Does Event Sponsoring Pay Off?

(1)

Does Event Sponsoring Pay Off?

The Effect of Social Media Activity and Involvement in Event

Sponsorships on Brand Loyalty

Sophie Kromhof

University of Groningen

Faculty of Economics and Business

MSc. Marketing Intelligence & Msc Marketing Management

Master Thesis

June 17, 2016

Eerste Helmersstraat 184-4V

1054 EL Amsterdam

Tel: +31(0)616173387

Email:

s.l.kromhof@student.rug.nl

Student number: S2005379

Supervisors

University of Groningen

(2)

2

University of Groningen

ABSTRACT

The global spending on sponsorships is still increasing each year in which the sponsoring of sports related events plays the most important role. This significant growth in global spending of all sponsorships, combined with the growing pressure on proving marketing investment decisions, triggered an increased focus on assessing the effectiveness of sponsorships as a marketing communication tool.

Literature on sponsorships has focused almost exclusively on evaluation outcomes such as brand image, brand recall and brand awareness in a sports sponsorship setting. However, assessing the effect of sponsorships on brand loyalty is still being called for by various scholars. Moreover, there is lack of incorporation of social media effects in evaluating sponsorship effects in current literature. Furthermore, it has been well established that effects of involvement in a sports event has a positive influence on brand recognition of the sponsor. However, research on event involvement has been limited to sports event which neglects the sponsorship between a brand and a cultural event (e.g., singing competition or arts event). Therefore, the aim of this research is to assess the effects of sponsorship on brand loyalty in a sponsorship setting between a cultural event and a telecommunication brand. This research takes the next step in sponsorship evaluation by investigating the possible effects of social media activity and event involvement within a sponsorship context. “Does sponsorship pay off?” is a fundamental question in established sponsorship literature. However, this study argues that a new generation of research in this area is needed and therefore this study addresses the following main research question: “does sponsoring of cultural events have an impact on brand loyalty accounting for the effect of social media interaction and event involvement?”

(3)

3

(4)

4 PREFACE

During my Msc. Marketing Management & Intelligence program I discovered my passion for brand marketing however, I also discovered my genuine interest for marketing intelligence. I studied at the University of Groningen since 2010 and I had the time of my life. The last two years being a student at the University of Groningen was fun but also required a lot of hard work. Moreover, combining my thesis with an internship at a large telecommunication operator, I am certain that I would like to develop myself as a true marketeer in my future career.

I could not have achieved the end-result of this thesis without the help of a few motivating people. Therefore, I would like to thank my supervisor dr. J.T. Bouma for providing me the opportunity to apply for an internship and for his insightful feedback. Moreover, I would like to dr. F. Eggers for providing me with useful feedback and steering me in the right direction. Furthermore, I would like to thank my supervisors at the telco company where I have enjoyed my time as an intern. Lastly, I would like to thank my parents for supporting my decisions regarding my study career.

Sophie Kromhof

Amsterdam

(5)

5

6

3.3.2 Brand Affect ... 25

3.3.3 Attitude toward Sponsorship ... 25

3.3.4 Attitude toward Event ... 25

3.3.5 Event Involvement ... 26

3.3.6 Social Media Activity regarding the Event ... 26

3.3.7 Control Variables ... 26 3.4 Method of Analysis ... 26 3.4.1 Factor Analysis ... 26 3.4.2 OLS Regression ... 27 3.4.3 Mediation Model ... 29 3.4.4 Model Fit ... 29 4. RESULTS ... 30 4.1 Characteristics ... 30 4.2 Correlations ... 31 4.3 Factor Analysis ... 32 4.4 Preliminary checks ... 33 4.4.1 Non-Zero Expectations ... 33

4.4.2 Homoscedastic Error Term ... 33

4.4.3 Autocorrelation ... 33

4.4.4 Non-Normality Assumption ... 34

4.4.5 Multicollinearity ... 35

4.5 Model 1 – Direct Effects of Variables on Brand Affect ... 35

4.6 Model 2 – Direct Effects and Moderator Effect of Involvement on Brand Affect ... 36

4.6.1 Moderating Effect of Involvement with the Event and Attitude toward Sponsorship ... 37

4.6.2 Moderating Effect of Involvement with the Event and Attitude toward Event ... 37

4.7 Model 3 – Effect of Brand Affect on Brand Loyalty ... 38

4.8 Comparison Model 1 and Model 2 ... 39

(7)

7

6. LIMITATIONS AND FURTHER RESEARCH ... 47

6.1 Limitations ... 47

6.2 Suggestions for Further Research ... 48

REFERENCES ... 50

APPENDICES ... 59

Appendix A – Timeline survey ... 59

Appendix B - Variables... 59

Appendix C – Factor Analysis... 60

Appendix D - Correlations ... 61

Appendix E – Non-Normal Errors ... 62

(8)

8 1. INTRODUCTION

In 1984, only 2 billion U.S. dollars was spend on global sponsorship. According to Statista 2016, the global sponsorship spending will increase to 60.2 billion U.S. dollars in 2016. The variety of sponsored activities has augmented gradually, though sport related activities remain the most important sector, collecting 68% of total sponsorship spending (International Events Group (IEG), 2010). Just recently, the American shoe brand Nike agreed on the most expensive football sponsorship in history of a value of 153 million U.S. dollars with the football club F.C. Barcelona (IEG, 2016). Another well-known sponsorship is the sponsorship agreement between the beverage brand Gatorade and the National Football League (NFL) in America. Gatorade sponsors the NFL since 1983 and still maintained its status as the most active sponsor of the NFL in 2015. According to IEG research conducted in 2016, the sponsorship revenue for the NFL is still increasing by 4% each year. The significant growth in expenditure of all sponsorships, combined with the growing pressure on proving the accountability of marketing investment decisions, caused an increased focus on evaluating the effectiveness of sponsorship as a marketing communication tool (Donlan, 2013).

(9)

9

As mentioned before, it is argued that many factors could influence the customer reactions to sponsorships. One of these factors that could influence customer perception is involvement with the sponsored event. Pham (1992) proved that effects of involvement with a sports event have a significant positive impact on the recognition of the sponsor of the event. Furthermore, it has been established that involvement with the event has an effect on brand use of the sponsor (Bennett, Ferreira, Lee & Polite, 2009). However, the research on event involvement and sponsorships has been limited to sport events. Therefore, more research is needed to investigate of these findings could be generalizable to cultural events (e.g., singing events and arts events). Grohs & Reisinger (2014) state that involvement with the event in a sponsorship setting needs to be further investigated. Additionally, as called by Gregg et al. (2009) more empirical research is needed to confirm a positive link between involvement and brand use.

Furthermore, in the article by Cornwell & Kwak (2015), the authors highlight several fields, which are linked to sponsorship-marketing, that are in need of further research. One of the areas that need further investigation is the integration of social media and (sport) sponsorships (Cornwell & Kwak, 2015). While there is a lack of formal definition of social media, a rich body of literature aims at establishing a description of social media (Xiang & Gretzel, 2010). Xiang & Gretzel (2010, p.180) describe social media as “internet-based applications that carry consumer-generated content which encompasses media impressions created by consumers typically informed by relevant experience, and archived or shared online for easy access by other impressionable consumers.” Ample research is done regarding the effects of social media on for example, music sales (Dewan & Ramaprasad, 2014), movie sales (Huang, Boh & Goh, 2011; Rui, Liu & Whinston, 2013), travel information search (Chung & Koo, 2015), brand image (Bruhn, Schoenmueller & Schäfer, 2012) and brand loyalty (Erdogmus & Cicek, 2012; Laroche, Habibi & Richard, 2013). However, there is a lack of research that analyses possible interaction effects of social media and sponsorships. Cornwell (2014) supports this statement and calls for further research to incorporate social media in sponsorship topics. Also, Cornwell & Kwak (2015) and Delia & Armstrong (2015) highlight the fact that integration of social media and sponsorships is a key area in which the impact of several social media platforms need to be further investigated and incorporated. Therefore, in this study both involvement with the sponsored event and social media activity will be analysed in a sponsorship setting.

(10)

10

positive impact on brand loyalty (Cobb-Walgren, Ruble and Donthu, 1995; Chaudhuri & Holbrook, 2001; Lin & Wang, 2001; Delgado-Ballester & Munuera-Alemán, 2005). Hence, in this research both sponsorship, social media and customer relationship literature will be used in order to satisfy the need for more comprehensive models of sponsorship effects in a cultural event setting. Therefore, the following main research question is answered:

Does sponsoring of events have an impact on brand loyalty accounting for the effect of social media interaction and event involvement?

In this research, data was collected by distributing an online survey to in total 2415 respondents about a Dutch reality singing competition on a national TV-broadcasting channel sponsored by a large telecom company in the Netherlands. Each respondent was presented with several questions about the Dutch talent event, awareness of its sponsors, several statements about loyalty, purchase intention and likeability of the sponsored brand, brand image, the respondents’ social media activity related to the event, involvement with the sponsored event and several questions about their demographics. To evaluate the effect of the sponsorship on brand loyalty the OLS-regression estimates are evaluated.

The contribution of this paper to the existing sponsorship literature is threefold. Firstly, the main contribution is to integrate the social media aspect of the sponsored event in analyzing brand loyalty in a sponsorship setting. Secondly, this research adds to the current sponsorship literature by accounting for involvement of the customer with the event. Moreover, most sponsorship literature is established in a sports sponsorship setting. This research is conducted in an (cultural) event setting and therefore adds to the little sponsorship literature conducted in a cultural event setting. Lastly, what makes this research even more interesting is the strong managerial relevance. From a managerial point of view, this research aims at uncovering if sponsoring a cultural event increases brand loyalty. Moreover, the interaction between event involvement and sponsorship of cultural events are analysed. If a significant effect is found it could be encouraged to motivate the customers to get involved with the event to increase brand loyalty of the brand that sponsors the event.

(11)

11 2. LITERATURE REVIEW

The purpose of this chapter is to discuss several aspects of sponsorship management. Furthermore, it provides several determinants that influence brand loyalty and brand affect. This chapter concludes with a discussion of relevant control variables and all hypothesized relations will be graphically summarized in a conceptual model.

2.1 Sponsorships

2.1.1 Sponsorship Marketing

(12)

12

away from mass communication “advertising” and toward integrated communication where the brand becomes part of the programming, part of the sharing, and part of the life experience.” As Kevin Spacey said during the Content Marketing World in 2014, people turn away from traditional marketing messages and want to be involved, experience and feel emotionally related to the brand. Cornwell (2014) also makes this distinction between simple sponsoring and sponsorship-linked marketing. Simple sponsoring could be viewed as some exposure of the brand in the sponsored programme or event of which the main goal is to obtain media coverage. Contrary to simple sponsoring, “sponsorship-linked marketing is more often an engagement at the event or involvement with activities that promote the brand” (Cornwell, 2014 p.18). Such consumer-focused sponsorships typically aim to advance the brand awareness, brand attitude, brand image, or behavioural aspects such as purchase intent and increase intent to utilize the brand’s products (Cornwell et al., 2005). In table 1, the most well established papers regarding sponsorship are summarized.

2.1.2 Advantages and Disadvantages of Corporate Sponsorships

(13)

13

competitive advantage and second, if this competitive advantage could be protected and kept explicitly to the sponsored brand. However, Fahy, Farrelly & Quester (2004) raised objections against this statement and concluded that successful sponsorships can create a competitive advantage in the “market of sponsorships”, which successively leads to a sustainable competitive advantage and greater performance in product markets.

Most sponsorship research focuses on modelling and comprehending the process by which sponsorships aid companies to reach their communication objectives (Mazodier & Merunka, 2012). In this conceptual model, which aims at establishing a relationship between attitude toward the sponsorship, attitude toward the event and social media activities regarding the sponsorship, moderated by the event involvement of an individual on brand affect and ultimately brand loyalty.

Table 1 Overview of Well-Established Papers Regarding Sponsorships Author(s) Year Type of

research

Purpose of article Main findings

Meenaghan 1983 Descriptive Consolidates and unites all available knowledge and literature about sponsorship.

Development of a definition of sponsorship marketing.

Meenaghan 1991 Descriptive Examines the development of corporate sponsorship usage as a marketing tool.

Several insights are provided in which sponsorship is proved to be an effective marketing tool. Moreover, several developments in the growing industry of sponsorships are presented.

Javalgi, Traylor, Gross & Lampman

1994 Empirical To investigate the value and effectiveness of sponsorships on corporate image.

Corporate sponsorships does ameliorate corporate image but the effects differ among firms. Furthermore, sponsorship is one of many sources consumers use to construct an impression of a firm.

Gwinner & Eaton

1999 Analytic Gain insight into the brand image factors of sponsorships.

If a fit between the sponsored event and the sponsor's image is present, the image transfer process is enhanced.

Gwinner 1997 Empirical Construct a model which entails the mechanisms by which brand image may be influenced through sponsorships.

Image of the event is associated with brand image however is moderated by a few factors such as degree of similarity, level of sponsorship, event frequency and product involvement.

Cornwell & Maignan

1998 Descriptive Offer a cross-disciplinary review of research conducted on sponsorships.

An extensive review of five research streams of sponsorship was investigated: nature of sponsorships, managerial aspects of sponsorships, measurement of sponsorship effects, strategic use of sponsorships and legal/ethical attentions in sponsorship.

Madrigal 2000 Analytic The purpose of the study is how social alliances present between fans and a favoured sports team affect purchase intentions.

(14)

14

University of Groningen Speed &

Thompson

2000 Analtytic Scholars aim to examine the effect of customers' attitudes about a sponsored sports event and rank them on a multidimensional measure of sponsorship responses.

The main findings are that sponsor-event fit, ubiquity of the sponsor, attitude toward the sponsor and perceived sincerity of the sponsor are of upmost importance generating a favourable response of the customer.

Meenaghan 2001 Descriptive Understanding sponsorship effects by suggesting a model.

The author identified several factors that are important for achieving customer response in sponsorships. The factors are goodwill in sponsorship, process of image transfer and concept of fan involvement.

Cornwell, Roy & Steinard

2001 Analytic Exploration of how managers perceive the brand-equity building capabilities of their sponsorship-linked marketing activities.

Active management of sponsorships takes part in the difficult task in differentiating the brand from its competitors and add financial value to the brand.

Walliser 2003 Descriptive Extends and updates the existing review of eighty studies on sponsorship.

The author reviewed 230 studies about sponsorship and thereby claiming that literature about sponsorship is in its growth phase. Still sponsorship impact (such as awareness and image) and different areas of sponsorship need more attention from scholars.

Cornwell, Weeks & Roy

2005 Descriptive Exploration of theoretical explanations of how sponsorship works.

Bringing together important factors, both individually or group related, as well as market and management factors, with the purpose of understanding its influence on sponsorship mechanisms and outcomes.

Simmons & Becker-Olsen

2006 Analytic Investigate the effects of sponsorships regarding corporate social responsibility.

The scholars show that the fit between a companies' specific associations and a sponsored cause can enforces or blur the companies positioning, influence liking of the sponsorship, and influence firm equity.

Barros, Santos & Chadwick

2007 Analytic Offer distinctive methods to evaluate the relationship between sponsorship and brand recall at a large sports event.

The authors proposed a framework for the evaluation of the Euro 2004 soccer tournament sponsorships and the justification of these activities.

Sirgy, Lee, Johar & Tidwell

2007 Analytic Extend self-image congruence studies into the corporate sponsorship literature regarding marketing communications.

The authors concluded that self-congruity with the sponsored sports event has a positive influence on brand loyalty.

Sözer & Vardar

2009 Emperical Aim of the paper is to provide brand managers a method how to leverage the equity of their brands through event sponsorship

Find a fit between the sponsored event and the firm you are representing. It will enhance the brand image of the firm.

Olson 2010 Analytic Create a complete model of high-level sponsorship effects which is suitable for both sports and cultural sponsorship contexts.

Developed a model for both sponsorship contexts in which sincerity and sponsorship attitude both influence sponsor equity.

Mazodier & Merunka

2012 Analytic Models and demonstrates the impact of sponsorship on brand loyalty.

Sponsorship has a positive influence on brand trust and brand affect. Moreover, the research uncovers that there are changes in perception of brand trust and brand loyalty after viewing the sponsorship.

Cornwell 2014 Descriptive Provide an extensive review of how sponsorships work.

(15)

15 2.2 Attitude toward the Sponsorship

Speed & Thompson (2010) show that the attitude toward the event and sponsor has a significant influence on the attitude toward the sponsorship. Moreover, Olson (2010) found that sponsorship attitude has a direct positive effect on sponsor equity. Ter Beek (2012) supports this by stating that the attitude toward the sponsorship is important in determining the effectiveness of a sponsorship. Moreover, it has been well established that sponsorship intervenes in the emotional link between customers and the specific event (Meenaghan, 2001). When customers perceive the association between the sponsor and the event as positive, one could assume a goodwill incentive from the customer and create affirmative feelings toward the brand (Mazodier & Merunka, 2010). As expected, Mazodier & Merunka (2010) found a positive relationship between attitude toward the sports sponsorship of Adidas sponsoring the Summer Olympics and brand affect. Also, Olson (2010) demonstrated a significant positive relationship between attitude toward the sponsorship and brand liking. Therefore, based on the research of Mazodier & Merunka (2010) and Olson (2010), the following hypothesis is proposed:

H1: Attitude toward the sponsorship relates positively to brand affect.

2.3 Attitude toward the Event

(16)

16

unconditioned stimulus. While the person is watching the event, the positive feeling they have with the event (the unconditioned response) will be projected to the conditioned stimulus which is the brand of the sponsor. A relatively large advantage of event sponsorship is that the viewer of the event chooses to be exposed to the sponsored event. Furthermore, Mazodier & Merunka (2012) established a positive link between attitude toward the event and brand affect of the sponsor. Hence, the more likely a customer is fond of a sponsored event, the more the individual will create a positive affect toward the brand that sponsors the event. Therefore, the following hypothesis is proposed:

H2: Attitude toward the sponsored event relates positively to brand affect.

2.4 Social Media Activity regarding the Event

2.4.1 The Rise of Social Media

As mentioned before, in this research the definition of social media of Xiang & Gretzel (2010) is adopted. The rise of social media had an enormous impact on several fields of expertise. For example, social media has changed certain aspects of journalism. Newman (2009) suggests that social media is prominently present and are key for effective storytelling in combination with mainstream media. Moreover, social media shifted the state of affairs in national politics (Shirky, 2011; Gunther & Mughan, 2000). According to Shirky (2011, p.2), “social media has become coordinating tools for nearly all of the world’s political movements, just as most of the world’s authoritarian governments are trying to limit access to it.” Recently, police forces are deploying social media in order to engage the community to solve a possible crime or gain important leads in crime cases. Crump (2011) noted that the police force also utilize social media to increase and maintain public trust, establish confidence in the police and improve the image of police officers. Furthermore, social media also shifted the business-to-consumer industry drastically. One of the main implications of social media for the business world is the inability of controlling the available information. As Kaplan & Haenlein (2010, p.60) suggested “firms have been increasingly relegated to the side lines as mere observers, having neither the knowledge nor the chance – or sometimes, even the right, - to alter publicly posted comments provided by their customers.”

2.4.2 Social Media in the Business-to-Consumer Market

(17)

17

television, magazines and tabloids (Mangold & Faulds, 2009) and take control in their own media consumption. Vollmer & Precourt (2008) noted that nowadays, customers are in control of their media intake and require on-demand and direct access to information at their own convenience. However, social media also offers a lot of opportunities for firms such as collaborative projects. Collaborative projects enable customers to comment, suggest and create new content for the company (Kaplan & Haenlein, 2011). The main idea is that collaborative projects lead to better outcomes compared to outcomes from one individual or the firm. Moreover, Qualman (2010) suggest firms to embrace social media in order to increase sales. Social media is also utilized by firms as a medium for educating (potential) customers about their products, services, causes or classes (Safko, 2010). There is a rich body of literature available on how firms could exploit social media. Rui et al. (2013) explored the link between the amounts of tweets on movie sales and found a significant positive effect. Furthermore, Dhar & Chang (2009) examined the content generated messages such as blogs and networking sites to predict music sales. It is also been established that brand communities on social media has a positive effect on brand loyalty (Laroche et al., 2013; Erdogmus & Cicek, 2012). Furthermore, it has been proven that social media predicts the tie strength; therefore, identify the trustworthy customers or the customers who are strangers of the company (Gilbert & Karahalios, 2009).

2.4.3 Social Media, Brand Affect and Brand Loyalty

(18)

18

that monitoring and measurement techniques can focus on the effectiveness and efficiency of different social media channels (Meenaghan et al., 2013).

There is quite some research establishing a link between social media activity and brand loyalty. Chaudhary, Asthana & Singhal (2015) claim that there is a link between the amount of likes of a Facebook page and loyalty; however, this is not empirically tested. In a recent paper, Liu (2016) showed that customers’ conversations on social media about brands has a significant impact on their valuation of brand features and ultimately on their purchase intent. Social media brand communities (e.g., fan pages on Facebook or owned brand pages on Facebook) offer firms the opportunity to enhance value, brand trust and brand loyalty (Muniz & O’Guinn, 2001; McAlexander, Schouten & Koening, 2002; Laroche, Habibi, Richard & Sankaranarayanan, 2012). Firms could also enhance its brand loyalty by offering different platforms and applications on social media in order to engage customers with the brand (Erdogmus & Cicek, 2012). Hence, companies should create more engaging and interesting contents to gain their customers’ interest. An interesting note is that individuals tend to prefer contents involving music, funny and extraordinary things (Erdogmus & Cicek, 2012). However, little research is present which confirms a relationship between social media activity of the sponsored event and the affect toward the sponsor. It could be assumed that membership and involvement in a brand community should also have an effect on the customer’s brand-behaviours. In the fashion retail industry, it has been proven that perceived social media activities has a mediated effect through brand equity, relationship equity and brand equity on brand loyalty (Kim & Ko, 2011). The authors regarded the customers’ entertainment, interaction and word of mouth as indicators for social media marketing activities. An interesting finding is that the social media marketing activities of the firm has a quite large significant impact on brand equity which in turn, has the biggest effect on the purchase intention (Kim & Ko, 2011). In the article of Kim & Ko (2011), brand equity is comprised of the customers’ perception of the brand’s uniqueness and distinctiveness. However, in most literature, customer’s brand equity consists of brand awareness and brand image (Keller, 1993). One could propose that if the link between the sponsored event and its sponsor is strong enough, social media activity regarding the event could also have effect on the brand affect of the telecommunication brand. So, one could assume that if customers engage in social media of both the cultural event and the sponsored related social media, it will have a positive effect on the brand equity of the firm that sponsors the event. Therefore, the following hypothesis is proposed:

(19)

19 2.5 Event Involvement

2.5.1 Customer Involvement

A rich body of literature is available focusing on customer involvement. Many different fields has researched the impact of involvement such as the effect of involvement on advertising (Krugman, 1965; Petty, Cacioppo & Schumann, 1983; Wu, 2001), involvement in health care (Lupton, Peckham & Taylor, 1998; Boote, Telford & Cooper, 2002) and involvement and its effect on psychological well-being (Kraut et al., 1998). The relationship between involvement and customers’ reactions to marketing communication has been extensively studied by many researchers present in the business and marketing fields (Park & Young, 1986). In the classic work of Howard & Sheth (1969), the authors suggested that customer involvement with brands or products has an effect on the degree of their information search, the scope of the customer’s consideration set and the origin of brand loyalty. Houston and Rotschild (1978) distinguished several customer involvement frameworks. One of those identified customer involvement frameworks is enduring involvement. According to Richins, Bloch & McQuarrie (1992), enduring involvement could be defined as “an individual difference variable representing the general, long-run concern with a product that a consumer brings to a situation.” Lardinoit and Derbaix (2001, p.170) supports this definition and adds that “enduring involvement corresponds to a kind of genuine enthusiasm, a strong and solid interest that comes from the relevance of an object or subject for the individual.” Involvement could also be conceptualized as genuine interest in the event or in the activity performed in the event, such as music (Grohs & Reisinger, 2014). Therefore, in the case of a national singing competition, enduring involvement could be observed as an individual participating in voting activities or downloading the music from the singing competition.

2.5.2 Elaboration Likelihood Model and Involvement

(20)

20

involved the customers are, the greater the commitment of the customer toward that specific brand. Moreover, Bruwer & Buller (2013) researched the link between product involvement and brand loyalty in the Japanese wine-industry. The authors found a positive association between wine product involvement and the loyalty to the wine brand. It is also been demonstrated that involvement could be analysed as a full mediator, partial mediator and moderator with regard to repurchase loyalty (Olson, 2006). However, the main finding by Olson (2006) entails that there is a direct effect and moderated effect of involvement and (repurchase) loyalty. One study supports this statement by showing that involvement has a direct influence on brand loyalty (Punniyamoorthy & Ray, 2007). However, in this research, involvement is viewed as event involvement and not as brand involvement or product involvement. Therefore, no direct effect of event involvement on brand loyalty is expected in this research.

2.5.3 Involvement and Sponsorships

As mentioned before, the theory of involvement creates quite simple predictions on the consequences of involvement on the behaviour of the consumer. Typically, if a customer is involved, one could expect that the customer engage in a number of activities (active search, in-depth choice process etc.); on the other hand, when customers are not involved, they should not be engaged in those activities (Krugman, 1965). Ample research has found a positive relationship between involvement and sponsorship effects (Madrigal, 2000). However, does the same hold for event involvement and the effect on affect of the brand that sponsors the event? Quite some sponsorship theory established that higher levels of enduring involvement improve awareness of event sponsors and knowledge of the event-sponsor link (Johar, Pham & Wakefield, 2006; Reisinger, 2013). As noted before, in this research enduring involvement is viewed as the individual expressing voting behaviour and for example downloaded the app of the event (Lardinoit & Derbaix, 2001). Furthermore, as proved by Grohs & Resinger (2014, p.1020), “higher activity involvement increases elaboration on the content of the event and processing of related activities, such as the sponsorships.” Gwinner (1997) empirically examines the relationship between involvement and brand attitude in a sponsorship setting. The author argues that especially low involvement products will be more effective in sponsorship marketing because of the peripheral nature of persuasion explained by Petty & Cacioppo (1986). Based on the available literature, it could be expected that respondents that are highly involved with the event could amplify the attitude toward the sponsorship and attitude toward the event. Therefore, the following hypothesis is proposed:

(21)

21

H5: Event involvement moderates the relationship between affect toward the event and brand affect.

2.6 Brand Loyalty and Brand Affect

2.6.1 Brand Loyalty

The concept of brand loyalty has a well-established and long presence in the marketing literature (Knox & Walker, 2010). According to Aaker (1991), brand loyalty of the customer base is a strong determinant of the core of the brand’s equity. Moreover, Aaker (1991) stated that brand loyalty could imply a reduction in marketing expenditures, increase in new customers and a better trade leverage. Oliver (1999, p. 34) defines brand loyalty as “a deeply held commitment to rebuy or re-patronize a preferred product/service consistently in the future, thereby causing repetitive same-brand or same same-brand-set purchasing, despite situational influences and marketing efforts having the potential to cause switching behaviour.” However, as stated in the definition of Oliver (1999), one could distinguish two kinds of brand loyalty: committed and purchase loyalty. Purchase loyalty, or behavioural loyalty, could be measured by analysing the repeat purchasing behaviour of the customer toward the preferred brand (Bandyopadhyay & Martell, 2007). Hence, behavioural loyalty is concerned only with observable activities (i.e., actual purchase behaviours, spending of customer) and therefore easy to collect (Russel-Bennett, McColl-Kennedy & Coote, 2007; Mellens, Dekimpe & Steenkamp, 1995). However, the behavioural brand loyalty lacks to incorporate the external influences and mental processes of the customer (Mellens et al., 1995). Therefore, in contrast to behavioural loyalty, committed or attitudinal brand loyalty includes “a degree of dispositional commitment in terms of some unique value associated with the brand” (Chaudhuri & Holbrook, 2001 p. 82). A rich body of literature argues that ultimately, attitudinal brand loyalty has a strong impact on behavioural brand loyalty (Bennett & Thiele, 2002; Back & Parks, 2003; Musa, 2005; Bandyopadhyay & Martell, 2007). Furthermore, compared to behavioural loyalty, attitudinal loyalty is less sensitive to short-run fluctuations (Mellens et al., 1995). However, attitudinal loyalty has also its shortcomings. Attitudinal loyalty is harder to collect and the validity of the representation of reality is not guaranteed (Mellens et al., 1995).

(22)

22

average. Therefore, Bennett & Thiele (2002) asserted that in those specific markets, it is of significant importance to monitor attitudinal loyalty and identify those customers who are dissatisfied with service encounters and may switch of provider. As a result, one could assume, attitudinal loyalty may have a significant impact on behavioural loyalty in the telecom industry. Moreover, according to Keller (1993), there are two different views on loyalty: the viewpoint of the firm and the viewpoint of the customer. The firm perceives customers’ loyalty in terms of market share or share of wallet, whereas customers perceive loyalty as personal associations with the brand (Chaudhuri & Holbrook, 2001). In this research, the perception of the customers’ view on loyalty is adopted.

2.6.2 Brand Affect

As mentioned before, it is proposed that sponsorship influences brand affect through different processes. However, the goal of this research is to clarify the impact of sponsorship on brand loyalty, therefore brand affect is related to brand loyalty (Chaudhuri & Holbrook, 2001). Brand affect is defined by Chaudhuri & Holbrook (2001, p. 82) as “a brand’s potential to elicit a positive emotional response in the average consumer as a result of its use.” Olson (2010) measured brand equity by measuring the customer’s feelings and likeability toward the brand, which both link to brand affect. It has also been demonstrated by various scholars that brand liking or brand attitude has a positive influence on brand affect (Sung & Kim, 2010). Hence, one may assume that a positive attitude toward the brand may result in a higher brand affect (Mazodier & Merunka, 2012). Moreover, there exists a rich body of literature that proves the positive relationship between brand affect and brand loyalty (Chaudhuri & Holbrook, 2001; Matzler, Grabner-Kräuter & Bidmon, 2008; Singh, Iglesias & Batista-Foguet, 2012). Therefore, brand affect is included in this research to predict brand loyalty in a sponsorship context. The following hypothesis is proposed:

H7: Brand affect is positively related to brand loyalty.

2.7 Control Variables

In this research age, gender and the number of season of the event are included as control variables. Research suggests that these factors may have an influence on the level of brand affect of customers, therefore it is of upmost importance to study their effects and control for them in the analyses performed in this research.

2.7.1 Age

(23)

23

people. Therefore, in this research age is added as a control variable in order to regulate for any differences in perception of attitude toward sponsorship and attitude toward event on brand affect.

2.7.2 Gender

Sufficient research proves that men are to a higher extent involved with telecom (Einav, Levin, Popov & Sundaresan, 2014). Therefore, male customers could form a more positive attitude toward the telecommunication brand compared to women.

2.7.3 Number of Events

The number of season of the event, which is ranging from season one to season 5, is also included as a control variable in this study. There may be differences in perception of the level in brand affect. For example, respondents of the second season of the event may be less aware of the sponsorship between the event and the telecommunication brand than the respondents of the fourth season. Therefore, the number of season is included in this research.

A conceptual model of all presented hypotheses is provided in figure 1.

3.

Attitude toward the sponsorship Attitude toward the event Social media activity regarding the event

Brand affect Brand loyalty

Event involvement

H2

H3

H7

H4

H1

H5

Control variables: o Age o Gender o Number of events

H6

(24)

24 3. RESEARCH DESIGN

This section elaborates on the design of the research. First, the research context is explained. Second, this chapter will provide an overview of the method and procedure of this research. Final, the plan of analysis of this research is explained.

3.1 Research Context

In this study, the effects of exposure to an event sponsorship on brand loyalty are researched. The Dutch talent show has been broadcasted by a large Dutch broadcast station since 2010. The yearly broadcasted talent show takes off in September and lasts until December or January with on average 20 episodes that takes about 130 minutes including commercial breaks. Every Friday, the talent show has 2.6 million audience ratings on average in the Netherlands. The talent show differentiates from other Dutch television talent shows by integrating the off- and online world in a new and interactive way. The real-time footages and information of different social media platforms will be displayed during the event to increase the experience of the viewer. The large Dutch telecom company sponsors the talent show from the beginning and facilitates the aim of the talent show to improve the viewers’ experience. The telecom company allows viewers to watch the talent show on a second screen such as a smartphone or a laptop to provide extra information and encourage the viewer to be interactive with the talent show. Adjacent to it, participants of the talent show can communicate via a mobile app in order to update all the viewers of their experiences. The app is also a platform to download their music, vote for the television show and receive news about the event. By incorporating social media into the talent show, it allows the viewers to be up-to-date regarding the developments of the talent show during and outside the broadcast of the event.

3.2 Method

(25)

25

which reminded the potential participant to fill in this survey. If the respondent declared that he or she was not familiar with the event, the questions about the event and the sponsorship were not displayed to the respondent. In this study, all constructs applied are established via an independent research firm who carried out the research for the telecommunication brand. All scale items appear in Appendix B.

3.3 Measurements

Several questions were asked in order to create the variables analysed in this study. The exact questions asked to the telco brand customers can be found in Appendix B.

3.3.1 Brand Loyalty

In order to measure this research’s dependent variable “brand loyalty”, purchase intention is utilized. As mentioned before, brand loyalty is described as the final aspect of consumer brand resonance symbolizing the customer’s decisive relationship and the extent of personal identification with the brand (Keller, 2001). However, in this study, behavioural loyalty is not observable and therefore attitudinal loyalty is observed. Attitudinal brand loyalty includes “a degree of dispositional commitment in terms of some unique value associated with the brand” (Chaudhuri & Holbrook, 2001 p. 82). In this study, the measurement of purchase intention is adopted to measure attitudinal loyalty.

3.3.2 Brand Affect

In this research, brand affect is measured by asking the respondents to answer a question about their feeling toward the telecommunication brand. The respondent was presented with four answer categories ranging from ‘a bad feeling’ to ‘a positive feeling’.

3.3.3 Attitude toward Sponsorship

The independent variable attitude toward sponsorship was measured by providing the customer with 5 different statements about the sponsorship between the telecommunication brand and the event. The statements contained topics such as discussing the sponsorship with friends, if the sponsorship made the telecommunication brand more appealing, feelings about the sponsorship, if the sponsorship made the telecommunication brand different compared to other providers and if the sponsorship contained new information about the telecommunication brand. The respondents were asked to what extent they agreed with the statement on a five-point Likert scale ranging from disagree to agree.

3.3.4 Attitude toward Event

(26)

26

that point in time. However, if the respondent graded the event with a ‘10’, the customer indicated that he or she does like the event and thereby formed a positive attitude toward the event.

3.3.5 Event Involvement

In this research, a dummy variable is created which signifies if the customer is involved or is not with the event. The moderator variable “event involvement” is measured by asking if the respondents have voted via the various channels on their favourite talent (by sending a text and/or calling by phone), downloaded the custom made app of the event, downloaded a song of the participants and/or visited the website of the event. The customer is involved when one or more actions that are previously described are performed by a respondent and indicated by a 1 in this research. A 0 value for the variable event involvement implies that the customer is not involved with the event. Moreover, while observing the variable event involvement, three outliers were detected. These respondents were involved in all five activities described earlier. These outliers were removed from the analysis since they could potentially bias the outcome.

3.3.6 Social Media Activity regarding the Event

The independent variable, social media activity regarding the event, of the respondent is measured by asking if the respondent posted something on Twitter related to the event, followed the event on Facebook and followed a reporter on Facebook, whose task was to inform the viewer about the show apart from the broadcasts. The variable is dummy coded, 0 meaning no social media activity has been performed and 1 meaning that the respondent was active on social media regarding the event. However, only 19 respondents declared that they were active on social media regarding the event.

3.3.7 Control Variables

The variable gender is dummy coded, a 0-value indicating that the respondent is a female while 1 indicates that the respondent is a male. Age is denoted as the age in numbers and is therefore considered as a continuous variable. Lastly, number of event is a categorical variable ranging from 1 to 5. The number indicates in which season the respondent answered the questionnaire.

3.4 Method of Analysis

3.4.1 Factor Analysis

(27)

27

Table 2 Statements Factor Analysis

procedures primarily used for data reduction and summarization.” The respondents’ attitude toward the sponsorship between the telecommunication brand and event were measured by five different questions. The five questions (see table 2) were asked with a five point Likert scale, ‘1’ being ‘Strongly Disagree and ‘5’ being ‘Strongly Agree’.

According to Malhotra (2010), there are several assumptions that need to be satisfied in order to perform a factor analysis. The first assumption is rejecting the null hypothesis that the population correlation matrix is an identity matrix evaluated by the Barlett’s test of Sphericity. The second assumption, the Kaiser-Meyer-Olklin (KMO), which measures the appropriateness of the factor analysis, must be significantly larger than the desired value of ,50 (Malhotra, 2010). The last assumption that needs to be satisfied is that all communalities should be above the extraction value of ,40 (Malhotra, 2010).

After creating the variable via factor analysis, validation is required to assume that the measurements truly represent the construct utilized in this research. In order to research the reliability of the construct, the Chronbach’s Alpha was deployed. The Chronbach’s Alpha is “an index of reliability associated with the variation accounted for by the true score of the underlying construct” (Santos, 1999 p.2). According to Nunnaly (1978), a score of ,7 is an acceptable reliability outcome. Therefore, in this study, the cut-off point ,7 of Chronbach’s Alpha is seen as the minimum score.

3.4.2 OLS Regression

In this paper, the analysis of the model is performed by a linear regression model in order to test the hypotheses estimated in SPSS edition 23. Furthermore, the moderating role of involvement with the sponsored event will also be analysed via linear regression. While performing the linear regression to test if a moderator effect is present, the variables involvement, attitude toward event and attitude toward sponsorship are mean centred for interpretation purposes. Moreover, the presence of non-normality is expected in this research. Therefore, the bootstrapping method is applied in this research.

3.4.2.1 Model 1

In order to examine level of brand affect to predict the brand loyalty of the customers, ‘Model 1’ uses brand affect as dependent variable and attitude toward sponsorship, attitude toward the event

The sponsoring of the event by the telecommunication brand is something I would talk about with friends The sponsoring of the event by the telecommunication brand let me believe that the brand is different from other brands

The sponsoring of the event by the telecommunication brand made the brand more appealing for me

The sponsoring of the event by the telecommunication brand was nice to watch

(28)

28

and social media activity regarding the event as independent variables. Moreover, the control variables age, gender and number of event were added in the model.

𝐵𝑟𝑎𝑛𝑑 𝑎𝑓𝑓𝑒𝑐𝑡 = 𝛽0 + 𝛽1𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑇𝑜𝑤𝑎𝑟𝑑𝑆𝑝𝑜𝑛𝑠𝑜𝑟𝑠ℎ𝑖𝑝 + 𝛽2𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑇𝑜𝑤𝑎𝑟𝑑𝐸𝑣𝑒𝑛𝑡 + 𝛽3𝑆𝑜𝑐𝑖𝑎𝑙𝑀𝑒𝑑𝑖𝑎𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦𝑅𝑒𝑔𝑎𝑟𝑑𝑖𝑛𝑔𝑇ℎ𝑒𝐸𝑣𝑒𝑛𝑡 + 𝛽4𝐴𝑔𝑒 + 𝛽5𝐺𝑒𝑛𝑑𝑒𝑟 + 𝛽6𝑁𝑢𝑚𝑏𝑒𝑟𝑂𝑓𝐸𝑣𝑒𝑛𝑡 + 𝜀

3.4.2.2 Model 2

In order to examine the level of brand affect and assess the moderating role of event involvement, ‘Model 2’ is similar to model 1 but adds the moderated effect of involvement on attitude toward the sponsorship, the moderating effect involvement on attitude toward the event and involvement as independent variables (see Appendix F for model formulation).

However, the cross-predictors Event Involvement x Attitude toward sponsorship and Event Involvement x Attitude toward event is likely to be correlated with the other independent variables attitude toward event and attitude toward sponsorship. This could be interpreted as a form of multicollinearity which makes it difficult to distinguish the separate effects of the moderator effect and the effect of the independent variables (Echambadi & Hess, 2007). To counteract multicollinearity, mean-centring the independent variables and moderator is recommended (Aiken & West, 1991; Jaccard, Turrisi & Wan, 1990). However, Echambadi & Hess (2007) withstand the recommendations of Aiken & West (1991) to use mean-centring to counteract multicollinearity. Echambadi & Hess (2007) suggest using mean-centring only for interpreting purposes. Therefore, to test the moderator effect and correctly interpret a moderation effect, the equation of model will be of the following form:

𝐵𝑟𝑎𝑛𝑑 𝑎𝑓𝑓𝑒𝑐𝑡 = 𝛽17 + 𝛽18(𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑇𝑜𝑤𝑎𝑟𝑑𝑆𝑝𝑜𝑛𝑠𝑜𝑟𝑠ℎ𝑖𝑝 − 𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑇𝑜𝑤𝑎𝑟𝑑𝑆𝑝𝑜𝑛𝑠𝑜𝑟𝑠ℎ𝑖𝑝̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅) + 𝛽19(𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑇𝑜𝑤𝑎𝑟𝑑𝐸𝑣𝑒𝑛𝑡 − 𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑇𝑜𝑤𝑎𝑟𝑑𝐸𝑣𝑒𝑛𝑡̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅) + 𝛽20𝑆𝑜𝑐𝑖𝑎𝑙𝑀𝑒𝑑𝑖𝑎𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦𝐼𝑛𝑣𝑜𝑙𝑣𝑖𝑛𝑔𝑇ℎ𝑒𝐸𝑣𝑒𝑛𝑡 + 𝛽21(𝐼𝑛𝑣𝑜𝑙𝑣𝑒𝑚𝑒𝑛𝑡 − 𝐼𝑛𝑣𝑜𝑙𝑣𝑒𝑚𝑒𝑛𝑡̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅) + 𝛽22(𝐼𝑛𝑣𝑜𝑙𝑣𝑒𝑚𝑒𝑛𝑡 − 𝐼𝑛𝑣𝑜𝑙𝑣𝑒𝑚𝑒𝑛𝑡̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅) ∗ (𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑇𝑜𝑤𝑎𝑟𝑑𝑆𝑝𝑜𝑛𝑠𝑜𝑟𝑠ℎ𝑖𝑝 − 𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑇𝑜𝑤𝑎𝑟𝑑𝑆𝑝𝑜𝑛𝑠𝑜𝑟𝑠ℎ𝑖𝑝̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅) + 𝛽23(𝐼𝑛𝑣𝑜𝑙𝑣𝑒𝑚𝑒𝑛𝑡 − 𝐼𝑛𝑣𝑜𝑙𝑣𝑒𝑚𝑒𝑛𝑡̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅) ∗ (𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑇𝑜𝑤𝑎𝑟𝑑𝐸𝑣𝑒𝑛𝑡 − 𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑇𝑜𝑤𝑎𝑟𝑑𝐸𝑣𝑒𝑛𝑡̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅) + 𝛽24𝐴𝑔𝑒 + 𝛽25𝐺𝑒𝑛𝑑𝑒𝑟 + 𝛽26𝑁𝑢𝑚𝑏𝑒𝑟𝑂𝑓𝐸𝑣𝑒𝑛𝑡 + 𝜀 3.4.2.3 Model 3

(29)

29

Figure 2 Mediation analysis

𝐵𝑟𝑎𝑛𝑑 𝐿𝑜𝑦𝑎𝑙𝑡𝑦 = 𝛽27 + 𝛽28𝐵𝑟𝑎𝑛𝑑 𝐴𝑓𝑓𝑒𝑐𝑡 + 𝜀 3.4.3 Mediation Model

Since the effect of the independent variables on brand affect and the moderating role of involvement on brand affect have been established, it is interesting to investigate if a direct relationship exists between the independent variables social media activity, attitude toward the sponsorship, attitude toward the event and involvement on brand loyalty. In the article of Baron & Kenny (1986), the appropriate analytic procedure of testing mediation is provided. A mediator is present when it accounts for the relation between the predictor variables and the dependent variable. According to Baron & Kenny (1986, p. 1174), “the statistical analysis must measure and test the differential effect of the independent variable on the dependent variable as a function of the moderator.” The authors state several assumptions that need to be met to analyse if a (partial) mediation is present. First, the variations of the predictor variables significantly

influences the mediator variable (path a). Secondly, the mediator variable must significantly influence the dependent variable (path b). Third, in order to test if the mediator partially mediates or fully mediates, the direct relationship between the independent variable and the dependent variable needs to be tested (c and c’). In figure 2, a graphical representation is displayed of the different paths in

order to test a (partial) mediation effect. Partial mediation occurs when M mediates the effect of X on Y (path c), but the relationship between X and Y controlled for M is also significant (path c’). Full mediation occurs in which the independent variable X no longer affects Y after M has been controlled for (path c’). All (moderated) mediation models are provided in Appendix F.

3.4.4 Model Fit

According to Leeflang, Wieringa, Bijmolt & Pauwels (2015, p.102), “an important measure for assessing the quality of a model is the extent to which fluctuations in the criterion variable are explained by the model.” A possible criterion measure is the coefficient of determination also known as R². However, the use of multiple independent variables may cause an artificially high R². Therefore, in most literature, the adjusted R² (𝑅𝑎2) is used because it punishes models with many predictor variables. The adjusted 𝑅𝑎2 is computed as follows:

𝑅𝑎2= 1 −

(30)

30

In literature, no clear cut-off value of the coefficient of determination is established. The value of 𝑅𝑎2 ranges from zero to one. A value close to one is desired (Malhotra, 2010).

4. RESULTS

4.1 Characteristics

The total sample of this research contained 2415 respondents. However, due to a substantial amount of missing values and inconsistency in the dataset only 279 respondents are analysed in this research. According to Hair, Black, Babin, Anderson & Tatham (2006), one would need a minimum of 200 respondents in order to create a sufficient representation of the population. In appendix A, a summary of the total respondents for each questionnaire in each year is provided. In table 3 the respondents’ characteristics of the total sample is summarized.

Table 3 Sample Characteristics

Answer category Frequency Percentage

Characteristic Age

15 – 24 262 10,8% 25 - 34 491 20,3% 35 - 44 593 24,6% 45 - 54 699 28,9% 55+ 370 15,3%

Characteristic Gender

Female 1292 53,5% Male 1123 46,5%

Characteristic Education

No education 20 ,8% Low education 368 15,2% Middle education 992 41,1% High education 944 39,1%

Characteristic Region

Northern region 268 11,1% West region 1011 41,9% Eastern region 572 23,7% Southern region 563 23,3%

Don’t want to say 1 ,0%

(31)

31

male (46,5%). According to CBS (2013), there are slightly more females present in the Netherlands than males. Moreover, the respondents in this sample are well educated. 39,1% of the respondents received high education (HBO/WO). 41,1% had middle education (MBO/VMBO) and counts for the majority of this sample. Lastly, most respondents (n = 1011, 41,9%) lives in the West Region. The densest area to live in is in the western region and the least dense area to live in is in the northern region of the Netherlands (CBS, 2013). To conclude, the sample of this research is viewed as a proper representation of the Dutch population. In table 4, the descriptives (N, minimum value, maximum value, mean and standard deviation) of the analysed variables of this research are presented.

N Min Max Mean Std. deviation (SD)

Brand Affect 529 1 4 2,57 ,811

Brand Loyalty 991 1 5 2,31 1,216

Social Media Activity 751 0 1 ,04 ,190

Attitude toward Event 972 2 10 7,73 1,285

Attitude toward Sponsorship 641 -2,18117 2,88545 -,0274483 ,984102

Involvement 607 0 1 ,30 ,461

Number of Event 2433 1 5 2,92 1,322

Age 2434 15 59 40,99 11,822

Gender 2434 0 1 ,53 ,499

4.2 Correlations

According to Malhotra (2010), it is insightful to summarize the strength of association between two predictors. The correlation (r) is the most widely accepted statistic to summarize the strength between two variables. In this research, the Pearson correlation test was performed in order to find correlations. The size of the correlations varies from particularly weak (r = ,003) for gender and number of events to quite high (r = ,594, p < .01) for attitude toward sponsorship and brand affect. If extreme high correlations are present this could cause multicollinearity. However, as will be established in chapter 4.2, no multicollinearity is detected in this research. In Appendix D the correlation coefficients matrix is provided of the variables used in this research.

When analysing the correlation matrix table, it is noticeable that brand loyalty correlates with brand affect (r = .328; p < .01). As expected, this implies that the customers with a positive attitude toward the brand, displays a higher brand loyalty. Furthermore, there exists a positive strong correlation between brand affect and attitude toward sponsorship (r = .594; p < .01), which infers customers with a more positive attitude toward the sponsorship displayed a higher level of brand affect for the telecommunication brand. Additionally, brand affect also correlates positively with attitude toward the event (r = .273 ; p < .01). So, one could state that the more positive attitude toward the event, the more affect the customers display for the telecommunication brand. Moreover, a positive small correlation was found between level of involvement and brand affect (r = ,159; p <.05).

(32)

32

Next, the correlations of brand loyalty were inspected. A medium correlation is present between attitude toward the sponsorship and brand loyalty (r = ,428; p < .01), which implies that customers who formed a positive attitude toward the sponsorship, are more loyal to the brand. As for the control variable age, there is a positive correlation between age and brand loyalty (r = .072; p < .05) which suggest that older customers are more brand loyal. Subsequent, the correlations of social media activity are also analysed. Social media activity correlates negatively with attitude toward sponsorship (r = -,126; p < .05). Moreover, the correlations of attitude toward sponsorship are also examined. As mentioned before, attitude toward sponsorship correlates positively with brand affect and brand loyalty, but negatively with social media activity. Furthermore, attitude toward sponsorship has a small positive correlation with involvement (r = ,143; p < .05), which implies that the more involved the customers are with the event, the more positive their attitude toward the brand.

4.3 Factor Analysis

In order to prevent multicollinearity in this research, a factor analysis for the five statements presented in table 2 is performed. The results of the factor analysis are provided in Appendix C. The Bartlett’s Test of Sphericity provides a Chi-square statistic of 344,689 with 10 degrees of freedom, which is significant at the ,00 level. Therefore, the null hypothesis entailing that the population correlation matrix is an identity matrix is rejected. Second, the Kaiser-Meyer-Olklin (KMO) is ,879, which is significantly larger than the desired value of ,50. The last assumption is that communalities should be above the extraction value of ,40. In this research, all communalities are above a value of ,83. Based on the factor loadings, the factor attitude toward the sponsorship is created.

The attitude toward the sponsorship measurement scales used in the analysis requires to be validated. The construct of the variable attitude toward sponsorship satisfied the proposed threshold with a Chronbach’s alpha value of ,911. A summary of the construct validity of this research is summarized in table 5.

Table 5 Overview of Attitude toward Sponsorship Variable

Variable Nr of items Reliability Mean Standard deviation

Attitude toward sponsorship 5 Chronbach’s Alpha ,911

(33)

33 4.4 Preliminary checks

Only three outliers were detected regarding the variable involvement. These three outliers were removed from the analysis. According to Leeflang et al. (2015) several assumptions need to be satisfied when the parameters are estimated according to the OLS method.

4.4.1 Non-Zero Expectations

The first assumption of nonzero expectation should not be violated. Nonzero expectation could be caused by incorrect functional form(s), omitted variable(s) and varying parameter(s) and has the consequence of a biased parameter estimate (Leeflang et al, 2015). In order to detect a possible zero-expectation, the unstandardized residuals were plotted against each predictor variable. In this research, the non-zero expectation is not violated.

4.4.2 Homoscedastic Error Term

The second assumption is that the error term should be homoscedastic meaning that it has the same variance in all cases. To detect for heteroscedasticity in the proposed model, the

Levene’s test was conducted. According to Wieringa (lecture 6, 2014), the Levene’s test statistic analyses if a group has different variance compared to another group. If the Levene’s test is significant, it could be concluded that the variances are significantly different. However, since the Levene’s test is not significant (table 6), one could assume that the error term does have the same variance in all cases. Hence, the second assumption is not violated in this research.

4.4.3 Autocorrelation

Autocorrelation, also known as correlated disturbances, is present if the residuals exhibit some systematic pattern over time (Leeflang et al., 2015). If autocorrelation is present it could cause biased parameter estimates. In order to test if autocorrelation is present in this research, the Durbin-Watson test statistic is computed. Durbin-Durbin-Watson is a measurement statistic of autocorrelation or serial correlation in the residuals of regression analysis (Michalakis, Varoutas & Sphicopoulos, 2008). The value of the Durbin-Watson decreases as the autocorrelation increases. The larger the autocorrelation is, the less reliable the results of the regression analysis become (Michalakis et al., 2008). According to Mendenhall, Sincich & Boudreau (1996), a value close to 2 is desired. In this research, the Durbin-Watson test statistic has a value of 1.938, which indicates that there is no relation between the residuals and no pattern could be detected implying no autocorrelation is present in this research.

Levene’s test Df1 Df2 Sig.

,109 1 279 ,905

Does Event Sponsoring Pay Off?