The effects of sponsoring and brand image on the willingness to pay : an empirical investigation of a music festival

(1)

The effects of sponsoring and brand image on the willingness

to pay: an empirical investigation of a music festival

Joren Harm Heuvels (10870946)

29th of June 2015

Master of Science in Business Administration track ‘Entrepreneurship and Management in Creative Industries’

Amsterdam Business School, University of Amsterdam, Faculty of Economics and Business

Supervisor: dhr. prof. dr. N.M. Wijnberg

(2)

Abstract:

A huge amount of literature studied the effects of sponsoring and brand image transfer on a sponsoring brand, however literature on the effects for the sponsee is sparse. This research investigates the effects of sponsoring and the brand image fit on the willingness to pay for the sponsee by using Veylinx, an online auction

platform. In this study the brand image fit is determined by genre overlap between the sponsor and the sponsee. This studied found significant results that a sponsor with a low brand image fit with the sponsee will lead to a lower willingness to pay for the sponsee compared to no sponsor. In addition, a sponsor with a high brand image fit with the sponsee will lead to a significant higher willingness to pay for the sponsee compared to a sponsor with a low brand image fit with the sponsee. Furthermore, previous attendance is a moderator on the willingness to pay for the sponsor with a low brand image fit with the sponsee. At last, other effects are described and explained. Important theoretical and managerial implications are offered, as well as directions for further research, since the findings in this study indicate a need for future research on the effects of sponsoring and brand image fit for the sponsee.

Statement of Originality

This document is written by Joren Heuvels who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

The effects of sponsoring and brand image on the willingness

to pay: an empirical investigation of a music festival

1. Introduction

The music festival market is a dynamic market. As a result, the music festival industry is placed on a commercial footing, hence sponsorships are becoming more

important. In line with the sponsorship of events and celebrity endorsement, music festival sponsoring has become an increasingly popular marketing communication tool (Cornwell et al., 2001). As a result, sponsorship expenditures increased from $44 billion in 2009 to $57,5 billion in 2015 (Statista.com). In addition, in 2012 there was a total of 38 million euros of art- and culture festivals sponsoring in the Netherlands (Marketingtribune.nl).

The two main revenue streams for the music festival industry are ticket sales and the sponsorship fees. As a result, it is interesting to study the effects of sponsoring on the willingness to pay for a festival ticket. Sponsorships could be beneficial to the success of a festival if the visitors feel familiar and confident with the brand

(Anderton, 2009). Hence, sponsoring is an important tool to position a festival, which is important for a festival marketing strategy.

Many researchers studied the effects of sponsorships and brand image transfer for the sponsors brand. Especially for celebrity endorsement and sport events there is an existing body of literature on the effects of sponsorship and brand image for the corporation. In addition, previous research studied the effects of brand image transfer (Cornwell and Maignan 1998; Crowley 1991; Fleck and Korchia, 2009; Gwinner 1997; Gwinner and Eaton, 1999; Keller, 1993; Meenaghan 1991;

Meerabeau et al. 1991; Till, 1998). However, little research exists on the effects of a sponsorship for the sponsee. As mentioned earlier, there has been research of the sponsorship effects of the event on the image of the sponsor; hence the literature

(6)

mainly studied the image transfer as a one-way image transfer from event to sponsorship. Most literature acknowledge that the brand image transfer works in two directions (Gwinner, 1999; Keller, 1993), however little research focused on the effects of sponsorships on the image of the event. Thus, this thesis will focus on the effects of sponsorships on music festivals and the willingness to pay for the festival. Hence the research question:

What are the effects of a sponsorship and the brand image fit with the sponsee on the willingness to pay for the sponsee?

Three hypotheses, followed by an experiment led to an answer of this research question. An experiment was conducted with the use of Veylinx, an online auction platform. A single ticket for the ‘A Campingflight to Lowlands paradise 2015’ festival was auctioned. Respondents were exposed to one of the four manipulated

treatments that showed different sponsors. Next to the bid amount, which

measured the willingness to pay, the respondents were asked to respond to several questions and statements. The combination of the bid and those answers led to the variables for testing the hypotheses of this research.

In the first place this study contributes on the general discussion on theories of sponsorship and brand image transfer and brand image fit. Especially on the side of the sponsee it gives new perspectives and opens new roads for future research. It also contributes on theories of celebrity endorsement. In addition, the research has several managerial contributions and implications to offer. Especially for festival organizers this research is quite relevant to optimize their brand image, sponsorship deals and ticket sales. At last, since the empirical setting of this research is the music festival industry, this research contributes for other related studies in the music festival industry.

(7)

The outline of this study is as follows: The paper starts with a review on the existing literature on sponsorship effects. Based on the gaps of the literature a total of three hypotheses are formulated. Then, the empirical setting is described together with the methodology and brand image operationalization. In the sections after, the results are presented and analyzed in more detail in the discussion section, which also give managerial implications, theoretical implications, limitations of this research and suggestions for future research. The paper is finished with a conclusion.

(8)

2. Literature review

In this section, the existing literature is reviewed. First, the literature on sponsorship is introduced. Second, the literature on the willingness to pay is presented. Third, the literature on brand image, brand image transfer and brand image fit is discussed. In addition, the three hypotheses are presented.

2.1 Sponsorship

Despite the massive growth in sponsorship activities as part of the marketing communication and promotional mix (Glaser and Lum, 2004), academic research on the effects of sponsorships for the sponsee is still sparse. However literature on the effects of sponsorship for the sponsor offers valuable insights into the sponsorship effects for a sponsee.

Companies engage in sponsorship activities for a variety of reasons and objectives. Non-commercial reasons for corporations to engage in sponsorship activities are to establish, strengthen or change brand image, discover new markets, positioning of the brand, generate goodwill, enhance community relations. In addition, commercial reasons are to increase sales/revenue and as a result gain more profit; hence

commercial sponsors are sponsors with a financial objective. However, sponsorship priorities and effects differ between corporations and countries (Cornwell and Maignan 1998; Crowley 1991; Dean, 2002; Gwinner 1997; Gwinner and Eaton, 1999; Keller, 1993; Khan and Stanton, 2010; Meenaghan 1991; Meerabeau et al. 1991; Quester and Thompson, 2001; Rowley and Williams, 2008). Next to the possible advantages of a sponsorship, a sponsorship could also damage the brand (Henseler et al., 2009; Javalgi et al., 1994; Till, 1998). This study contributes to the existing literature by looking at the effects of a sponsorship on the sponsee side. More specifically, to the willingness to pay for a festival ticket.

(9)

Definition - Many definitions of sponsorships exists. Abratt et al. (1987) defined sponsorship as ‘an agreement in terms of which a sponsor provides some aid to a beneficiary, which may be an association, a team, or an individual, to enable the latter to pursue some activity and thereby derive the benefits contemplated in terms of its promotion strategy. The aid provided may be financial or a service of expertise, whereas the benefits may be media exposure, creating name awareness of the product/company, and overall publicity’. In addition, Gardner and Shuman (1988) defined sponsorship as ‘investments in causes or events to support corporate objectives, such as an enhancement of corporate image or an increase in brand awareness’. Lee, Sandler and Shani (1997) define event sponsorship as the provision of resources by an organization directly to an event, cause or activity in exchange for a direct association to the event, cause or activity. The providing organization can then engage in sponsorship-linked marketing to achieve either their corporate, marketing or media objectives. However, all those definitions only look at the sponsor side of sponsorship. At last, Cornwell and Maignan (1998) stated that ‘sponsorship involved two activities: (1) an exchange between a sponsor and a sponsee whereby the sponsee recevies a fee and the sponsor obtains the right to associate itself with the activity sponsored; and (2) the marketing of the association by the sponsor. Both activities are necessary if the sponsorship is to be a meaningful investment’. Hence, a sponsorship is based on both (sponsee and sponsor)

objectives (Madrigal, 2001). However, Johar, Pham, and Wakefield (2006) found that the audience often does not correctly discriminate actual offcial sponsors from other brands, hence event sponsors often do not receive the right proper credit for their sponsorship efforts.

2.2 Sponsorship effect on the willingness to pay of the product of the sponsee

Previous research has studied the effects of sponsorships on the willingness to pay and the purchase intention. Spears and Singh (2004) define purchase intention as ‘an individual’s conscious plan to make an effort to purchase a brand’. Simonin and Ruth (2003) found that attitudes towards brands participating in a product bundle

(10)

influenced consumers willingness to pay for bundled products, since it affects the cognitive and affective response of a consumer. Hence, the consumer’s cognitive and affective response to a brand were significantly linked to the willingness to pay (Koo et al., 2006). Lu, Chang, Chang (2014) found that high awareness products are positively valued by consumers, which improves the willingness to pay.

Commercial sponsors might have a negative influence on the effect. For example, the Olympic Games in Atlanta were sponsored by Coca-Cola, however this

sponsorship negatively affected the credibility and attractiveness of the Olympic Games (Pope et al., 1997). In addition, if the sponsorship is announced by the sponsor, then consumers have a less positive reaction to the sponsorship

relationship then when the sponsorship is announced by the sponsee (Becker-Olsen and Simmons, 2002). Concluded, Ruth and Simonin (2003) conclude that consumer perception of sponsor motives will have an effect on evaluations by consumers. Sponsorship motives will not only influence attitudes toward the event, but also the likelihood of purchasing tickets for the event and attending the event. Hence, the prior attitudes toward sponsoring brands will influence consumer evaluations of the sponsored event and affect the willingness to pay. As a result, the prior attitudes toward each sponsor are positively related to attitudes toward the event.

Furthermore, Forehand (2000) reports lower brand attitudes and a lower willingness to pay for sponsorship that are commercial, compared to goodwill. At last, the concept of purchase intention is used in some literature, however this study will use the concept of willingness to pay, since it studied the actual willingness to pay and not just the intention of purchasing.

2.3 Brand associations and brand image transfer

Brand image should first be explained in order to explain the concept of brand image transfer. Brand image is ‘the set of associations of the brand as reflected by the brand held in consumer memory’ (Keller, 1993). Positive brand image is achieved if consumers react more favorable to elements of the brand than to elements of other

(11)

brands. A brand is characterized by the brand associations in a consumer’s mind, hence the brand image and brand meaning is created. Brand judgements and brand feelings are a result of this brand image (Keller, 2001). Hence, a brand image is based upon brand associations and linkages a consumer holds in memory regarding the brand (McCracken, 1989). Grohs et al. (2004) defines brand image (in accordance with Keller, 1993) as the perceptions about a brand as reflected by the brand associations held in consumer memory.

Companies try to link their brand to an entity to increase their brand awareness and as a result to increase their brand equity and brand image (Keller, 1993). In addition, brands could build brand equity by leveraging their secondary associations (Keller, 2005; Till et al., 2011). Linking a brand to other entities creates associations with those entities, hence the consumer can identify and associate the brand with those other entities. Those entities could be people (endorsement), other brands (alliance or sponsorship) or places (location). The consumer knowledge of the entity, the meaningfulness of consumers knowledge on the entity and the transferability of consumers’ knowledge on the entity are important dimensions of brand association and are important aspects of creating brand meaning and brand image. However, leveraging secondary associations might be risky, since there is no full control on the brand image, which will be mentioned in the celebrity endorsement section.

Marketeers often attempt to increase brand equity by borrowing equity from other entities (Keller, 2001).

Associations are developed through brand and product category experiences, product attributes, price information, positioning in promotional communications, packaging, user imagery and user occasion (McCracken, 1989). Chavanat et al. (2009) conclude that the image of a sponsee and the image of a sponsor do interact within a sponsorship arrangement. Hence, a company creates a link with an entity with a sponsorship, hoping to influence the audience behavior by image transfer, since brand image associations are influenced by sponsorships. Hence, a brand becomes

(12)

linked with an entity through sponsorship activities, because the associations held in consumers’ memory regarding the entity become linked in the memory with the endorsing or sponsoring brand (Bashiri et al., 2010; Chavanat et al., 2009; D’Astous and Bitz, 1995; Gwinner, 1997; McCracken, 1989). Brand associations are developed from product use, informational sources and secondary associations with other entities (Gwinner, 1997; Keller, 2003). Those brand associations are defined as the degree of favourability, strength and uniqueness (Keller, 1993). Secondary

associations lead to brand image transfer. Concluded, the associations that are linked to one object in the mind are transferred to another object within the mind (Keller, 1993). Hence, the image is transferred from one entity to another entity.

Henseler et al. (2009) studied the transfer of associations and brand image from a sponsor to a sponsee. The image of a sponsor has an impact on the image of the sponsee. Sponsorships can create or enhance but also destroy or damage value for the sponsee, next to the sponsorship fee. As a result of image transfer, the image of a sponsor has an impact on the image of the sponsee because of their association, hence the brand image transfer mechanism works in both directions (Henseler et al., 2009). In addition, brand image transfer from the sponsor to the sponsee is likely to occur, since it is possible to transfer associations in both directions, according to the associative network theory (Keller, 1993). As a result, previous research that studied the effects of sponsorships and brand image transfer for the sponsor can provide valuable insights of the role of brand image transfer for the sponsee. Especially, the literature on celebrity endorsement and sport event sponsorship offers valuable insights into the effects of sponsorship. If an event or celebrity is repeatedly paired through sponsorship with a brand having strong and favorable associations, then the event or celebrity is likely to obtain the same favorable and strong associations. However, there is a clear need for further research into image transfer through sponsorship relationship and the strength of the directionality of the image transfer.

(13)

As a result, image transfer is the transfer of some of the associations that are linked between a sponsor and a sponsee during a sponsorship, when both sponsor and sponsee benefit from each other associations by transferring their brand image and brand meaning to each other. The aim of the sponsorship is to evoke positive associations, feelings and attitudes towards each other by closely linking the brand image of the sponsor and sponsee (Grohs et al., 2004; Gwinner, 1997).

Authors differ on which definition they use for brand image fit, since brand image congruence, brand image match, brand image similarity and brand image

relatedness are also mentioned in the literature (Cornwell, 1995; Cornwell, Pruitt, and Van Ness, 2001; Ferrand and Pages, 1996; Gwinner 1997; Gwinner and Eaton, 1999; Johar and Pham, 1999; McDaniel, 1999; Musante, Milne and McDonald, 1999; Rifon et al., 2004; Speed and Thompson, 2000). However, there is consensus that brand image fit is important. Hence, the effect of brand image transfer of a sponsorship is moderated by the perceived fit between the brand image of the company and the brand image of the sponsored entity or cause (Becker-Olsen and Simmons, 2002). The significant impact of consumers’ perceived brand image fit between sponsor and sponsee have been discussed in sponsorship literature (Grohs & Reisinger, 2005; Gwinner, 1997; Gwinner & Eaton, 1999; Koo, Quarterman, & Flynn, 2006; McDonald, 1991; Speed & Thompson, 2000). Brand image transfer will be stronger when the brands have a higher brand image fit (Simonin and Ruth, 2003). Hence, brand image fit could be used to explain brand attachment and willingness to pay (Jagre et al., 2001; McDaniel 1999).The perceived fit is the degree of similarity between the sponsor and sponsee based on the image and the function. Brand knowledge, power, fit and quality are considered as influences on the brand image transfer process (Smith, 2004). Keller (1993) found six types of associations that may constitute a brand image: product attribute, user imagery, brand

personality, functional benefits, experiential benefits, and symbolic benefits. Roth and Romeo (2000) found that similar product categories and higher perceived fit between co-promotion partners contributed to a higher willingness to pay and satisfaction.

(14)

2.4 Brand image transfer from event sponsorship

Brand associations can be influenced when a brand becomes linked with an event through sponsorship activities, hence the event image is transferred to the brand and vice versa. Bashiri et al. (2010) and Walliser (2003) studied the goals and effects of sponsorship in sport events. Event sponsorship differs from regular

advertisement, since it creates a linkage between the sponsor and the event. This linkage is subjected by the interpretation by consumers (D’Astous and Bitz, 1995; Dean, 1999).

In line with the previous section (2.3), the amount and strength of associations that are linked between the sponsor and the event is called brand image fit. The

literature distinguish between functional related fit and image related fit to refer to the potential fit between an event and a brand that act as a sponsor (Gwinner, 1997; Gwinner and Eaton, 1999). Functional related fit describes the thematic relatedness between a sponsor and an event and the actually utilization of a product during the event. Image related fit encompasses the attributes associated with the brand image of a sponsor and a sponsored event, also in terms of prestige (Grohs et al., 2004; Gwinner, 1997). The image transfer between the event and the brand is more effective when the degree of image and/or functional fit between the event and the brand that acts a sponsor is higher. In addition, the magnitude of image transfer depends on two factors, sponsorship leverage and event-sponsor fit (Grohs et al., 2004). The image transfer will be less when brands have less or no brand image fit with the event (Gwinner and Eaton, 1999), however image transfer always exist (Grohs et al., 2004). Moderating variables between the event image and brand image are: degree of similarity, level of sponsorship, event frequency and product involvement (Gwinner, 1997)

Low brand image fit of a sponsorship generated less favorable thoughts and attitudes toward the sponsorship. In addition, it generated less favorable affective

(15)

and behavioral responses; hence it has a negative impact on the sponsorship and the willingness to pay. Concluded, incongruity or low brand image fit is negatively

valued. In addition, low brand image fit makes people less certain of the firm’s positioning and audience targeting, which result in a lower willingness to pay

(D’Astous and Bitz, 1995). Furthermore, low brand image fit or incongruity decreases the brand equity relative compared to a situation where no sponsorship occurs (Becker-Olsen and Simmons, 2002). When the sponsor-event link is weak, consumer evaluations are less positive, however this evaluation increases as sponsorships are continuous (D’Astous and Bitz, 1995). People with close ties to the event may be less critical of low brand image fit than a more general audience (Becker-Olsen and Simmons, 2002). Concluded, the level of brand image fit impacts the attitude toward the sponsor, which has a positive influence on the consumer’s willingness to pay (Gwinner and Bennet, 2008).

Chavanat et al. (2009) studied the effects of multiple sponsorships on a sport event. Multiple sponsorships occurs when one sponsee has several partners in the same events, as well as one organization that is partner of several sponsee. They studied the impact of commercial sponsorship on the intention to purchase sponsor products in relation to brand image and brand attachment. Multiple sponsorship create interactions between the sponsor and the sponsees. In addition, multiple sponsorship would activate brand cognitive, affective, and conative consumer’s perceptions. The hierarchy of effects are cognitive first, then affective and at last conative. Hence, first the brand image is established, then there is brand attachment and at last there is a purchase intention that could be measured by the willingness to pay for the sponsor’s products. Since brand image transfer works in two directions, it will affect the willingness to pay for the sponsees products as well.

2.5 Brand image transfer from celebrity endorsement

As mentioned before, celebrity endorsement literature offers valuable insights in the effects of brand image transfer, since celebrities are often used as endorsers in

(16)

advertising and sponsorships; hence they act as the sponsee. Till (1998) found that celebrity endorsement creates an associative link between a brand and a celebrity, hence each is part of the other’s association set.

The brand image fit between the brand and the celebrity seems to be important in celebrity endorsement. This fit between a brand and a celebrity occurs when some of the association that are part of each other association set have similarities and are shared (McCracken, 1989). If a celebrity and a brand are repetitive paired, then this strengthens this link between the celebrity endorser and the sponsoring brand (Till, 1998). This will lead to higher brand image fit and is as much important as the attitude toward the celebrity when considering a celebrity endorsing a brand (Fleck and Korchia, 2009). Fit of the image of the celebrity and the brand image will have a positive impact on the brand, because of the brand image transfer process. In addition, the stronger the link between the sponsor and the sponsee, the more impact the celebrity associations will have on the brand. (Kahle and Homer, 1985; Kamins, 1989 and 1990, McCracken, 1989, Misra and Beatty, 1990; McCracken 1989; Lynch and Schuler, 1994; Till, 1998).

However, literature also found that a moderate level of incongruence between an expectation and an object may be beneficial as well, provided it is perceived as interesting and positive. A slight inconsistency between the sponsor and the sponsee should lead to enhanced recall and more favorable attitudes and activities than in the case of total consistency. However, if the incongruence is too big, then it will have a negative impact (Fleck and Korchia, 2009).

In addition, negative publicity for celebrity endorsers may tarnish a brand; since the brand image transfer works in two directions, hence people associate the celebrity endorser with the brand as well (Gwinner, 1999; Till, 1998). However, the literature did not study the effects of negative publicity for the sponsor on the sponsee.

(17)

Furthermore, celebrities are perceived more negatively when consumers were told if this celebrity endorsed multiple products, since respondents consider the endorser as less distinctive. However, this does not affect the attitude toward the ad, the attitude toward the brand or the willingness to pay of the brand (Tripp, Jensen and Les Carlson, 1994). Concluded, celebrity endorsement can create positive effects for the brand, as well as for the endorser.

As a result of this literature review, the hypotheses are:

Hypothesis 1a: The display of a commercial sponsor with a high brand image fit with the sponsee will lead to a higher willingness to pay for the festival.

Hypothesis 1b: The display of a commercial sponsor with a low brand image fit with the sponsee will lead to a lower willingness to pay for the festival.

Hypothesis 1c: The display of multiple commercial sponsors with both a high and low brand image fit will lead to a higher willingness to pay for the festival.

In addition, the second hypothesis will compare the difference between a sponsor with a high brand image fit and a low image fit with the sponsee.

Hypothesis 2: The display of a commercial sponsor with a high brand image fit with the sponsee will lead to a higher willingness to pay compared to the display of a low brand image fit with the sponsee.

(18)

The third hypothesis will test the level of moderation of previous attendance on the festival on the willingness to pay for the festival.

Hypothesis 3a: Previous attendance will moderate the effect of the display of a commercial sponsor with a high brand image fit with the sponsee on the willingness to pay.

Hypothesis 3b: Previous attendance will moderate the effect of the display of a commercial sponsor with a low brand image fit with the sponsee on the willingness to pay.

Hypothesis 3c: Previous attendance will moderate the effect of the display of a multiple commercial sponsors with both a high brand image fit with the sponsee and a low brand image fit with the sponsee on the willingness to pay.

(19)

3. Methodology

This section focuses on the empirical setting of the research. It consists of the following sections: the empirical setting, the brand image fit measurement, the research design, the data collection, the sample and the variables.

3.1 Empirical setting

The empirical setting of this research was the festival ‘A Campingflight to Lowlands Paradise’ or better known as the Lowlands festival. Lowlands is a three days, commercial, yearly festival that already exist since 1993 (D’Astous and Bitz, 1995). Lowlands offers a variation of entertainment. The main focus of Lowlands is

alternative music, however they also offer literature, movies, cabaret, ballet, theatre and comics. In addition, they focus on sustainablity, politics and science. Lowlands is held at the eventarea next to Walibi Holland in Biddinghuizen, the Netherlands and it usually takes place in August. Lowlands is organized by Mojo, the biggest concert organizer in the Netherlands.

Experiment – An appropriate research strategy must be chosen in order to test the hypotheses introduced in the literature review and to answer the research question. Most appropriate is to test the hypotheses through an empirical quantitative

research, in particular an experiment. The experiment research strategy is most suitable for one good reason, related to testing the hypotheses. An experiment is perfect for examining the manipulation of an independent variable, in our case a sponsor, and measuring the effects on dependent variables, in our case willingness to pay. The results of this experiment consist of comparing the willingness to pay of different sponsor.

3.2 Brand image fit operationalization

The brand image is determined by the genres that were part of the Lowlands 2014 line up, since genres are a signal of brand image. Most music festival cover multiple music genres (Leenders et al., 2005; Paleo and Wijnberg, 2006), hence genre-labels

(20)

have a crucial role in the organizational identity of a festival (Leenders et al., 2005; Paleo an Wijnberg, 2006).

Brands often use music in their commercials. Hence, genre overlap is a measure of brand image fit. The music genres that are featured in the last thirty videos online on the sponsors YouTube channel are compared with their overlap with the line up of Lowlands 2014. As a result, high brand image fit is operationalized if a sponsor and Lowlands share similar genres and the amount of bands in the same genre.

The Lowlands 2014 line up featured 94 music styles (defined by allmusic.com). The main music style at Lowlands (also defined by allmusic.com) is Alternative/indie rock (35). In addition, Indie rock (11), Alternative singer/songwriter (9) and Club/dance (9) are music styles that are featured a lot at Lowlands as well. Indie folk (6), Indie electronic (5), Heavy Metal (5), Alternative R&B (4), Contemporary singer/songwriter (3), Alternative pop/rock (3), Adult alternative pop/rock (3) and Dubstep (3) were featured at Lowlands 2014 as well. Music styles/genres that were counted less than three times were excluded from this count.

As mentioned before, the music displayed in the videos of the sponsor on their YouTube channels were used to measure the brand image fit with the festival. Lowlands has eight sponsorship agreements, next to their partnership with

radiostation 3FM, music platform 3voor12 and public broadcaster VPRO. The eight sponsors of Lowlands 2015 are Bacardi, Converse, Douwe Egberts (DE), Heineken, Hi, Ilovemyears.nl, Nedap and NRC Next.

It is the first year that Heineken is the sponsor of the Lowlands festival; hence it is not suitable to test. In addition, McDonald (1991) found that many people believe strongly that alcohol or tobacco sponsorships are not beneficial for the entity.

(21)

Furthermore, Ilovemyears.nl, Nedap and NRC Next have a limited of videos on their YouTube channel, respectively 1, 8, 6, hence they are also not suitable to test the genre overlap with Lowlands. The videos on the YouTube channels of the other sponsors (Bacardi, Converse, DE and Hi) were observed. All artists that featured in the video and the music of an artist that was featured in the video were counted and categorized by their genre. The following tables show the results.

Table 1: The number of artists featured on the online channels of the sponsors and by allmusic.com

Bacardi Converse DE Hi

Number of artists featured in videos 17 35 9 32

Number of artists on allmusic.com 12 21 6 28

Numbers of artist that are specified by music style at allmusic.com

7 13 6 11

Number of different music styles in videos

21 25 20 25

Number of total music styles in videos 27 41 21 39

Table 2: Genres and music styles overlap between the online channels of the sponsors and Lowlands 2014

Lowlands genres: Bacardi Converse DE Hi

Alternative indie/rock (35) 2 8 2 4 Indie rock (11) - 2 - - Alternative singer/songwriter (9) 1 1 - 1 Club/Dance (9) 2 - - 2 Indie folk (6) - - - - Indie electronic (5) 1 3 - 2 Heavy Metal (5) - 2 - - Alternative R&B (4) - - - 1 Contemporary singer/songwriter (3) - - 1 - Alternative pop/rock (3) - 2 1 -

Adult alternative pop/rock (3) - - 1 1

(22)

High brand image fit is based on the high amount of genre overlap between the sponsor and the sponsee. Of all eleven main music styles featured at Lowlands 2014, six were featured in Converse and Hi commercials. However, the main genre

Alternative indie/rock is featured in 8 videos of Converse commercials compared to four in Hi. In addition, Converse feature music styles that are more often featured at Lowlands 2014 than Hi. At last eighteen music styles that were among the main music styles at Lowlands 2014 were counted in the Converse videos, compared to eleven in the Hi videos. Concluded, Converse has the highest genre overlap and hence the highest brand image fit with Lowlands (table 1 and table 2).

Low brand image fit is determined by the low amount of genre overlap between Lowlands and the sponsor. Bacardi and Douwe Egberts both only feature two times Alternative indie/rock in their videos and both only feature four main music styles of Lowlands 2014 in their videos. However, Bacardi feature music styles that are

featured more on Lowlands 2014 than Douwe Egberts. In addition, Bacardi feature three genres in their videos that were not featured at Lowlands 2014, however Douwe Egberts featured twelve genres in their videos that were not featured at Lowlands 2014. Concluding, Douwe Egberts has the lowest genre overlap with Lowlands and hence is perceived of having the lowest brand image fit with Lowlands (table 1 and table 2).

Converse – Converse is an American shoe company, founded in 1908, that is known for their Chuck Taylor All Star shoe. All shoes have the distinguished Star Insignia logo on the shoe. Converse sells it footwear in over 160 countries and is a subsidiary of Nike, Inc. One of Converse main sponsorship activities is Converse Rubber Tracks. Converse offers free studio time, mixing, mastering and publicity for upcoming artists. Mojo (the organizer of Lowlands) state that the link between Converse and the festival is obvious. In addition, Converse won the ‘Cool Brand Awards’ and is one

(23)

of the biggest brands on Facebook. The 100-anniversary of Converse was the direct reason of sponsoring the Lowlands festival.

Douwe Egberts – Douwe Egberts is the main brand name of D.E. Master Blenders, founded in 1753. Douwe Egberts is a Dutch tea and coffee brand with offices in 14 countries over the world. Douwe Egberts sponsors many cultural institutions as the ‘Rijksmuseum Amsterdam’ and the ‘Stadsschouwburg Utrecht’. On the website of Mojo (www.mojo.nl) the organization state that the visitors of Lowlands valued Douwe Egberts more positively compared to young people that did not attend Lowlands. Hence, they explicitly state the brand goals of Douwe Egberts on their website and not the goals for the Lowlands festival.

3.3 Data collection

The main goal is this experiment is to show the audience different treatments and test the differences on the willingness to pay. An auction was set up with the help of Veylinx. Veylinx is an online platform that conducts auctions. Individuals are invited to subscribe their e-mail address at www.veylinx.nl and are randomly invited to participate in an auction. Respondents are led to an unique auction where they see a picture of a product on which they can bid. The respondents only have a limited amount of time to bid on the product and they do not know the bids of other

respondents, hence it is a sealed bid single-item auction. Some control questions are asked after the bid. The auction closes after one day and then the highest bidder wins the auction, however he pays the second highest bid for the product. Hence, Veylinx is a Vickrey auction (Vickrey, 1961) The Vickrey auction will lead to the highest expected outcome, since bidders are given an incentive to bid their maximum willingness to pay due to counter speculations. In other words, it

encourages the respondents to bid truthfully. They do not know the bids of others and it is also possible to bid zero euros. Hence, a Vickrey auction will result in the respondent maximum willingness to pay and as a result the item ships to the person who values the product at the highest price.

(24)

Veylinx is usefull to obtain a huge dataset, which is often a struggle for students. Every student that uses this concept is asked to recruit 100 new respondents by subscribing to Veylinx. Lots of variables are already in the dataset

Another advantage is that Veylinx makes it easy to ask additional questions after the auction. The last advantage is that Veylinx provides the opportunity for testing more than one treatment, hence it makes it possible to set up an experiment. However, counter arguments are that a Vickrey auction might lead to overbidding (Kagel, Hardstadt & Levin, 1987).

3.4 Research design

Four treatments were designed for the experiment with Veylinx. The sponsoring brand was altered during the Veylinx experiment but everything else was kept constant. The ad should be simple and free of irrelevant design elements to obtain the best results (Heckler and Childers, 1992; Pham and Vanheule, 1997). Also, the association should focus on the link between the event and brand; hence the advertisement should focus on communicating the brand-event pairing (Till, 1998). Concluded, an associative link between the event and the brand increases when there are few other elements in the advertisement. The respondents of Veylinx were randomly assigned to one of the four experimental groups. As a result, it was

possible to auction one single festival ticket for Lowlands 2015.

- Treatment 1 showed no sponsor

- Treatment 2 showed the logo of the sponsor, which has a high brand image fit with the festival (Converse).

- Treatment 3 showed the logo of the sponsor, which has a low brand image fit with the festival (Douwe Egberts).

(25)

In this setting, the respondents of Veylinx got a unique email in which they got a random link to one of the four treatments. The respondents were asked to bid on one single ticket for Lowlands 2015.

In addition, the respondentens were asked to respond to five questions/statements.

- Have you visited Lowlands before? (Ben je eerder op Lowlands geweest?)

Yes – No

- Are you familiar with the brand ‘Converse’? (Ken je het merk Converse?)

Yes – No

- Are you familiar with the brand ‘Douwe Egberts’? (Ken je het merk Douwe Egberts?)

Yes – No

- Converse has the same associations as Lowlands! (Converse brengt dezelfde associaties bij mij naar boven als bij Lowlands)

Totally agree – agree – neutral – disagree – totally disagree

- Douwe Egberts has the same associations as Lowlands! (Douwe Egberts brengt dezelfde associaties bij mij naar boven als bij Lowlands)

Totally agree – agree – neutral – disagree – totally disagree

Treatments – The four treatments were edited by the researcher, using material found at google images. Those images were manipulated, which led to four different treatments. All treatments displayed the logo of Lowlands (A campingflight to Lowlands paradise). In addition, all treatments display the date (21+22+23 Aug 2015), location (Biddinghuizen) and the text: ‘Eén ticket voor Lowlands 2015’. The first treatment didn’t display a sponsor. The second treatment displayed the image

(26)

of Converse. The third treatment also displayed the image of Douwe Egberts. The four treatment displayed both, Converse and Douwe Egberts, images.

Advantages – Due to a few reasons, Lowlands is a relevant case for this research. First, timing wise, Lowlands festival takes place in August 2015. Hence, the ticketsale for Lowlands festival overlap with the period in which this thesis is written and the tickets are auctioned (January 2015 - July 2015). Second, Lowlands is a Dutch festival, as so the respondents of Veylinx. Third, Lowlands is sponsored by many brands, hence it is a relevant and suitable setting to study sponsorship effects.

Disadvantages – Lowlands is not a new festival, so respondents could know the price and the sponsors of the festivals. It might also be possible that the brands are not recognized by respondents, hence the experiment check if the respondent is familiar with the brands. Third, it might be possible that Lowlands is not appealing to the respondent, because of her interest in other music genres. At last, only one single ticket is auctioned, so the final bid of the respondents may also depend on the availability of friends.

3.5 Sample

Due to the use of Veylinx, there was access to an existing panel consisting of people who subscribed for the platform. In order to use Veylinx, a total of 100 extra

respondents had to be recruited, which resulted in a huge penal.

From the entire database, a total of 870 participants opened the actual link for the auction and gave information for the dependent variable by bidding on the festival ticket, while 790 respondents finished the whole auction by bidding on the festival ticket as well as filling in all the questions. Of these 870 participants, 2 were outliers

(27)

in terms of age, leaving 868 participants with an oldest participant of 80 years old and the youngest participant of 16 years old.

This remaining sample of 868 participants consists of 476 men (54,8%) and 392 women (45,2%). The average age of the sample was 41,4 years old. This was divided into an average age of 41,6 for men and an average age of 41,2 for women. The sample average age for men is 0.4 years higher than the average age for women.

Table 3: Age distribution of the respondents

Data (table 3) shows that the age is a 0,299 positively skewed and the kurtosis is -1,044, so it is normal distributed since values for asymmetry (a.k.a. skewness) and kurtosis between -2 and +2 are considered acceptable in order to prove normal univariate distribution (George & Mallery, 2010).

(28)

3.6 Variables

The dependent variables were collected during the live second price sealed-bid Vickrey auction (Vickrey, 1961). In addition, there was access to demographic data from the respondents such as age, education and gender. Also, control questions were asked to control for certain assumptions.

Dependent variable - During the data collection one dependent variable was

collected: willingness-to-pay. This dependent variable is the willingness to pay of the respondents for a single Lowlands 2015 ticket after seeing the Lowlands ticket in combination with no sponsor, Douwe Egberts, Converse or both sponsors.

Independent variable - The independent variable in this research are the four different treatments that differ on what sponsor was displayed. The treatments were manipulated, as described in the treatments part, resulting into four different treatments: no sponsor, Converse, Douwe Egberts, and Converse/DE.

Moderating variables - Previous attendance was tested as a moderating variable in hypothesis 3.

Control variables – The treatments were controlled for age and gender in an additional regression analysis (male = 0 and female = 1). In addition, the familiarity with the sponsors were checked and the assumption that Converse has a high brand image fit with Lowlands and Douwe Egberts has a low brand image fit with Lowlands were tested by asking the respondents if the sponsor and Lowlands had the same brand associations.

(29)

4. Results

The result section is divided into four parts. First, the data is described with the use of descriptive statistics. Then the control questions from Veylinx are described with the use of descriptive statistics. In addition, it is tested on reliability. The section is closed with tests related to correlations, the three hypotheses and some additional analysis.

4.1 Descriptive statistics

In this section the descriptive statistics about the data of the different treatments are provided. The assumptions of normality in the distribution and homogeneity of variance is also tested. The main findings are described below.

4.1a Willingness-to-Pay

The Veylinx experiment led to the following distribution for the four treatments: no sponsor (N = 224), Converse (N = 193), Douwe Egberts (N = 204), Converse/DE (N = 247), while the total sample is 868. The N value for all treatments is sufficient and even though the distribution of the respondents over the treatments is not completely equal it is not expected to cause problems. This led to the following mean bids for the four treatments: no sponsor (M = 20.08 euro), Converse (M = 19.44 euro), Douwe Egberts (M = 16.11 euro), Converse/DE (M = 19.88 euro), while the mean of the total sample is 18.95 euro. The distribution of the treatments is: no sponsor (25.8%), Converse (22.2%), Douwe Egberts (23.5%), Converse/DE (28.5%).

Data show that the willingness to pay is skewed and non-normally distributed. The bids are clustered towards zero euros. The data shows a positive skewness of 2.369 and kurtosis of 5.631. Values for asymmetry (a.k.a. skewness) and kurtosis between -2 and +-2 are considered acceptable in order to prove normal univariate distribution (George & Mallery, 2010). In addition, The Kolmogorov-Smirnov and Shapiro-Wilk

(30)

are both significant (p = .00), hence the data shows that it is not normally distributed. At last, the data shows no homogeneity of variance, since p= .198.

4.1b Willingness-to-pay (excluding under median bids)

Another way to test the willingness to pay data is to exclude the respondents who bid les then the median, hence only the relevant respondents who have an actual willingness to pay that is higher than the median are included.

This led to the following distribution for the four treatments: no sponsor (N = 112), Converse (N = 97), Douwe Egbters (N = 103) and Converse/DE (N=124), while the total sample is 436. The N value for all treatments is sufficient and even though the distribution of the respondents over the treatments is not completely equal it is not expected to cause problems. This led to the following mean bids for the four

treatments: no sponsor (M = 40.16 euro), Converse (M = 38.67 euro), Douwe Egberts (M = 31.90 euro), Converse/DE (M = 39.49 euro), while the mean of the total sample is 37.69 euro.

The distribution of the treatments is: no sponsor (22.94%), Converse (22.25%), Douwe Egberts (23.62%), Converse/DE (28.44%).

The skewness of this sample is 1.488 and the kurtosis is 1.748. Values for asymmetry (a.k.a. skewness) and kurtosis between -2 and +2 are considered acceptable in order to prove normal univariate distribution (George & Mallery, 2010), however the Kolmogorov-Smirnov and Shapiro-Wilk are both significant (p = .000), hence it shows that the data is not normally distributed. At last, the data shows no homogeneity of variance, since p= .274.

(31)

4.1c Willingness to pay (including all bids)

The bids were transformed into a natural logarithm of the bids in order to normalize the distribution of the bids. The natural logarithm including all bids has the same sample as the regular willingness to pay (N = 868); hence the N values for all treatments are sufficient. This led to the following mean bids for the four

treatments: no sponsor (M = 3.77 euro), Converse (M = 3.83 euro), Douwe Egberts (M = 3.55 euro), Converse/DE (M = 4.13 euro), while the mean of the total sample is 3.84 euro. The distribution of the treatments is:

The skewness of the sample of the natural logarithm is 0.133 and the kurtosis -1.802. Values for asymmetry (a.k.a. skewness) and kurtosis between -2 and +2 are

considered acceptable in order to prove normal univariate distribution (George & Mallery, 2010). However, the Kolmogorov-Smirnov and Shapiro-Wilk are both significant (p = .000), so this proves that the natural logarithm is still not normally distributed. At last, the data shows no homogeneity of variance, since p= .508.

4.1d Willingness to pay (Natural Logarithm excluding under median bids)

The natural logarithm of all bids is clustered around zero euros. Hence, for the last descriptive analysis all bids under median are excluded, hence the natural logarithm excluding under median bids is tested. Veylinx recommends this way of data testing.

The natural logarithm excluding under median bids has the same sample as the regular willingness to pay excluding under median bids (N = 436), hence the N values for all treatments are sufficient. This led to the following mean bids for the four treatments: no sponsor (M = 7.54 euro), Converse (M = 7.56 euro), Douwe Egberts (M = 7.04 euro), Converse/DE (M = 7.75 euro), while the mean of the total sample is 7.49 euro.

(32)

The distribution of the treatments is: no sponsor (22.94%), Converse (22.25%), Douwe Egberts (23.62%), Converse/DE (28.44%).

The data is negatively skewed (-1.515) and the kurtosis is 4.479, hence it is a leptokurtic distribution. Values for asymmetry (a.k.a. skewness) and kurtosis between -2 and +2 are considered acceptable in order to prove normal univariate distribution (George & Mallery, 2010). So the skewness level is acceptable, however the kurtosis violates the rule of normal distribution. In addition, the Kolmogorov-Smirnov and Shapiro-Wilk are both significant (p= .000), so the natural logarithm is still not normally distributed. At last, the data shows homogeneity of variance, since p= .001.

Table 4: Average willingness to pay for all samples

N all bids M M natural logarithm N above median M M natural logarithm No Sponsor (929) 224 20.08 3.77 112 40.16 7.54 Converse (930) 193 19.44 3.83 97 38.67 7.56 Douwe Egberts (931) 204 16.11 3.55 103 31.90 7.04 Convers/Douwe Egberts (932) 247 19.88 4.13 124 39.49 7.75 Total/mean 868 18.95 3.84 436 37.69 7.49

(33)

Table 5: The average willingness willingness to pay of the natural logarithm excluding under median bids by the different treatments

Table 6: Graph of the distribution of the average willingness to pay of the natural logarithm excluding under median bids

(34)

4.1e Concluding

Concluding, the data on willingness to pay seems to violate the assumption of normality in the

distribution. However, the natural logarithm of the willingness to pay excluding under median bids has homogeneity of variance. In addition, the histogram (table 6) shows that the data is almost normal distributed. Because of the

assumption of normality in the natural logarithm we will continue analysing the data with statistical tests by only using the natural logarithm excluding under median bids, hence the sample size is 436. This sample is divided into 252 men and 184 women. The average age of this sample is 37.51 (38.25 for men and 36.48 for women). The graph (table 7) shows that the distribution of the age is positively skewed (.633) with a kurtosis of -.742.

Running ANOVA, t-tests and regressions analysis in the hypothesis section might lead to inaccurate results because of the violation of normal distribution.

Table 7: Age distribution for the average willingness to pay for the natural logarithm excluding under median bids

(35)

4.2 Control questions Veylinx

4.2a Visit Lowlands before

The 436 respondents were asked if they have visited Lowlands before. 305 out of the 436 respondents have visited Lowlands before, however there were 38 missing values. Hence, 23.4% of the respondents have visited Lowlands before.

4.2b Familiarity

An important aspect for the reliability of this research is to test if the respondents knew the sponsor related to the different treatment. The respondents were asked if they were familiar with both sponsors (Converse and Douwe Egberts). 348 out of the 436 respondents were familiar with Converse, however there were 38 missing values. 393 out of the 436 respondents were familiar with Douwe Egberts, however there were 38 missing values. Hence, 87.4% of the respondents are familiar with Converse and 98.7% is familiar with Douwe Egberts.

4.2c Brand association sponsor and Lowlands

The respondents were asked if they think that the sponsors have the same brand associations as Lowlands. 791 respondents (77 missing values) replied to those statements, hence data for 398 respondents (38 missing values) was gathered if the under median bids were excluded.

Converse and Lowlands

Including all bids Mean = 3.81

Median = 4

Excluding under median bids Mean = 3.66

Median = 4

Table 8: Likert scale if Converse and Lowlands have the same associations for the respondent. 1=totally agree while 5=totally disagree

(36)

Douwe Egberts and Lowlands

Including all bids Mean = 4.29

Median=5

Excluding under median bids Mean = 4.15

Median = 5

4.2d Concluding

Both sponsors are well recognized, although Douwe Egberts is recognized more than Converse.

The respondents believe that Converse has on average more the same associations as Lowlands compared to Douwe Egberts, since the median and mean is lower. This proves that Converse has a higher brand image fit regarding the genres than Douwe Egberts, however most respondents think that both sponsors do not have that much in common regarding the brand associations. Summarizing this means that all the treatments seem reliable.

4.3 Correlation

In this part the correlations between all variables are explored and described.

Table 9: Likert scale if Douwe Egberts and Lowlands have the same associations for the respondent. 1=totally agree while 5=totally disagree

(37)

M

Willingness to pay

Gender Age No Sponsor (willingness to pay) Converse (willingness to pay) Douwe Egberts (willingness to pay) Converse/DE (willingness to pay) Previous attendance Familiar with Converse Familiar with Douwe Egberts Converse association with Lowlands Douwe Egberts association with Lowlands Willingness to pay 7.49 - Gender -.024 - Age -.418** _-.062 _-

No Sponsor (willingness to pay) 7.54 .022 .050 -.008 -

Converse (willingness to pay) 7.56 .026 .001 -.051 -.315** _-

Douwe Egberts (willingness to pay) 7.04 -.160** _.017 _.064 _-.327** _-.297** _-

Converse/DE (willingness to pay) 7.75 .106* _-.065 _-.005 _-.371** _-.337** _-.351** _-

Previous attendance .328** _.115* _-.235** _.005 _-.004 _.028 _-.028 _-

Familiar with Converse .196** _.042 _-.315** _.012 _-.010 _-.085 _.076 _.138** _-

Familiar with Douwe Egberts .076 .007 -.046 -.038 .008 .007 .023 -.044 .229** _-

Converse association with Lowlands -.200** _.021 _.170** _.023 _-.051 _.144** _-.108* _-.121* _-.122* _.135** _-

Douwe Egberts association with Lowlands -.079 .018 .003 .050 -.067 .054 -.037 -.130** _.097 _.195** _.423** _-

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

(38)

The full correlation matrix shows the average willingness to pay, as well as all correlations between the variables.

The table shows that age, the Douwe Egberts treatment and the same association between Converse and Lowlands had significant negative correlations with the willingness to pay (resp., r = -.418, p < .01; r = -.16, p < .01; r = -.200, p < .01). In addition, the

Converse/DE treatment, previous attendance and familiarity with Converse had significant positive correlations (resp., r = .106, p < .05; r = .328, p < .01; r = .196, p < .01). In addition, gender, familiarity with Converse, the same associations between Converse and Lowlands and the same associations between Douwe Egberts and Lowlands and previous attendance were positively correlated (resp., r = .115, p < .05; r = .138, p < .01; r = -.121, p <.05; r = .-130, p < .01).

The correlation matrix also show correlation effects for the Veylinx questions. It shows that previous attendance, familiarity with Converse and the same association between Converse and Lowlands are significant correlated (resp., r = -.235, p < .01; r = -.315, p < .01; r = .170, p < .01). In addition, familiarity with Douwe Egberts and the same associations between Converse and Lowlands are correlated with familiarity with Converse (resp., r = .229, p < .01; r = -.122, p < .05) and the same associations between Converse and Lowlands and Douwe Egberts and Lowlands are significant correlated with familiarity with Douwe Egberts (resp., r = .135, p < .01; r = .195, p < .01). At last, the same associations between Converse and Lowlands is both positively correlated with the same associations between Douwe Egberts and Lowlands and the Douwe Egberts treatment (resp., r = .423, p < .01; r = .144, p < .01).

Regarding the hypothesis, the Converse treatment, Douwe Egberts treatment and Converse/DE treatment are correlated with the no sponsor treatment (resp., r = -.315, p < .01; r = -.327, p < .01; r = -.371, p < .01). In addition, the Douwe Egberts treatment and

Converse/DE treatment are both significant negatively correlated to the Converse treatment (resp., r = -.297, p < .01; r = -.337, p < .01). At last, the Douwe Egberts treatment is significant correlated with the Converse/DE treatment (r = -.351, p < .01).

The variable ‘Education’ is left out of this research, since most respondents are WO or HBO students (72.6%).

4.4 Hypotheses

4.4a Hypothesis 1

Hypothesis 1a: The display of a commercial sponsor with a high brand image fit with the sponsee will lead to a higher willingness to pay for the festival

Hypothesis 1b: The display of a commercial sponsor with a low brand image fit with the sponsee will lead to a lower willingness to pay for the festival

Hypothesis 1c: The display of multiple commercial sponsors with both a high and low brand image fit will lead to a higher willingness to pay for the festival

(39)

The willingness to pay is measured as natural logarithm excluding under median bids. The display of a commercial sponsor is measured by comparing the treatment with no sponsor with the treatments with a sponsor. The single treatments with sponsor(s) were compared separately with the treatment with no sponsor (no sponsor vs. Converse, no sponsor vs. Douwe Egberts, no sponsor vs. Converse/DE). For this analysis ANOVA and independent sample t-tests were used.

Descriptive statistics show an average willingness to pay of 7.54 euro for the no sponsor treatment, 7.56 euro for the Converse treatment, 7.04 euro for the Douwe Egberts treatment and 7.75 euro for the Converse/DE treatment.

ANOVA shows that there is a statistically significant difference between the treatments (p = .006). Post hoc tests (Tukey HSD and Scheffe) show a significant difference between the Douwe Egberts and the Converse/DE treatment (p = .008 (p < .01)).

Hypothesis 1a: The no sponsor treatment had an average bid of 7.54 euro, while the Converse treatment had an average bid of 7.56 euro. Levene’s test is insignificant (p = .667), so homogeneity of variances within the populations cannot be assumed. The two groups did not differ significantly from each other with t (206) = -.097, p = .923 (not significant)

Hypothesis 1b: The no sponsor treatment had an average bid of 7.54, while the Douwe Egberts treatment had average bid of 7.04 euro. Levene’s test is insignificant (p = .085), so homogeneity of variances within the populations cannot be assumed. The two groups differed significantly from each other with t (213) = 2.065, p = .04 (p < .05)

Hypothesis 1c: The no sponsor treatment had an average bid of 7.54 euro, while the Converse/DE treatment had an average bid of 7.75 euro. Levene’s test is significant (p = .023), so the ‘T-test for equal variances not assumed’ provides the solution to violation of the assumption of homogeneous variances of the two treatments. As a result, the two groups did not differ significantly from each other with t(205) = -1.167, p = .244 (not significant).

The independent samples T-Tests shows a significant difference between the no sponsor treatment and Douwe Egberts treatment (p = .04), however there is no significant difference between the no sponsor treatment and Converse treatment (p = .923) as well as for the no sponsor treatment and Converse/DE treatment (p = .244).

Concluding, hypothesis 1 is partly (1/3rd) accepted, since hypothesis 1b is accepted and 1a and 1c are rejected. The bids for no sponsor (7.54 euro) were significant higher than the bids for Douwe Egberts (7.04 euro), so there is significant proof that the display of a commercial sponsor that has a low brand image fit with the sponsee will lead to a lower willingness to pay. Although the bids for Converse (7.56 euro) and Converse/DE (7.75 euro) were higher than the bids for no sponsor (7.54 euro), both differences are not significantly different.

(40)

4.4b Hypothesis 2

Hypothesis 2: The display of a commercial sponsor with a high brand image fit with the sponsee will lead to a higher willingness to pay for the festival compared to the display of a commercial sponsor with a low brand image fit with the sponsee.

The second hypothesis is tested by using an independent sample T-test. The treatment with Converse and treatment with Douwe Egberts were compared.

The Converse treatment had an average bid of 7.56 euro (N=97), while the Douwe Egberts had an average bid of 7.04 euro (N=103). Levene’s test is significant (p = .045), so the ‘T-test for equal variances not assumed’ provides the solution to violation of the

assumption of homogeneous variances of the two treatments. As a result, the two groups differed significantly from each other with t (179) = 2.12, p = .035 (p < .05).

Concluding, hypothesis 2 is accepted. The bids for Converse (7.56 euro) were significant higher than the bids for Douwe Egberts (7.04 euro), so there is significant proof that audiences have a higher willingness to pay for a festival that has a higher brand image fit with the sponsor compared to a festival that has low brand image fit with its sponsor.

4.4c Hypothesis 3

Hypothesis 3a: Previous attendance will moderate the effect of the display of a commercial sponsor with a high brand image fit with the sponsee on the willingness to pay

Hypothesis 3b: Previous attendance will moderate the effect of the display of a commercial sponsor with a low brand image fit with the sponsee on the willingness to pay

Hypothesis 3c: Previous attendance will moderate the effect of the display of a multiple commercial sponsors with both a high brand image fit with the sponsee and a low brand image fit with the sponsee on the willingness to pay

Running a regression with PROCESS, which includes the moderator ‘Previous attendance’ tests the third and last hypothesis. The moderated regression analysis with PROCESS shows a significant moderating effect of previous attendance on the Douwe Egberts treatment.

Hypothesis 3a studied if previous attendance is moderating the effect of display to the Converse treatment on the willingness to pay. The overall model is significant F (3, 394) = 23.9342, p < .01 (p= .00), R2= .1081. The predictor ‘previous attendance’ shows b =

1.2048, t (394) = 7.4272, p = .00 – significant (p < .01), hence if visited before, there is a 1.2048 increase in willingness to pay, however there is no interaction effect (p= .9405).

(41)

Hypothesis 3b studied if previous attendance is moderating the effect of display to the Douwe Egberts treatment on the willingness to pay. The overall model shows F (3, 394) = 28.4451, p < .01 (p = .00), R2= .1500. The predictor ‘previous attendance’ shows b = .9544, t (394) = 6.0272, p = .00 – significant (p < .01), hence if visited before, there is a .9544 increase in willingness to pay. In addition, the predictor Douwe Egberts shows b = -.8702, t (394) = -4.3902, p = .002 – significant (p < .01), if in Douwe Egberts treatment, there is a -.8702 decrease in the willingness to pay. The data shows interaction b=1.0859, t (394) = 3.1106, p = .002 – significant (p < .01). The slopes for Douwe Egberts predicting the willingness to pay for levels of attendance are:

0: No previous attendance, DE b= -.8702, t (394) = -.4.3902, p = .00 (p < .01)

1: Previous attendance, DE b= .2157, t (394) = .6261, p = .5316 (not significant)

So previous attendance is a statistically significant moderator in the Douwe Egberts treatment.

Hypothesis 3c studied if previous attendance is moderating the effect of display to the Converse/DE treatment on the willingness to pay. The overall model shows F (3, 394) = 25.3776, p < .01 (p = .00), R2= .1189. The predictor ‘previous attendance’ shows b =1.2416, t (394) = 7.0198, p = .00, which is significant (p < .01), hence if visited before, there is a 1.2416 increase in willingness to pay. In addition, the Converse/DE treatment shows b= .3847, t (394) = 2.4034, p= .017, which is significant (p < .05), hence if in the Converse/DE treatment, there is a .3847 increase in willingness to pay. However, there are no interaction effects (p = .679)

Table 10: Average willingness to pay for the natural logarithm excluding under median bids for the different treatments and by previous attendance

Variable N M

Previous attendance, no sponsor 24 8.07 euro

No previous attendance, no sponsor 77 7.48 euro

Previous attendance, Converse 21 8.52 euro

No previous attendance, Converse 70 7.34 euro

Previous attendance, Douwe Egberts 23 8.63 euro

No previous attendance, Douwe Egberts 67 6.59 euro

Previous attendance, Converse/DE 25 8.67 euro

No previous attendance, Converse/DE 91 7.54 euro

Concluding, hypothesis 3 is partly (1/3rd) accepted, since hypothesis 3b is accepted and 3a and 3c rejected. Previous attendance has a moderating effect on the willingness to pay in the Douwe Egberts treatment. However, previous attendance is not a significant moderator for the Converse treatment and Converse/DE treatment.

4.5 Additional analysis

In order to get every possible interesting finding out of the data, some additional analysis was done.

4.5a Regression analysis (without the brand association variables, including gender and age controls)

In the first step of this regression, two predictors were entered: gender and age. This model was statistically significant F (2, 394) = 39.45; p < .01 (p= .000) and explained 16,7% of variance in the willingness to pay. After entry of the treatments and other variables at Step 2 the total variance explained by the model as a whole was 25,8 % F (8, 388) = 16.906; p< .01 (p= .000). The introduction of the

The effects of sponsoring and brand image on the willingness to pay : an empirical investigation of a music festival

The effects of sponsoring and brand image on the willingness

to pay: an empirical investigation of a music festival

Table of Contents

The effects of sponsoring and brand image on the willingness

to pay: an empirical investigation of a music festival

1. Introduction

2. Literature review

3. Methodology

4. Results