Woolworths Wrewards, Edgards Thank U and Dischem Benefit. (Businesstech.co.za 2021). According to the retailer’s annual investment reports for 2020, the retailers performed as follows: • Pick n Pay – Turnover increased from R87.2bn in FY 19 to R89.2bn in FY 20 while improving their GP from 19.1% to 19.7%. Their net profit margin remaining flat at 2.1% for both periods under review. Market share of food stables such as maize, sugar and oil reported growth. (Picknpayinvester.co.za 2021) • Dischem – Revenue increased by 12% to R24bn. Total income increased by 9.8% to R6.8bn. Total income margin remained at 28.5%. Operating margin improved to 5.2%. Market share in all core categories reported increases. (thevault.exchange.co.za 2021) • Clicks – Turnover increased from R8.65bn in FY 19 to R9.375bn in FY 20. Net profit improved by 11.8% from R1.681bn to R1.880bn for the same period. Market share gains were reported across all core categories except for Clicks Pharmacy, which saw a slight decline of -0.3%. (Clicksgroup.co.za 2021) • Woolworths – Turnover decreased from R78.335bn in FY19 to R78.262bn in FY20. Gross profit decreased from R27.473bn in FY19 to R25.336bn in FY20. 3 (Woolworthsholdings.co.za 2021) Profit before tax decreased from R4.6bn to R2.5bn. Market share within the food turnover and concession sales reported increased. Much of the available research on brand loyalty is focused on its antecedents and consequences and on various strategies to create and enhance it (Moretta Tartaglione, Cavacece, Russo, & Granata, 2019). Within the literature, the author has identified several principal factors that influence brand loyalty. The factors to be explored include loyalty schemes and customization. 1.1 Purpose of the study The purpose of this study is to determine the effect on brand loyalty of loyalty programmes together with loyalty programme drivers, and customization within the retail sphere in South Africa. 1.2 Context of the study Today’s marketplace is dynamic and competitive. Customers are astute and informed and accustomed to having a wide variety of choices and offerings available to them. Customers can easily move their business to competitors who promise similar products or services at lower prices (Rathod, 2016). Contini (2018) proposed that the critical factor in gaining market share in today’s environment is building long-term relationships with stakeholders. A loyalty programme or a reward programme is a marketing tool designed to build customer loyalty by incentivising profitable customers. Loyalty programmes are generally used in strongly competitive markets to build customer loyalty through planned reward schemes based on customers’ purchasing histories (Moretta Tartaglione et al., 2019). Loyalty programmes were first introduced in developed countries, such as the United States of America and the United Kingdom. Over the last decade, the use of loyalty programmes has grown significantly in the South African retail market, with the average South African adult 4 subscribing to at least three of these programmes and the average household subscribing to a total of 10 (Corbishley, 2017). The concept of brand loyalty has been widely discussed in traditional marketing literature and has been the subject of research for more than ninety years (Copeland, 1923). It is widely accepted that there are two facets to brand loyalty, namely, behavioural loyalty and attitudinal loyalty. Behavioural loyalty is defined and measured by the frequency of repeat purchases by consumers, whereas attitudinal loyalty is defined and measured by consumers’ willingness to purchase a particular product or service and to recommend this product or service to others. Brand loyalty is a multidimensional concept and can be defined and measured with reference to behavioural loyalty, attitudinal loyalty or a combination of these (Cengiz & Akdemir-Cengiz, 2016). 1.3 Problem statement The available literature reflects many different views on the effectiveness of loyalty programmes. Partch (1994) suggests that if all companies are forced to offer loyalty programmes in order to obtain a competitive edge, then such programmes serve only to increase operating costs and act as a short-term promotion. Dowling and Uncles (1997) opine that loyalty programmes are unlikely to have a significant impact on customer behaviour, especially in established competitive markets. Doubt in the value of loyalty programmes continues to grow as researchers criticize that they fail to understand consumer behaviour (Xie & Chen, 2014). Other research suggests that loyalty programmes do increase brand loyalty by creating switching costs, thereby increasing operational profits by avoiding price competition (Caminal & Matutes, 1990; Kim, Shi, & Srinivasan, 1997; Klemperer, 1987). O’Malley (1998) suggests that loyalty schemes create false loyalty when customers view them merely as point accumulation systems, but also that, when viewed as a part of a coherent value system, loyalty programmes can play an integral role in developing sustainable loyalty. He views this as being the only viable, long-term role for customer loyalty 5 schemes. Lewis (2004) claims that loyalty programmes represent only one possible technique for increasing customer retention. He suggests that repeat buying may also be encouraged through other means such as short-term discounts on merchandise or reduced shipping charges. According to Rust and Zahorik (1993), customers’ loyalty to a service provider is influenced by their general satisfaction with that provider. Past research has found a positive correlation between customer satisfaction and customer retention (Rust & Zahorik, 1993). Bolton (1998) proposes that the relationship between a customer and a service provider will endure for far longer when that customer is satisfied. Shoemaker and Lewis (1999) found that brand loyalty is increased when a company engages in customization. Gommans, Krishman, and Scheffold (2001) suggest that a satisfied customer tends to be more loyal to a brand over time than a customer whose purchase is motivated by other factors, such as time restrictions or information deficits. According to the literature, card-based loyalty programmes are the most used system in the retail environment (Sharp & Sharp, 1997). These programmes often entail similar mechanics, such as points accumulation for a reward, discounts at the point of purchase, or future discounts based on purchase history. Offering similar loyalty schemes however, may lead to competitive parity (Uncles et al., 2003), Therefore the motivation for retailers to differentiate their loyalty programmes becomes even more pertinent. The primary contribution of this research is to consider the impact of loyalty to towards a loyalty programme, the drivers of this behaviour and the consequence on brand or retailer loyalty. Furthermore, this research aims to consider the impact of drivers of programme loyalty and retailer loyalty. This research will contribute to existing literature by investigating the effect of loyalty schemes, in conjunction with customization, on long term brand loyalty. No other published work, to the best of the researcher’s knowledge, on the retail environment in South Africa has examined the above-mentioned factors. 6 The problem statement can be summarized as the following: Can loyalty schemes directly influence long term brand loyalty? The primary objective of the study is to determine the effect of loyalty schemes on brand loyalty. The secondary problems of the research include the following: - What drives consumers to be loyal to a loyalty programme? - Is there a relationship between customisation and Brand Loyalty? - Is there a relationship between loyalty schemes and customization? - What influence do loyalty schemes, in conjunction with customization have on brand loyalty? 1.4 Significance of the study This study conceptualizes a model that can be used to determine the effects of loyalty schemes and customization on brand loyalty. The hypothesis of this study is that loyalty schemes increase brand loyalty in conjunction with customization. This study provides a novel contribution to the existing body of literature on loyalty schemes and brand loyalty; to the researcher’s knowledge no other study focusing on the effect of these particular factors on brand loyalty has been undertaken in the South African retail sphere. The results of this study provide key insights and implications for managers. Determining whether loyalty schemes influence brand loyalty directly or only in combination with other marketing mechanisms will assist managers in allocating resources efficiently, enabling them to make the best use of the resources available to them. A more nuanced understanding of the impact of loyalty schemes will enable managers to make wise investments in relation to strategies to increase brand loyalty. Stronger brand loyalty positively affects the long-term success of a company and its ability to retain its best customers. This translates into increased market share and competitive advantage for an organization. 7 1.5 Delimitations of the study The research was limited to retail marketing. The research questionnaire asked the respondent to answer the questions in relation to a particular store. 1.6 Definition of terms Loyalty is frequently defined as repeat purchasing over time or similarly, the same or increased volume of same-brand purchasing over time (Tellis, 1988). Newman and Werbel (1973) defined loyal customers as those who re-bought a brand, considered only that brand, and did no brand-related information seeking. Loyalty can be defined not only by the actual repurchase of a product or service but also with reference to the positive attitude of customers towards a brand that may result in the repeat purchase of the product or service (Moretta Tartaglione et al., 2019). Yi and Jeon (2003) describe a loyalty scheme as a marketing programme that aims to build customer loyalty by offering certain incentives to a specific, profitable group of customers. The researcher makes use of the terms loyalty schemes and loyalty programmes interchangeably throughout this paper. Customisation, also commonly referred to personalisation, has been defined as a product or service which is tailored to an individual’s needs or preferences as opposed to the conventional goods or services (Fels, Falk & Schmit, 2017). Programme social benefits refer to the concept of a consumer feeling part of a group or a community, for example, being recognized in a retail store (Hennig-Thurau et al., 2002). Programme value can be defined as a consumer’s perceived value of the loyalty programme in terms of aspirational value, cash rewards and convenience (Yi & Yeon, 2003). 8 1.7 Assumptions • The researcher assumed that the respondents engage in retail shopping. • The researcher assumed that the respondents reflect normal viewpoints and experiences, that would be representative of South African consumers at large. • The researcher assumed that the sample population subscribes to and partakes in a loyalty scheme. 9 CHAPTER 2: LITERATURE REVIEW 2.1 Introduction Brand loyalty has been a prominent topic of discussion in marketing literature for at least five decades (Howard & Sheth, 1969). It has been noted that the establishment of brand loyalty can lead to certain marketing advantages such as reduced marketing costs, increase in number of new customers, and better trade leverage (Aaker, 1991). Other advantages of brand loyalty that are referred to in the literature include positive referrals and greater resistance among loyal customers to competitors’ strategies (Dick & Basu,1994). Despite the clear relevance of brand loyalty to managerial decision-making, conceptual and empirical gaps remain. Constructs that have been linked to brand loyalty in the literature include customization, loyalty schemes, and the respective drivers of loyalty towards a loyalty programme. The purpose of the literature review is to examine these constructs and relate them to brand loyalty. This is done by reviewing past literature on these constructs, delineating them, and providing a concise overview. 2.2 Theoretical Grounding Seminal literature on brand loyalty started around 1923 when Copeland began to define brand loyalty (Copeland, 1923). Since then, researchers have continued to define the brand loyalty construct due to the complex nature of the brand loyalty construct. Oliver (1999) provided a now widely adopted definition and defined brand loyalty as a ‘deeply held psychological commitment to re-buy or re-patronize a particular product or service consistently over time’. The brand loyalty concept has since developed into a multidimensional construct involving multiple determinants (Mustafa, Rahman & Nawai, 2020). The study of brand loyalty in social psychology draws on theories involving both attitudes and behaviours. The classic Tripartite model of attitude (Rosenberg & Hovland, 1960). Is widely accepted and discussed in literature and consists of cognitive, affective, and conative components. The cognitive component of attitude is related to thoughts and 10 beliefs about a brand. The affective component of attitudes refers to emotions or feelings towards a brand. The conative component deals with the past experiences or past behaviours with a brand. As a consequence of this tripartite relationship, loyalty has been considered as a sequential process over time. For example, a person may become cognitively loyal based on beliefs about the brand’s attributes. This person may then go on to become affectively loyal post a positive fulfilment on an experience with the brand. Finally, this person may become conatively loyal by a preference or a commitment towards a particular brand. The literature also suggests that the overarching theory of behavioural brand loyalty is the persistent purchase of a brand over time and that these behavioural insights will shed light on pertinent brand loyalty issues (Bandyopadhyay & Martell, 2007). Behavioural brand loyalty has been simply defined as an individual’s repeat purchasing patterns (Back & Parks, 2003). The repeat purchasing pattern, can be looked at by purchase frequency, the proportion of times that the brand is purchased over competing brands or categories as well as the actual value purchased (Back & Parks, 2003). The literature on brand loyalty reflects the emergence of the relationship marketing paradigm (Gummerus, Von Koskull, & Kowalkowski, 2017). Brand loyalty is grounded in consumer behaviour theory; it is a relational construct where a psychological bond to a particular brand or store is linked to repeat purchase over time (Yuen & Chan, 2010). The literature also underpins that customers who are loyal, may be loyal to different things withing a retail setting. This includes loyalty to the brand, loyalty to the loyalty programme, loyalty to the employees or loyalty to the channel intermediaries (Evanschitzky, et al., (2012). Yi and Jeon (2003) theorized that programme loyalty is the concept of a consumer having positive attitudes towards the benefit or value of a loyalty programme whereas brand loyalty is the concept of the consumer having positive attitudes towards the brand. 11 2.3 Drivers of programme loyalty 2.3.1 Loyalty programme value The theory behind perceived value is rooted in equity theory. Researchers have defined perceived value as the outcome between the perception of what is received and what is actually given, based on the utility of a product or service (Zeithaml, 1988). Accordingly, perceived value is a positive feeling a consumer has when evaluating the benefits received from the loyalty programme versus the effort the consumer has put into being a member of the loyalty programme. These perceived value constructs are the assessment between the costs and the consequent value received. Costs include perceived costs, time and energy expended, and items related to feelings of stress. According to Yang and Peterson (2004), consumers also perceived value by comparing a product or service to a competitive offering. Researchers have cited that perceived value is positively linked to purchase intentions (Parasuraman & Grewal 2000). Research has also pointed out that perceived value may lead to brand loyalty (Parasuraman & Grewal 2000), however researchers also warn that perceived value while creating brand loyalty, may not necessarily lead to loyalty to a loyalty programme (Nobre & Rodrigues, 2018), therefore it is necessary that the items are empirically tested. That stated however, other researchers have found a positive link between perceived value and programme loyalty (Yi & Yeon, 2003; Evanschitzky et al., 2012). Accordingly, the researcher hypothesizes: H1 A: perceived value has a positive relationship with programme loyalty. 2.3.2 Social Benefits of the programme The literature has underlined social benefits as a key driver into what makes consumers loyal to a particular loyalty programme. This concept was identified as one of the relationship benefits by Gwinner et al. (1998), which they concede focuses on the 12 relationship itself rather than the outcome of the relationship. Researchers have found that social relationship aspects have been positively related to fostering commitment between a consumer and a brand/retailer (Evanschitzky et al., 2012). Emotional engagement has also been shown to increase when a consumer feels these social relationship constructs (Szczepanska, et al;. (2011). Social benefits are achieved when members of the loyalty programme feel a sense of belonging and the commonality of shared values (Mimouni-Chaabane & Volle, 2010). The key benefit of establishing social benefits is that they are difficult and take time to replicate, hence offering a competitive advantage (Sanchez-Casado, et al., (2019). Researchers have found a direct relationship between programme social benefits and loyalty towards the loyalty programme (Hennig-Thurau et al. 2002; Evanschitzky et al, 2012). Other researchers have also argued that because of the personalized customer- retailer relationship that social benefits build, social benefits offer a distinct competitive advantage as they take time to perfect and are difficult to replicate (De Wulf et al. ,2001). Accordingly, the following hypothesis is offered: H2 A: Programme social Benefits have a positive effect on brand loyalty 2.4 Customization and loyalty schemes Customization refers to the personalization of content, products, and services. It is the creation of individually suited items that the customer wants and desires. This is normally achieved through the collection of data and then using the data insights to tailor make products, services and content to suit a consumer’s individual needs. The literature suggests that this customization makes it easier for the marketer to bond themselves to the consumer and ultimately leads to brand loyalty (Cho, 2019). Nandal, Nandal, and Malik (2019) state that a successful loyalty scheme is one that the customer views as an asset and not a liability. These researchers state that loyalty to a 13 company or brand will occur when the loyalty scheme provided, offers the customer customized and instant rewards. The literature suggests that, owing to the rewards offered, customers view loyalty schemes as being beneficial to frequent customers who purchase a large volume of products or services from a given retailer or service provider. It is apparent, however, that many customers do not claim the benefits they have earned; many available rewards under loyalty schemes are left unredeemed. Researchers therefore emphasise the importance of using the data collected through these schemes to improve and customize the rewards offered in order to ensure that customers are actually incentivised by the loyalty scheme to repurchase the product or service, thereby increasing brand loyalty. Dorotic, Bijmolt, and Verhoef (2012) hold the view that the future of loyalty schemes lies in customization of the actual scheme and the potential benefits offered to the customer. Accordingly, the following hypothesis is offered: H3A There is a positive relationship between customization and programme loyalty. 2.5 Customization and brand loyalty According to Lewis and Shoemaker (1999), customers were more likely to be loyal to a hotel group if that hotel group provided the following services: customized packages, special requests allowed for certain rooms, alerts during busy times, and lastly, the customer was greeted by name upon arrival at the hotel. Shugan (2005) further emphasized the importance of customization, proposing that service offerings could be customized across different industries and that the very action of a firm offering customized services would create customer loyalty and strengthen the relationship with the customer. Shugan (2005) believes that customization should be a marketing goal. 14 Coelho and Henseler (2012) found that customization has the potential to increase perceived service quality, customer satisfaction, customer trust and ultimately customer loyalty to a product/service. H4A There is a positive relationship between customization and brand loyalty. 2.6 Loyalty Schemes and brand loyalty Copeland (1923) appears to have been the first to suggest a phenomenon related to brand loyalty, which he described as brand assertion. He described this phenomenal as an extreme attitude towards a particular brand that may influence purchase behaviour. Brown (1953) analysed summary measures of brand purchase patterns and found marked consistencies in consumers’ purchase patterns of brands of various products. They concluded that individuals exhibit strong and operative brand loyalty. These spurred continuous investigation into brand loyalty behaviour. Historical research suggests that loyalty programmes do increase brand loyalty (Caminal & Matutes, 1990; Kim, Shi & Srinivasan, 2001; Klemperer, 1987). Airline managers have found loyalty programmes to be extremely successful in increasing brand loyalty by increasing customer switching costs and building barriers to entry for competing airlines. In some industries, loyalty programmes have become a competitive necessity (Winer, 2001) Today’s proliferation of loyalty schemes makes finding a single definition complicated. However, what is common is the main objective to drive loyal behavior of customers by offering rewards for purchase (Sharp & Sharp, 1997). Bijmolt, Dorotic and Verhoef (2010) define loyalty schemes as encompassing the following attributes: • A loyalty programme needs to be long term. The programme needs to a continuous long-term investment that ties consumers in for the consumer lifecycle. 15 • A loyalty programme needs to foster both attitudinal and behavioural loyalty. This should increase the frequency of the purchase and the duration of the purchases thereof over time. It should encourage upselling, cross-selling and increase the share of wallet. • A loyalty programme needs to be formally structured with the customer. The programme needs to be membership based and should be based on certain specified redemption criteria. The programme must be able to recognize purchasing behaviour from members to help build relationships with other members. • A loyalty programme needs to be rewarding. It should reward members based on past, current, and future expected purchases. This is usually done via a specified points accumulation system that stems from the type and frequency of purchases. • The programme provider needs to engage with ongoing marketing efforts to the member. The member should receive personalized communication such as targeted emails, event notifications and individualized promotional offers. Loyalty schemes are defined in the literature as any company initiative that boosts repeat purchases by providing some type of an incentive that persuades customers to purchase more goods more often (Wang, Lewis, Cryder, & Sprigg, 2016). Wait and Lekhuleni (2020) report that the South African market has an abundance of loyalty schemes across various business sectors and that South Africans in general, tend to subscribe to loyalty schemes for several years. Despite the reported high rate of retention of loyalty scheme members, they cite researchers who deny the overall efficacy of such schemes. 16 Evanschitzky et al. (2012) discuss the importance of determining whether customers are loyal to the loyalty scheme or to the retailer. They found that although loyalty towards a retailer influences a customer’s purchase behaviour, it is not a strong predictor of purchase behaviour. Conversely, there results show that loyalty towards a loyalty scheme proves to be a far more notable driver of purchase behaviour. Due to the proliferation of loyalty schemes in the South African market, it is imperative to research consumers’ opinions on them, therefore the following hypothesis is offered: H5A There is a positive relationship between programme loyalty and brand loyalty 17 2.7 Conceptual model Figure 1: Conceptual model 2.8 Conclusion of Literature Review The purpose of this literature review is to provide insight on the topics of customer satisfaction, loyalty schemes, customization, and brand loyalty. The literature underpins the need to research the problem statement in this research paper – the effect on brand loyalty of loyalty schemes in conjunction with customer satisfaction and customization. The available literature on these topics shows clearly how customer satisfaction, loyalty programmes, and customization all contribute to brand loyalty. All these constructs are related to the emerging theory of marketing relationships. These constructs are all attempts to increase patronage to a brand and/or store. 2.8.1 Research hypotheses In concluding the literature review, we list the hypotheses and summarize the relevant authors who have previously studied these relationships. H10: There is no relationship between loyalty programme value and loyalty programmes H1A: There is a positive relationship between loyalty programme value and loyalty programmes 18 H20: There is no relationship between programme social benefits and programme loyalty H2A: There is a positive relationship between programme social benefits and programme loyalty H30: There is no relationship between customization and programme loyalty H3A: There is a positive relationship between customization and programme loyalty H40: There is no relationship between customization and brand loyalty H4A: There is a positive relationship between customization and brand loyalty H50: There is no relationship between programme loyalty and brand loyalty H5A: There is a positive relationship between programme loyalty and brand loyalty Table 1: Summary of the research hypotheses and relevant literature Hypotheses Variables Literature review H1 H10: There is no relationship between loyalty programme value and loyalty programmes H1A: There is a positive relationship between loyalty programme value and loyalty programmes Loyalty programme value Programme loyalty Nobre and Rodrigues (2018) Yi and Yeon (2003); Evanschitzky et al. )2012) 19 H2 H20: There is no relationship between programme social benefits and programme loyalty H2A: There is a positive relationship between programme social benefits and programme loyalty Programme social benefits Programme loyalty Mimouni-Chaabane and Volle (2010) Hennig-Thurau et al. (2002 Evanschitzky et al. (2012) H3 H40: There is no relationship between customization and loyalty schemes H4A: There is a positive relationship between customization and loyalty schemes Customization & Loyalty Schemes Kumar and Shah (2004) Shugan (2005) Schreier (2006) Henderson et al. (2011) Dorotic et al. (2012) Nandal et al. (2019) Cho (2019) H4 H30: There is no relationship between customization and brand loyalty H3A: There is a positive relationship between customization and brand loyalty Customization & Brand Loyalty Lewis and Shoemaker (1999) Morais et al. (2004) Shugan, (2005) Coelho amd Henseler (2012) H5 H20: There is no relationship between loyalty schemes and brand loyalty H2A: There is a positive relationship between loyalty schemes and brand loyalty Loyalty Schemes & Brand Loyalty Wait and Lekhuleni (2020) Bijmolt, Dorotic and Verhoef (2010) Evanschitzky et al. (2012) Kopalle et al. (2012) Ckasai and Chauke (2017) 20 21 CHAPTER 3: RESEARCH METHODOLOGY This chapter outlines the various tasks to be undertaken in the research investigation, how samples were selected, how results were recorded and how diverse research challenges were administered. 3.1 Research methodology / paradigm The research philosophy in this study was positivism. This is the most applicable pragmatic paradigm for a quantitative study. According to Charmaz and Bryant (2011), a positivist researcher will conduct research by identifying a research topic, putting together appropriate and applicable research questions and hypotheses and then adopting the most suitable research methods. 3.2 Research Design The design of the study was a quantitative, non-experimental, cross-sectional, correlation design. The study was quantitative because a set questionnaire was administered once-off to many people within a short space of time. The study was non-experimental because there was no control group, no independent variable manipulation, and no random assignment for a causal conclusion to be made. The study was also cross-sectional which entailed that the data be gathered once and not over a long period of time (Malhotra & Birks, 2007). Quantitative research can be defined as a research method in which structured responses that are gathered from respondents can be condensed into some numerical measurement system (Cannon, Perreault, & McCarthy, 2008). This type of research is used to test for structural relationships, differences, couplings and interactions among assembled variables in a structured manner (Aaker, Kumar & Day, 2004). Quantitative research methods are founded on the premise that it is possible to draw inferences about general populations based on conclusions drawn from sample parameters. Such methods are therefore capable of providing accurate representations of the studied population based on studying sample behaviours and 22 characteristics in a systematic fashion. Quantitative techniques provide one major advantage over qualitative research: unlike qualitative techniques, quantitative measurements can be repeated over time once accurate results have been achieved and accurate strategies have been designed. The literature suggests that the most effective quantitative mechanism available is the use of surveys to obtain structured responses. Survey questionnaires elicit from a target sample, structured responses that can be readily mathematically manipulated for analysis. The efficiencies associated with questionnaire procedures in established literature justified its use for the study of brand loyalty in South Africa. 3.3 Population and sample 3.3.1 Population A population is defined as all individuals, objects and events that meet the sample criteria for inclusion in a study (Burns & Grove, 2005). For the current investigation, the focal population was South Africans who typically engage in purchase related decision making. Based on the population requirements, a suitable sample should be drawn. The respondent criteria for this study would consist of loyalty programme members in major retail stores in Gauteng. In 2014 Woolworths Holdings Ltd had three million members and Pick n Pay Holdings Ltd had more than double this with eight million loyalty card members (Kasai & Chauke, 2017). Retailers do not disclose these databases due to confidentiality and protection of consumer’s personal data, therefore these databases are not available in media or online sources. The researcher allowed the respondents to indicate the particular loyalty programme they subscribed to, based on the major retailers in South Africa. A list of these loyalty programmes with corresponding retailers were provided to the respondents and they were requested to answer the survey questions with the particular retailer, and corresponding loyalty scheme in mind. 23 The respondents needed to be between the ages of 18 and 60 years of age. The population sample are active retail shoppers. 3.3.2 Sample and sampling method A sampling frame is a list of population members used to obtain a sample (Aaker, Day & Kumar, 2004). Due to the sampling method that was used, the sampling frame was identified later. A sample is defined as a subset of the population used and measured for the purpose of representing true population parameters (Cateora & Graham, 2002). In order to draw such a representative sample, adequate sampling procedures and methods were put in place, and these are outlined below. The sampling method used for this investigation included a convenience technique to achieve the objectives and to obtain representative results from the target population. A convenience sampling method was used, as this method is affordable, and the respondents are readily available. According to Etikan, Musa and Alkassim (2016), non-probablity sampling techniques have limitations, they are useful when randomization is not possible. 3.4 The research instrument Marketing research literature defines a questionnaire as a survey method of obtaining structured responses from sample respondents by asking them to answer a series of questions relevant to a particular topic (Aaker, Day & Kumar, 2004; Wegner, 2007). Questionnaires are generally used to collect responses from a large sample audience and serve as the primary communication channel between the researcher and the respondent. The researcher made use a pilot study to test the questionnaire for understandability. Although a questionnaire is defined as a method, various instruments are used to implement and administer such surveys including mail, face-to-face and web-based questionnaires. Since this study aimed to investigate brand loyalty from the consumer’s perspective, online questionnaires were utilized whereby consumers 24 answered the questionnaires to gain insight into consumer’s opinions. In order for these questionnaires to be administered, a study population and sample needed to be identified. The questionnaire sent to respondents included a cover sheet, demographic information and the five scales needed to test customer satisfaction, service, loyalty schemes, customization, and brand loyalty. - To measure Programme value, a 7-point Likert scale developed by Evanschitzky et al. (2012) was used. *Refer to appendix A. - To measure programme social benefits, a 7-point Likert scale developed by Evanschitzky et al. (2012) was used. *Refer to appendix A. - To measure brand/company loyalty, a 7-point Likert scale developed by Evanschitzky et al. (2012) was used. *Refer to appendix A. - To measure customization and brand loyalty, a 7-point Likert scale developed by Coelho, Pedro and Henseler (2012) was used *Refer to appendix B. 3.5 Procedure for data collection The researcher made use of an online structured questionnaire to collect primary data from respondents. Online questionnaires are becoming more and more prevalent due to economic feasibility and convenience for the respondents. It is convenience for respondents as they can answer in their own time and skip through any questions that do not apply to them. This procedure allows researchers to collect large amounts of data efficiently (Regmi, Waitbaka, Pudyal, Simkhada and Van Teijlingen, (2016). The researcher made use of the Qualtrics online platform. The questionnaire took approximately 10 minutes to complete. 25 3.6 Data analysis and interpretation An appropriate analysis technique for quantitative data collection is the use of descriptive tactics, that is, the quantitative description of a collection of data to make reasonable inferences regarding the representative population (Aaker, Day & Kumar, 2004). A wide variety of statistical tests can be used for this purpose, but they are limited to three distinct factors: the types of data, the research design and the assumptions of the test used (Aaker, Day & Kumar, 2004). Descriptive statistics such as frequency distribution, mean, and standard deviation were used to summarize the data. The mean was used to summarize metric data such as respondent age while frequencies were used to summarize data for categorical variables such as the respondent’s gender and highest level of education. Exploratory Factor analysis (EFA) was conducted to assess the validity of the constructs. Principal Axis Factoring method with a Promax rotation was applied as the constructs that were expected to be related. Items with factor loadings of less than 0.6 and those that were loading onto more than one factor were excluded during factor analysis. Cronbach’s Alpha was computed for each construct that was retained after EFA to assess reliability of the scale of multi-item scales. Reliability describes the extent to which all the items in a multiple item scale measure the same concept or construct. The value of the Cronbach’s Alpha ranges from zero to one and the closer the Cronbach’s alpha coefficient is to 1, the greater the internal consistency of the items in the scale. A value greater than 0.7, is widely accepted in research, values below 0.7 but above 0.5 can still be used (Hair, Black, Babin, & Anderson, 2010). Pearson’s correlation analysis was also conducted to assess the relationship between variables. Correlation measures the strength of a relationship between two variables. A correlation coefficient is weak if it is between 0 and 0.29, moderate if between 0.3 and 0.49, and strong if between 0.5 and 1 (Cohen, 1988). The sign of the correlation coefficient shows the direction of the relationship. A positive correlation means that as one variable increases, the other variable increases as well while a negative correlation coefficient implies that one variable increases as the other one decreases and vice versa. A p-value of the Pearson’s Correlation less than 0.05 implies that the 26 relationships is significant while a p-value greater than 0.05 is an indication of an insignificant relationship. Path analysis was conducted using IBM Amos version 21 to assess the causal relationship among variables. This was used to determine both the direct and indirect impact of various variables on brand loyalty. Path analysis has an advantage of allowing the research to have several dependent variables in one model; this is not possible with multiple regression, which could have been used. The dependent variable had three sub-constructs which rendered path analysis the most suitable analysis technique to establish the impact of the independent variables on the dependent variables. 3.7 Validity and reliability of research 3.7.1 Validity Scale validity refers to the extent to which a scale measures what it is intended to measure. Factor analysis is used to test scale validity by incorporating Eigenvalues to discriminate between scale items in terms of their respective factor loadings. Items that are assigned to a particular factor with loadings of above 0.5 should be retained, provided that these are grouped with other items of a similar nature (Distefano, Zhu and Mindrila, (2009). Scale items that load high on one factor are deemed to be measuring the same construct as other scale items that load similarly high on the same factor. Items loading below 0.5 and assigned to a second factor should be removed from the initial scale as, in such cases, uni-dimensionality of the scale is hindered (Hair et al., 2010). Convergent validity was examined by using item loadings, item-to-total correlation values and average variance as indicators. 3.7.2 Reliability Scale reliability refers to the internal consistency [of the scale] and the extent to which a questionnaire correlates with itself based on different respondent characteristics (Corbishley, 2017). Cronbach coefficient alpha was used to test for scale reliability. 27 Scale items with alphas lower than 0.7 should be removed from the original scale. The researcher also ran a summative scale to increase reliability. 3.7.3 Ethical Consideration Ethical considerations were considered during all phases of the research. To ensure that all participants were sufficiently informed to provide meaningful consent, a cover letter explaining the research was issued at the beginning of the questionnaire. The online questionnaire was set up so as not to allow the participants to answer any questions until they had confirmed that they had read the cover page. The cover page clearly stated that: - all responses remained strictly anonymous; - respondents were not required to supply their names or identity; - participation in the research was completely voluntary; - respondents could withdraw from completing the questionnaire at any time; and - responses to the questionnaire were for research purposes only. The participants were not subject to any physical or psychological stress or other harm during the research process. The researcher’s contact details were also issued to ensure transparency of the research; the respondents were able to contact the researcher should they have had any further questions regarding the research. Once the data was obtained, it has been securely stored and archived in a password protected computer and will be deleted after one year. 28 CHAPTER 4: PRESENTATION OF THE RESEARCH RESULTS 4.1 Introduction This chapter presents the findings of the statistical analysis of the research. Sample characteristics collected from the respondents is presented first. The researcher then unconventionally presents the testing of the model fit, this then presents a springboard to present the rest of the statistical analysis and results carried out. 4.2 Data screening A total of 203 responses were received. Of the 203 responses, three responses were incomplete and were therefore excluded during data cleaning. Accordingly, 200 responses were analysed. 4.3 Sample Characteristics / Demographic 4.3.1 Gender The gender distribution, illustrated in Figure 1, indicates that most of the respondents were female (65% female, as compared to 34% male and 1% non-binary/ third gender). Figure 2: Respondent gender 34% 65% 1% Male Female Non-binary / third gender 29 4.3.2 Age The respondents were on average 40.61 ± 12.330 years old with a range of (19,69) years. Not all respondents indicated their age; 187 respondents indicated their age, while the other 13 respondents elected not to disclose this information. Table 2: Respondent age distribution N Minimum Maximum Mean Std. Deviation Age 187 19.00 69.00 40.61 12.330 4.3.3 Level of education The results presented in Figure 3 show that only 7% of the respondents had high school as their highest attained level of education. 32% of the respondents had earned Diplomas, 32% had earned degrees and 30% had earned post-graduate degrees. These results indicate that the data may represent an education bias and the respondents are clearly highly educated. Figure 3: Highest level of education 7% 32% 32% 30% 0% 5% 10% 15% 20% 25% 30% 35% High school Diploma University degree Post-Graduate degree P e rc e n ta g e o f re s p o n d e n ts Level 30 4.3.4 Income Level Figure 4 shows the respondent’s monthly income levels. One in every three respondents earned R50 000 or above. Only 8% of the respondents earned less than R15 000 per month. This data shows that the respondents are from a very high earning group within the South African population. Figure 4: Income levels 4.4 Discussion regarding Sample Characteristics / Demographic The population sample by way of the convenience sampling technique presented some bias. There is a clear educational and income bias. However, the bias presented offer interesting insights for retailers. Conventionally, South African marketers have used the segmentation tool called the Living Standards Measure (LSM). This tool was developed by the South African Advertising Research Foundation (SAARF) and segmented the South African population into different groupings based on household’s degree of material items owned and urbanization. This tool was introduced in the 1980s and as such, due to fast changes in the South African society, the tool is no longer considered the appropriate population segmentation tool (Langschmidt, 2017). In 2017, a new measure was developed, called the Socio-economic Measure (SEM). This measurement is said to be more accurate as it segments the South African 33% 9% 8% 21% 12% 10% 4% 4% 0% 10% 20% 30% 40% Above R50 000 R45 000 – R50 000 R40 000 – R44 999 R24 999 – R39 999 R20 000 – R24 999 R15 000 – R19 999 R10 000 – R14 999 R5 000-R10 000 Percentage of respondents M o n th ly i n c o m e 31 population more accurately based on income, lifestyle, geography, race and more. Based on the way SEM segments the South African population, this tool provides a more realistic picture into South Africans lifestyle and provides marketers with greater insight into targeting these segmentations accordingly (Langschmidt, 2017). One of the SEM groupings called SEM 8-10 group corresponds with the sample demographic presented in this study. This group is said to be the most educated, constituting 19.3% of the South African population. Characteristics of this grouping includes educated people who earn approximately 2.5 times more than the average South African household. The average age of these SEM 8-10 households is 42 years of age which is very close to the age of the average age of the respondents who answered this study’s questionnaire. This group is at the highest end of the SEM groups as they enjoy the best standard of living (Reid, 22 February 2018). 4.5 Model fit The results on the model fit indices shows that the RSME, NFI, NNFI (TLI), and CFI indices were within acceptable ranges. The AGFI and the GFI were slightly below the acceptable range. Since most of the indices were met, it can be noted that the model represents a good fit for the data. Further pruning of the model did not improve the indices. Less stringent goodness of fit indices that can be applied were AGFI and values that are required to be ≥ 0.8 (Ishiyaku, Kasim, & Harir, 2017). The model fit indices are shown in the tables below. Table 3: Absolute Fit Indexes Absolute Fit Indices Acceptable Value Reference Value Outcome GFI >0.9 MacCallum and Hong (1997) 0. 889 Slightly below acceptable range RMSEA <0.08 Steiger (1990) 0.074 Acceptable NFI >0.9 Bentler (1992) 0.922 Acceptable CFI >0.9 Gerbing, and Anderson (1992) 0.957 Acceptable Absolute Fit Indexes Acceptable Value Value Outcome 32 GFI >0.9 0.889 Slightly below acceptable range AGFI >0.9 0.843 Slightly below acceptable range RSME RSMEA<0.08 0.074 Acceptable NFI >0.9 0.922 Acceptable NNFI (TLI) >0.9 0.946 Acceptable CFI >0.9 0.957 Acceptable CMIN /DF < 5 2.094 Acceptable 4.6 Validity Exploratory factor analysis (EFA) was conducted to assess the validity of the constructs. All items from all the constructs were loaded to assess how they will be classified by EFA. The results are presented below. The Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy value was 0.863 as shown in Table 5. This was higher than the required minimum of 0.5, thus the sample was adequate to run EFA (Exploratory factor analysis). As is also reflected in Table 5, the Bartlett's Test of Sphericity had a p-value of 0.000. This indicates that the statistic was significant, which implies that the items were sufficiently strongly correlated to enable EFA to be conducted. Table 4: KMO and Bartlett's Test Kaiser-Meyer-Olkin Measure of Sampling Adequacy. 863 Bartlett's Test of Sphericity Approx. Chi-Square 2827.729 df 136 Sig. .000 The total variance presented in Table 6 shows that there were 6 factors extracted after running EFA. The retained 6 factors explained 79.130% of variation in the initial 24 items retained in the 6 constructs. This is after eliminating all the items that had a factor loading less than 0.6. 33 Table 5: Total Variance Explained Facto r Initial Eigenvalues Extraction Sums of Squared Loadings Rotation Sums of Squared Loadingsa Total % of Variance Cumulativ e % Total % of Variance Cumulativ e % Total 1 6.969 40.992 40.992 6.731 39.596 39.596 4.772 2 3.045 17.909 58.901 2.822 16.602 56.197 4.393 3 1.526 8.976 67.876 1.307 7.687 63.884 4.770 4 1.213 7.135 75.012 .882 5.188 69.073 3.956 5 1.064 6.260 81.272 .724 4.256 73.329 3.998 6 .573 3.368 84.640 7 .487 2.865 87.505 8 .433 2.550 90.055 9 .320 1.882 91.937 10 .299 1.758 93.695 11 .282 1.658 95.352 12 .214 1.258 96.611 13 .178 1.046 97.656 14 .148 .868 98.525 15 .119 .701 99.226 16 .075 .443 99.668 17 .056 .332 100.000 Extraction Method: Principal Axis Factoring. a. When factors are correlated, sums of squared loadings cannot be added to obtain a total variance. 34 The scree plot shown in Figure 5 illustrated that there were 5 factors that explained most of the variance in the initial factors. This is because the line graph flattened after factors. This confirms the results shown in table 3. Figure 5: Scree plot The final construct competition after factor analysis is presented in table 7. Each item loaded highly one factor and all the items that were loading on more than one factor we excluded during EFA. Items with factor loadings less than 0.5 were excluded as well. The results show that there were 5 valid constructs. The retained constructs were Programme Social Benefits, Brand loyalty, Programme Loyalty, Customization, and Programme Value. 35 Table 6: Pattern Matrix Pattern Matrixa Factor 1 2 3 4 5 Program Social Benefits PSB3 I am recognized by certain employees. .964 PSB1 I have developed a friendship with the staff at this retailer. .963 PSB2 I am familiar with the employees who perform the service. .947 PSB4 The staff at this retailer know my name. .828 Brand loyalty CS5 Based on all of your experience with this retailer, how satisfied are you? .944 CS6 Based on all of my experience I am .883 BL1 I would repurchase products and services from this retailer. .711 BL4 How likely are you to continue to purchase goods from this retailer? .679 Program Loyalty PL3 I have a strong preference for this loyalty program. .933 PL2 I would recommend this loyalty program to others. .915 PL1 I like this loyalty program more than other programs. .910 Customizatio n CZ1 This retailer offers products and services that I am not able to find at another retailer. .914 CZ2 If I changed retailers, I wouldn't be able to obtain products and services that are as customized as those I can access at this retailer. .692 CZ3 This retailer offers me products and services that satisfy my specific needs. .550 Program Value PV1 This loyalty program is easy to use. .738 PV3 It is highly likely that I will be eligible for the rewards of this loyalty program. .737 PV2 This loyalty program's rewards are what I want. .676 Extraction Method: Principal Axis Factoring. Rotation Method: Promax with Kaiser Normalization. a. Rotation converged in 6 iterations. 36 Confirmatory factor analysis (CFA) was also conducted using IMB Amos version 21 to assess the validity of the proposed model using factors retained during EFA. The composite reliability (CR) measure was used to assess the reliability while Average variance Extracted (AVE) was used to assess convergent validity and discriminant validity was assessed using the maximum shared variance (MSV). The model with all the constructs and items retained during EFA show in Figure 5. Figure 6: CFA Model of New model 37 The model shows that all the items loads highly onto their respective factors and the model indices are summarise in Table 5. 4.7 Reliability Reliability was assessed by running Cronbach’s Alpha for each of the factors retained after EFA. The results presented in Table 5 indicates that Program Social Benefits (4 items, α = 0.962), Loyalty Schemes - Program Loyalty (3 items, α = 0.940), Brand Loyalty - Retailer Commitment (5 items, α = 0.907), and Program Loyalty (3 items, α = 0.940) had excellent reliability level as they had Cronbach’s Alpha values greater than 0.9. There was good reliability for Brand loyalty (4 items, α = 0.875) and Customization (3 items, α = 0.817) as the values was greater than 0.8, while Program Value (3 items, α = 0.770) had acceptable reliability with values greater than 0.7. The reliability level was thus acceptable for all constructs retained after EFA. A composite scale was computed for each construct since all constructs were valid and reliable. This was done by calculating the average of the items retained with each construct. This means that each construct can have a lowest score of 1 (strongly disagree) and a highest possible of 5 (strongly agree) Table 7: Reliability of scale Construct/Sub-construct N of Items Cronbach's Alpha Reliability level Program Social Benefits 4 0.962 Excellent Program Loyalty 3 0.940 Excellent Brand loyalty 4 0.875 Good Customization 3 0.817 Good Programme Value 3 0.770 Acceptable 38 4.8 Correlation analysis The descriptive statistics shown in table 8 indicates that Brand Loyalty (mean = 5.92 ± 0.95) was the highest rated construct, followed by Programme Value (mean = 5.38 ± 1.05) and Programme Loyalty (mean = 5.04 ± 1.47). Programme Social Benefits construct was (mean = 3.43 ± 1.80) was the lowest rated construct. The correlation coefficients shows that there was no risk of multicollinearity as all the correlations were less than 0.8. Table 8: Descriptive Statistics and Pearson Correlation *. Correlation is significant at the 0.05 level (2-tailed). **. Correlation is significant at the 0.01 level (2-tailed). Descriptive Statistics Pearson Correlation Mean Std. Deviation 1 2 3 4 5 Program Social Benefits 3.43 1.80 1 Program Loyalty 5.04 1.47 .393** 1 Customization 4.96 1.34 .503** .425** 1 Program Value 5.38 1.05 .277** .523** .410** 1 Brand Loyalty 5.92 0.95 .133 .444** .404** .475** 1 4.9 Hypothesis Testing Path analysis was conducted using IBM Amos version 21 to assess the hypotheses. The Path analysis model is presented in Figure 6. A Structural equation modelling (SEM) model was fitted with the items that were retained in the CFA mode. The model is presented below. 39 Figure 7: Detailed SEM analysis 40 Figure 8: Simplified SEM analysis The regression weights, t-values and p-values are presented in Table 9. It can be noted that Programme Social Benefits, Programme Value, and Customization explained 43.5% of Programme Loyalty. While Programme Loyalty and Customization explained 22.9% of variation in Brand Loyalty. Table 9: SEM Model - Regression Weights ***, p-value < 0.001 Estima tes Standardi zed Estimates T- valu e P- valu e R- Squar e Programme Social Benefits Program Loyalty ,120 ,167 2,57 6 ,010 0.435 Programme Value Program Loyalty ,971 ,568 6,72 3 *** Customization Program Loyalty ,124 ,135 2,11 5 ,034 Programme Loyalty Brand Loyalty ,185 ,340 4,49 0 *** 0.229 Customization Brand Loyalty ,147 ,293 3,69 4 *** 41 Estima tes Standardi zed Estimates T- valu e P- valu e R- Squar e SEM Fit Indices: 𝑥2 = 318.628 (112); 𝑥2/df = 2.354; RMSEA = .096 CFI = .925; TLI = .909; GFI = .858; NFI = .890, The CFI, TLI, and CMIN/DF were acceptable as per the stringent goodness of fit cut-off points while the GFI, RMSEA and the NFI were acceptable as the stringent cut-offs (Ishiyaku, Kasim, & Harir, 2017). Results pertaining to hypothesis 1 (H1): There is a positive relationship between Loyalty Programme value and Programme Loyalty. H0: There is a no relationship between Loyalty Programme value and Programme Loyalty H1: There is a positive relationship between Loyalty Programme value and Programme Loyalty. The results presented in Table 9 show that Loyalty Programme value (β = 0.568, t=6.723, p-value < 0.001) had a positive significant impact on Programme Loyalty. The relationship was positive because the coefficient for Programme value (β = 0.568) was greater than zero. It was significant because the p-value was less than 0.05. This indicates that the null hypothesis was rejected in favour of the alternative hypothesis. It was thus, concluded that there is a positive relationship between Loyalty Programme value and Programme Loyalty. Results pertaining to hypothesis 2 (H2): There is a positive relationship between Programme Social Benefits and Program Loyalty. H0: There is a no relationship between Loyalty Programme value and Programme Loyalty 42 H2: There is a positive relationship between Programme Social Benefits and Programme Loyalty. The results presented in Table 9 show that Programme Social Benefits (β = 0.167, t=2.576, p-value = 0.010) had a positive significant impact on Programme Loyalty. The relationship was positive because the coefficient for Programme Social Benefits (β = 0.167) was greater than zero. It was significant because the p-value was less than 0.05. This indicates that the null hypothesis was rejected in favour of the alternative hypothesis. It was thus, concluded that there is a positive relationship between Programme Social Benefits and Programme Loyalty. Results pertaining to hypothesis 3 (H3): There is a positive relationship between customization and programme loyalty H0: There is no relationship between customization and programme loyalty H3: There is a positive relationship between customization and programme loyalty The results presented in Table 9 show that Customization (β = 0.135, t=2.115, p-value = 0.034) had a positive significant impact on Program Loyalty. The relationship was positive because the coefficient for Customization (β = 0.135) was greater than zero. It was significant because the p-value was less than 0.05. This indicates that the null hypothesis was rejected in favour of the alternative hypothesis. It was thus, concluded that there is a positive relationship between customization and loyalty schemes. Results pertaining to hypothesis 4 (H4): There is a positive relationship between customization and brand loyalty H0: There is no relationship between customization and brand loyalty. 𝐇𝟒: There is a positive relationship between customization and Brand Loyalty The results presented in Table 9 show that customization (β = 0.293, t=0.3694, p-value > 0.001) had a positive and significant impact on Brand Loyalty. The relationship was positive because the coefficient for Customization (β = 0.293) was greater than zero. The 43 relationship was significant because the p-value was less than 0.054. It is thus concluded that there is a positive relationship between customization and brand loyalty Results pertaining to hypothesis 5 (H5): There is a positive relationship between programme loyalty and brand loyalty H0: There is no relationship between programme loyalty and brand loyalty. H5: There is a positive relationship between programme loyalty and brand loyalty. The results presented in Table 7 show that Program Loyalty (β = 0.340, t=4.490, p- value < 0.001) had a positive significant impact on Brand Loyalty. The relationship was positive because the coefficient for Programme Loyalty (β = 0.340) was greater than zero. It was significant because the p-value was less than 0.05. This indicates that the null hypothesis was rejected in favour of the alternative hypothesis. It was thus concluded there is a positive relationship between loyalty schemes and brand loyalty. 4.10 Summary of hypotheses Table 10: Summary of all hypotheses Outcome H1 There is a positive relationship between Loyalty Programme value and Program Loyalty. Supported and Significant H2 There is a positive relationship between Programme Social Benefits and Program Loyalty. Supported and Significant H3 There is a positive relationship between customization and loyalty schemes Supported and Significant H4 There is a positive relationship between customization and brand loyalty Supported and Significant H5 There is a positive relationship between loyalty schemes and brand loyalty Supported and Significant 44 CHAPTER 5: RESEARCH DISCUSSION 5.1 Introduction The purpose of this chapter of this study is to delineate, analyse, discuss, and evaluate the results presented in the preceding chapter. The primary and secondary objectives of this paper are revisited, and the final section of this chapter outlines theoretical and managerial implications of the research. Also, suggestions for future research are emphasized to broaden academic understanding pertaining to brand loyalty. 5.2 Discussion of the research problem The researcher sought to determine the effect of loyalty schemes on long term brand loyalty. Research question: Can loyalty schemes influence brand loyalty directly? The research problem was derived as a result of an extensive review on the available literature pertaining to loyalty schemes. This is supported since all four constructs of programme loyalty schemes namely, (1) Programme Loyalty, Loyalty Programme Value, Programme loyalty and customization all had a positive and significant impact on Brand Loyalty. Thus, this research question was supported. 5.3 Sub-Problems The researcher sought to determine secondary problems which are outlined below. What drives consumers to be loyal to a loyalty programme? This was addressed by examining programme perceived value and programme social benefits. This statement can be supported by arguing that social benefits and perceived value do drive loyalty to a loyalty program as both these constructs were found to be both positive and significant. 45 Is there a relationship between programme loyalty and customization? This is supported as customization was found to have a significant and positive effect on programme loyalty. Is there a relationship between customization and brand loyalty? This is supported as this relationship proved to be positive and significant. 5.4 Discussion regarding the Sample Characteristics / Demographic The study, due to the use of a convenience sampling method, lead to the population sample bias towards to an older group of consumer, the mean age was 40.16. while it was not the researcher’s intention to sample particularly older adult consumers, it is noteworthy to discuss previous authors findings on the study of an older population sample. The literature suggested that while older populations may have problems with changing their purchasing habits (Pillemer et al., 2011). Other research suggests that an older adult was more likely to be influenced by affection rather than cognition. It has been suggested that marketers should design products that better satisfy this particular population by enhancing affect connections (Deepraj et al., 2018). 5.5 Evaluating the results based on the retailer and loyalty scheme chosen by the respondents One way analysis of variance was conducted to assess whether the constructs were rated differently across the retainers. Multiple comparison was conducted where there were significant differences by retailer to ascertain the retailers that were rated statistically differently. The results revealed that the rating of Customization (p-value = 0.000) differed significantly by retailer. It can be noted that Woolworths WRewards (mean = 5.57) was the highest rated and was rated significantly higher than both Pick n Pay Smart Shopper (mean = 4.31, p-value = 0.000) and Edgars Thank U (mean = 4.98p-value = 0.000). The second highest rated in terms of customization was Dischem Benefit (mean = 5.30), this 46 was also significantly higher than Pick n Pay Smart Shopper (p= 0.001) and Clicks Clubcard (mean = 4.51, p-value = 0.010). Edgars Thank U (mean = 4.98) was third rated and was significantly higher than Pick n Pay Smart Shopper (p= 0.042). There was no significant difference among the other retailer combinations. The results also showed that Brand Loyalty rating (p-value = 0.006) differed significantly per retailer. Dischem Benefit (mean = 6.17) was the highest rated and was rated significantly higher than all the other retailers; Pick n Pay Smart Shopper (mean = 5.73, p-value = 0.032), and Edgars Thank U (mean = 5.43, p-value = 0.003). The second highest rated was Woolworths WRewards (mean = 6.12), which was significantly higher than Pick n Pay Smart Shopper (p-value =0.002) and Clicks Clubcard (p-value 0.020). The third best rated retailer was Clicks Clubcard (mean = 5.99) and it was significantly higher than Edgars Thank U (p-value = 0.019). Edgars Thank U was rated the lowest and was significantly other than all other retailers except Pick n Pay Smart Shopper (p-value = 0.194). Table 11: Summary statistics One Way ANOVA N Mean Std. Deviation P-value Program Social Benefits Pick n Pay Smart Shopper 46 3.26 1.73 .108 Woolworths WRewards 65 3.26 1.67 Clicks Clubcard 37 3.22 1.86 Dischem Benefit 29 4.13 1.83 Edgars Thank U 20 3.96 2.10 Total 197 3.45 1.81 Program Loyalty Pick n Pay Smart Shopper 46 5.02 1.23 .412 Woolworths WRewards 65 4.92 1.52 Clicks Clubcard 37 5.20 1.51 Dischem Benefit 29 5.47 1.28 Edgars Thank U 20 4.80 1.76 Total 197 5.06 1.45 Customization Pick n Pay Smart Shopper 46 4.31 1.26 .000 Woolworths WRewards 65 5.57 0.94 47 Clicks Clubcard 37 4.51 1.47 Dischem Benefit 29 5.30 1.33 Edgars Thank U 20 4.98 1.30 Total 197 4.98 1.32 Programme Value Pick n Pay Smart Shopper 46 5.22 1.05 .519 Woolworths WRewards 65 5.35 1.01 Clicks Clubcard 37 5.62 0.90 Dischem Benefit 29 5.47 1.20 Edgars Thank U 20 5.42 1.16 Total 197 5.40 1.04 Brand Loyalty Pick n Pay Smart Shopper 46 5.73 0.90 .006 Woolworths WRewards 65 6.12 0.73 Clicks Clubcard 37 5.99 0.64 Dischem Benefit 29 6.17 0.76 Edgars Thank U 20 5.43 1.50 Total 197 5.94 0.89 Multiple Comparisons LSD Dependent Variable Mean Difference (I-J) Sig. Customization Pick n Pay Smart Shopper Woolworths WRewards -1.263* .000 Clicks Clubcard -.202 .456 Dischem Benefit -.987* .001 Edgars Thank U -.672* .042 Woolworths WRewards Pick n Pay Smart Shopper 1.263* .000 Clicks Clubcard 1.061* .000 Dischem Benefit .276 .315 Edgars Thank U .591 .061 Clicks Clubcard Pick n Pay Smart Shopper .202 .456 Woolworths WRewards -1.061* .000 Dischem Benefit -.785* .010 Edgars Thank U -.470 .168 Dischem Benefit Pick n Pay Smart Shopper .987* .001 Woolworths WRewards -.276 .315 Clicks Clubcard .785* .010 Edgars Thank U .316 .376 Edgars Thank U Pick n Pay Smart Shopper .672* .042 Woolworths WRewards -.591 .061 Clicks Clubcard .470 .168 48 Dischem Benefit -.316 .376 Brand Loyalty Pick n Pay Smart Shopper Woolworths WRewards -.391* .020 Clicks Clubcard -.265 .169 Dischem Benefit -.444* .032 Edgars Thank U .303 .194 Woolworths WRewards Pick n Pay Smart Shopper .391* .020 Clicks Clubcard .126 .482 Dischem Benefit -.053 .784 Edgars Thank U .694* .002 Clicks Clubcard Pick n Pay Smart Shopper .265 .169 Woolworths WRewards -.126 .482 Dischem Benefit -.179 .406 Edgars Thank U .568* .019 Dischem Benefit Pick n Pay Smart Shopper .444* .032 Woolworths WRewards .053 .784 Clicks Clubcard .179 .406 Edgars Thank U .747* .003 Edgars Thank U Pick n Pay Smart Shopper -.303 .194 Woolworths WRewards -.694* .002 Clicks Clubcard -.568* .019 Dischem Benefit -.747* .003 5.6 Discussion pertaining to hypotheses 5.6.1 Hypothesis 1 H10: There is no relationship between loyalty programme value and loyalty programmes H1A: There is a positive relationship between loyalty programme value and loyalty programmes As discussed in the previous chapter, the null hypothesis was rejected in favour of the alternate hypothesis as this study found that there is a positive and significant relationship between programme value and loyalty to the loyalty programme. The following constructs were included in loyalty programme value: - It is highly likely that I will be eligible for the rewards of this loyalty programme. - This loyalty programme is easy to use. 49 - This loyalty programme's rewards are what I want. In line with Yi and Jeon (2003), the researcher found the relationship between programme value and brand loyalty to be positive. This is also aligned to previous empirical evidence found by Koo, Yu, and Han (2020), Ball et al. (2006), and Grace and O’Cass (2005). Evanschitzky et al. (2012) argue that this finding creates significant implications in practice. If the loyalty scheme can offer consumers’ value, then this may encourage purchases from the retailer from customers who previously had no desire to purchase from the particular retailer. In other words, the perceived value derived from the loyalty scheme may offset the negative feelings towards the retailer. 5.6.2 Hypothesis 2 H20: There is no relationship between programme social benefits and programme loyalty H2A: There is a positive relationship between programme social benefits and programme loyalty As discussed in the previous chapter, the null hypothesis was rejected in favour of the alternate hypothesis as this study found that there is a positive and significant relationship between programme social benefits and loyalty to the loyalty programme The positive relationship found between programme special benefits and retailer commitment is aligned to previous research carried out by Gwinner, Gremler, and Bitner (1998) who reported that customers value special treatment benefits. These findings contradict the results of Hennig-Thurau, Gwinner and Gremler (2002), who do not find special treatment benefits to have a significant direct influence on customer loyalty, and only a small indirect influence on word-of-mouth communication via commitment was found. 50 The positive relationship found between programme loyalty and social benefits of the retailer are closely related concepts that have both been proved to influence loyalty positively. The constructs included in social benefits of the retailer include the following: - I am familiar with the employees who perform the service - I have developed a friendship with the staff at this retailer - I am recognized by certain employees. - The staff at this retailer know my name This relationship is said to be positive as social bonds lead to an increase in loyalty. Direct connections between social benefits and loyalty have been empirically tested and are discussed in the literature. Hennig-Thurau, Gwinner, and Gremler (2002) discuss that social benefits allow consumers to feel a part of a community and hence, this relationship is fostered on feelings on togetherness and belonging rather than on a transaction. This theory is in line with Oliver (1999) who persists that consumers who are ‘a part’ of a brand are more likely to be loyal to the brand, for example, a club, membership, etc. 5.6.3 Hypothesis 3 & 4 H30: There is no relationship between customization and programme loyalty H3A: There is a positive relationship between customization and programme loyalty As discussed in the previous chapter, the null hypothesis was rejected in favour of the alternate hypothesis as this study found that there is a positive and significant relationship between customization and loyalty to the loyalty programme H40: There is no relationship between customization and brand loyalty H4A: There is a positive relationship between customization and brand loyalty As discussed in the previous chapter, the null hypothesis was rejected in favour of the alternate hypothesis as this study found that there is a positive and significant relationship between customization and brand loyalty 51 Customization was found to have a positive and significant impact on both constructs of loyalty in this study. Customization involves products or services that appeal to customers on a personal level. The theory is that customization strengthens the emotional bond relationship between the consumer and the brand, thereby leading to repeat purchases over time (Fels, Falk, & Schmitt, 2017). In this context, the more customized the loyalty scheme, the higher the usage will be. The constructs included in customization include the following: - This retailer offers products and services that I am not able to find at another retailer - If I changed retailers, I wouldn’t be able to obtain products and services that are as customizes as those I can access at this retailer - This retailer offers me products and services that satisfy my specific needs. The results were akin to previous studies, whose authors also stressed the importance of customization. Dorotic et al. (2012) persist that loyalty programme success lies in the leveraging of customer data to introduce personalized, targeted campaigns to increase sales. Vargo and Lusch (2004) suggest that marketing goals of firms should include customization rather than standardization. This research is in support of that. Dorotic, Bijmolt, and Verhoef (2012) hold the view that the future of loyalty schemes lies in customization of the actual scheme and the potential benefits offered to the customer. 5.6.4 Hypothesis 5 H50: There is no relationship between programme loyalty and brand loyalty H5A: There is a positive relationship between programme loyalty and brand loyalty As discussed in the previous chapter, the null hypothesis was rejected in favour of the alternate hypothesis as this study found that there is a positive and significant relationship between loyalty to the loyalty programme and brand loyalty 52 The constructs included in programme loyalty include: - I have a strong preference for this loyalty programme - I would recommend this loyalty program to others - I like this loyalty programme more than other programmes The constructs included in brand loyalty include: - I would repurchase products and services from this retailer - How likely are you to continue to purchase goods from this retailer - Based on all your experience with this retailer, how satisfied are you? - Based on all of my experience I am (how satisfied) The satisfaction constructs were included in the final constructs as there is empirical evidence supporting the strongly linked relationship between satisfaction and brand loyalty. Customer satisfaction is defined consistently across the literature as a consumer’s post-consumption evaluation. However, this definition has evolved in the literature over time; it is no longer considered transaction-specific. Lombart (2017) cites authors to explain that satisfaction is a fundamental antecedent of a consumer’s long-term behavior. It is because of this that the satisfaction-loyalty relationship is central to marketing theory and practice. This construct included items such as: staff are always willing to assist customers, I am extremely happy when the quality delivered is of a high standard, and a rating of the consumer’s satisfaction level, to name a few. The finding that customer satisfaction leads to a positive increase in a consumer’s commitment to a retailer and subsequent brand loyalty is aligned with previous research. A previous study completed in the South African retail industry found customer satisfaction to be a predictor of customer trust, customer loyalty and customer repurchase intention (Chinomona & Sandada, 2013). Retailer commitment has been defined as the emotional attachment to the retailer and programme loyalty is defined as a consumer’s favorable attitude towards the loyalty scheme. The research results above are significant as this indicates that creating loyalty 53 towards to the loyalty scheme can result in creating loyalty towards this retailer. This result was the main concern evaluated in the study conducted by Evanschitzky et al. (2012). There is extensive research supporting the notion that loyalty schemes create loyalty towards the programme and not to the retailer or brand (Dowling & Uncles 1997; Meyer- Waarden 2007; Yi & Jeon 2003). The findings presented in this study contradict the previous literature and offer a more positive scenario. These results are consistent with the study conducted by Pandit and Vilches-Montero (2016), where they persist that their novel findings represent an inexplicable link between commitment to an object (i.e.., the loyalty card) translates into an attachment with an object (the retail store). They too, propose that attachment and commitment should be used as a strategy to create positive consumer responses 5.7 Demographic data and share of wallet One way analysis of variance was conducted to assess whether the average percentage share of wallet differed significantly by retailer with multiple comparison to ascertain the retailers that were statistically differently. The percentage of the wallet was calculated by dividing the amount spent by a respondent on a retailer by the mid-point of their salary range. 5.7.1 Drivers’ Share of Wallet A regression model with Share of Wallet as the dependent variable and Programme Value, Programme Social Benefits, and Customization as the independent variables was fitted. The results presented in the model summary indicates that Programme Value, Programme Social Benefits, and Customization explained the 9.4% of variation in Share of Wallet as indicated by an r-square of 0.094. 54 Table 12: Model summary Model Summary Model R R Square Adjusted R Square Std. Error of the Estimate 1 .307a .094 .080 7.36234 a. Predictors: (Constant), Program Value, Program Social Benefits, Customization The model was valid since the p-value was less than 0.05 (p-value = 0.000). Table 13: ANOVA ANOVAa Model Sum of Squares df Mean Square F Sig. 1 Regression 1096.563 3 365.521 6.743 .000b Residual 10515.591 194 54.2