Brand Loyalty of Mobile Operating Systems Factors defining brand loyalty in an online community setting by J. Veldmeijer March, 2012

(1)

Brand Loyalty of Mobile Operating Systems

Factors defining brand loyalty in an online community setting

by

J. Veldmeijer

(2)

Brand Loyalty of Mobile Operating Systems

Factors defining brand loyalty in an online community setting

by

J. Veldmeijer

University of Groningen Faculty of Economics and Business

Department of Marketing

March, 2012

Course name: Master thesis

Program: Marketing management and Marketing research

Jeroen Veldmeijer de Bree 33 8381 BS Vledder +31 (0) 625405968 J.Veldmeijer@student.rug.nl S1916300

Supervisor & examiner: dr. J.E.M. van Nierop

(3)

III

PREFACE

With the utmost respect and gratitude I thank the following individuals for their support and encouragement during the process of creating this enervating Master's thesis at the University of Groningen. First and foremost, I extend a ‘thank you’ that no words can ever do justice to dr. Erjen van Nierop; advisor, mentor and sounding board, who has always been positive and critical to the completion of the project. His investment in my success and the success of this work is significantly dependent on the interaction with him as a person and scholar, but also on his devotion and wisdom. Although he moved back to his home town, in order to express my gratitude the freely translated

Groninger expression sums it up: “I could certainly been worse off”. I would also like to thank Floortje Westerink for her advice and wisdom throughout this process; thank you for sharing your experience and joy. Moreover, I appreciate it tremendously that you have provided me the opportunity to work at times, when I needed to continue working relentlessly on the project. I would also like to thank the rest of my direct colleagues at HJ Heinz, who are second to none.

I would also like to thank my cohorts; Anne Marije and Peter. Thank you for your friendship and support during our four to six years at various institutions together. The laughter and joy we have shared on this ‘educational’ rollercoaster together made it that much sweeter to finalize the end note of this chapter in my life.

(4)

IV

MANAGEMENT SUMMARY

The vivid streams (Hepburn, 2011) of online communication might indicate a new source to increase brand loyalty. If looked closer however, no answer to how to increase online brand loyalty is found. A vast amount of academic research is currently present, but there is still a great void in research including the online aspect in brand loyalty. This is an important topic since the antecedents of online brand loyalty might be significantly different from offline brand loyalty and its subsequent behaviour. Additionally, understanding online brand loyalty is an essential requirement for optimizing community performance, customer retention and the exploitation of other online advantages.

A literature study is carried out to identify the determinants of online brand loyalty. In conducting this research the setting of a high involvement product is selected. The advantage of conducting this particular research in a high involvement product setting contributes to the odds that participants already exercise some form of online brand loyalty, increasing the exposure of statistical relationships. Factor analysis was performed in order to abstract components to test the model. Three determinants were found significant, which are functional value, product involvement and consumer satisfaction. Four steps (Baron and Kenny, 1986) were used to test whether the model was subject to moderated mediation. Among others, multiple regression analysis was performed to investigate if the moderated mediation construct exists. It is found that functional value, product involvement and (partially) consumer satisfaction through brand commitment significantly influence online brand loyalty. Additionally, brand commitment influences online brand loyalty proving the existence of a mediation effect. Moreover, brand trust significantly moderates the relationship between brand commitment and online brand loyalty. Thus, five variables have been found to significantly influence online brand loyalty. Additionally, evidence was found in support of a moderated mediation model.

(5)

V

1. INTRODUCTION

This chapter contains the introduction to this master thesis. The introduction will consist of the background problem, followed by an introduction to the topic an sich, the problem statement and the relevance of this thesis. Finally, the organization of the rest of this thesis will be described.

1.1 BACKGROUND PROBLEM

Currently, 57 percent of people talk to people more online than they do in real life (Hepburn, 2011). For instance, with an approximate value of about 80 billion US dollars Facebook is by far the largest online community to date. Facebook has in excess of 650 million active users (Bullas, 2011), 250 million users log in every day (Hepburn, 2011), and each user has about an average of 130 friends. This enormous potentially global database, not to mention traffic, is very interesting for companies in general and for marketers in particular to leverage. According to a study by the Chadwick Martin Bailey and iModerate Research Technologies (Renfrow, 2010), 60 percent of Facebook fans are more likely to recommend brands and 51 percent are more likely to buy the brands they follow. At first sight, these numbers might indicate a possible high level of potential in, for instance, brand loyalty. Online communities such as Facebook, are a perfect fit for creating and expanding brand loyalty for many marketers, assumed from the 16,000,000 fan pages (Bullas. 2011) that exist worldwide on Facebook.

By noticing a shift in 1) marketing thinking from consumer acquisition to consumer retention (Verhoef, 2003) and 2) an exponentially growing attention for online communities, i.e. social media, confusion emerged why and with what result Facebook fan pages were emerging. Business sense (the author runs a small internet enterprise, www.evansnwatson.nl) triggered to research how brand loyalty is defined in an online setting. Furthermore, which characteristics are deemed most important in the process of creating and / or expanding brand loyalty in an online setting.

1.2 INTRODUCTION TO THE TOPIC

(7)

7

important reasons to friend a brand on Facebook was to receive discounts, followed by ‘being a consumer‘ and a ‘desire to show others that they support the brand’ (Renfrow, 2010). This indicates a not clear cut answer to how to make people friends firstly, but more important what makes people brand loyal. Online communities are determined quite different than physical communities. For instance, the process of bonding and forming relationships is significantly different online compared to offline. In an offline setting people meet each other face to face, get to know another and form relationships. Online however, people get to know another, form a relationship, and then they may choose to meet the other face to face or not (Rheingold, 2000). More and more firms are recognizing the advantages of on-line brand communities, which include the opportunity for effective communication with consumers and of obtaining valuable ideas. Brand communities do not simply provide companies with an additional communication channel; they also enable the possibility of establishing linkages to devoted users. A brand community starts with its core asset, the brand itself, and grows by building relationships among members interested in the brand.

1.3 PROBLEM STATEMENT

The effects of the independent variables on brand loyalty are not specifically studied in an elaborated online fashion. Therefore the questions that remain are “what factors define brand loyalty in an online community setting?” and “which factors have the highest influence on brand loyalty in an online community setting?”

1.4 RELEVANCE OF THIS THESIS

(8)

8

long-term relationships with consumers (Heere et al., 2011). For instance, the marketing costs for attracting a new consumer have been found to be about six times higher than the costs of retaining an old one, brand loyal consumers are willing to pay higher prices and are less price sensitive (Rosenberg and Czepiel, 1984; Krishnamurthi and Raj, 1991; Reichheld and Sasser, 1990). Kuenzel and Halliday (2010) argue for instance that people go beyond their personal identity to develop a social identity, by identifying and associating themselves with brands in order to reflect and reinforce their self-identities. This makes way, for instance, for companies or brands to be the focal point within a community’s identity (Muniz and O’Guinn 2001). Heere et al., (2011) argue in their study that brand communities specialized in a particular subculture (i.e. Harley Davidson, Apple) enable an organization to communicate, establish, and foster rich consumer relationships, in order to positively influence consumer behaviour. In order to facilitate this, Cova and Pace (2006) have proposed consumption-oriented community-like platforms, in particular by looking at Nutella (well known chocolate sandwich spread), involving the consumer-brand relationship.

(9)

9

2. LITERATURE REVIEW

This chapter contains an overview of the relevant literature on brand loyalty written to date, investigating the relevant variables. The existing literature to date is not quite extensive on brand loyalty in an online setting. Therefore, in this study’s starting point is just brand loyalty. If possible brand loyalty in an online setting will be used, succeeding profound argumentation. Based on the literature review, the author has generated the different hypotheses associated with the model. These hypotheses focus on the relationships of the independent variables satisfaction, perceived value, involvement, to the mediating variable brand commitment, and indirectly to brand loyalty. Moreover, the relationship between brand commitment and brand loyalty is also subject to investigation. Furthermore, the moderating effects of identification, type of community and brand trust on the association of brand commitment with brand loyalty have been proposed.

2.1 BRAND LOYALTY

Since its appearance in the thirties of the twentieth century, over 200 definitions have been used for brand loyalty (Knox and Walker, 2001). An extensive literature review on just one topic provides enough evidence as to how important this topic is perceived in both the academic world as well as for professionals in practice. Since its appearance however, only agreement on the conceptual definition of brand loyalty seems to have emerged, without coming to agreement on a single unified measure (Knox and Walker, 2001). This study follows Mellens, Dekimpe and Steenkamp (1995) in that it is convenient to distinguish conceptual definitions (abstract definition of the phenomenon) from operational definitions (measurement method). Forming this bridge is necessary for assessing the validity of the construct by the measurement method. Without, a reference schema is absent subsequently disabling assessment of the correctness of the brand loyalty measure and the meaningfulness / meaninglessness of the results.

Considering the indecisiveness on brand loyalty in ‘Academia’ the definition of Jacoby and Chestnut (1978) will be used, because it covers the seemingly most important aspects of brand loyalty. In short, brand loyalty might be considered to be a complex concept consisting of both a psychological (attitude) and a behavioural construct (Knox and Walker, 2001; Day, 1969; Pritchard et al., 1999). The definition provides a clear picture on both single-brand loyalty versus loyalty towards a portfolio of brands, as well as spurious loyalty versus a longer term, deep-rooted psychological brand loyalty (Knox and Walker, 2003; Mellens, Dekimpe, and Steenkamp, 1996).

(10)

10

The second stream is more common in a low involvement context, where consumers purchase on a portfolio basis (Knox and Walker, 2001). According to Delgado-Ballester and Munuera-Alemán (1999) the literature concerning brand loyalty can be classified into two other separate streams, namely a psychological orientation and a sociological orientation. The first stream concerns primarily the cognitive processes supporting the development of brand attitude strength (Delgado-Ballester and Munuera-Alemán, 1999). The second stream is more interested in the hedonic motive aspects of brand loyalty (Delgado-Ballester and Munuera-Alemán, 1999). The latter orientation might be seen as the relational perspective in which brand loyalty is examined, considering relational characterizing variables such as for instance commitment and trust.

2.1.1 BRAND LOYALTY IN AN ONLINE SETTING

Loyalty may be a relevant variable to represent the degree of influence a brand has on consumers in an online community (Jang et al., 2008). Online community participation for instance may increase the brand value, perceived by community members. This happens since it may help the community members have a positive attitude towards the brand as well as being brand loyal (Jang et al, 2008). Jang et al. (2008) found that two community characteristics (interaction and reward for activities) significantly affected community commitment. Another moderating variable is identification. Carlson, Suter and Brown (2008) found that identification with a brand is weaker for respondents who were a member of a social brand community than for respondents who were not. This might also mean that identification with the brand and / or identification with the other users might moderate the influence on brand loyalty (Carlson, Suter and Brown, 2008). Brand trust is also considered as an influencer (Punniuamoorthy and Prasanna Mohan Raj, 2007).

2.1.2 SEQUENCE OF LITERATURE REVIEW

(11)

11 2.2 COMMITMENT

It is generally accepted that commitment leads to loyalty (Day, 1969; Morgan and Hunt, 1994). For instance, in Jacoby and Chestnut’s (1978) study commitment is considered the foundation for the development of brand loyalty. In their exploration of the link between brand community commitment and level of brand loyalty in online communities, Jang et al. (2008) found that brand loyalty is directly influenced by brand community commitment. According to Oliver (1999) a high level of commitment drives loyalty, which leads to product or service repurchase. From literature may be concluded that commitment plays one of the key (mediating) roles (Carlson, Suter, and Brown, 2008) in creating consumer loyalty (Garbarino and Johnson, 1999; Morgan and Hunt, 1994; Pritchard et al., 1999). Commitment is “an enduring desire to maintain a valued relationship” (Moorman et al., 1992) with the “tendency to resist change” (Pritchard et al., 1999). Lam et al. (2010) argue that brand commitment is the pledging or binding of a person to his or her brand choice within a product class.

Johnson, Hermann and Huber (2006) distinguish in relationship commitment two underlying different commitment types: 1) affective commitment and 2) calculative commitment, replenished by attitudinal commitment (Delgado-Ballester and Munuera-Alemán, 1999). Affective commitment, the “warmer” more emotional factor, relates to the degree to which a consumer identifies and is personally involved with a company (Garbarino and Johnson 1999; Morgan and Hunt 1994). Affective commitment has direct effect on actual behaviour, both in relationship maintenance (i.e. retention) and relationship development (Verhoef, 2003). Calculative commitment on the other hand is the more “colder” rational, economical factor. Calculative commitment depends for instance on product benefits, the lack of choice or the switching costs (Anderson and Weitz 1992; Dwyer, Schurr, and Oh 1987). In this study, affective commitment is considered only for the construct commitment. Although calculative commitment is considered and expected to exist, significant calculative commitment is not considered for the sake of simplicity of this study.

H1: the higher an individual’s commitment towards a brand, the higher an individual’s

loyalty towards a brand.

2.3 PERCEIVED VALUE

(12)

12

represents the utilitarian driver of the consumer–brand relationships. Punniyamoorthy and Prasanna Mohan Raj (2007) argue that perceived value is a broader construct that encompasses perceptions of quality given price and inputs versus outputs relative to the competition. In their broader construct Punniyamoorthy and Prasanna Mohan Raj (2007) focus on perceived value as a “consumer’s overall assessment” and that this assessment comprises of a product’s ‘get’ and ‘give’ component. According to Stryker (1968) it might be argued that if an individual’s perceived value of a certain brand is higher than for other brands one is more likely to resist change (Pritchard et al., 1999). In other words, if an individual perceived value of brand is higher it is likely that the individual will be more committed to that particular brand. Therefore, it is proposed that if the perceived value of a brand is relatively higher than that of the alternative(s) it has a positive influence on brand commitment. Thus,

H2a: The higher an individual’s perceived value of a brand, the higher the individual’s brand

commitment.

As one can note above, several latent dimensions seem to linger under the surface. Notably in the marketing context, perceived value is not limited to functional aspects, such as quality and price (Sheth et al., 1991). It may be apparent that perceived value is a broader and richer construct than a trade-off between just utility and price (Khan and Kadir, 2011). In accordance with Arnold and Reynolds (2003) and Sweeney and Soutar (2001), products might be evaluated as a two dimensional utilitarian and hedonic motive-structure. In line with Arnold and Reynolds (2003), Lindgreen and Wynstra (2005) proposed to focus on two main perspectives: 1) value of the products, and 2) value of the relationship. Punniyamoorthy and Prasanna Mohan Raj (2007) propose that a direct relationship exist between perceived value and brand loyalty. Furthermore, Sweeney and Soutar (2001) propose that perceived value is a construct of 1) functional value, 2) price-worthiness, 3) emotional value, and 4) social value. By using Lindgreen and Wyntstra (2005) view on perceived value, combined with the four dimensions proposed by Sweeney and Soutar (2001), the first two dimensions would fall into Lindgreen and Wynstra’s (2005) first stream ‘value of the product’, namely: 1) functional value, and 2) price-worthiness. The latter two dimensions would then fall into Lindgreen and Wynstra’s (2005) second stream ‘value of the relationship’, namely: 3) emotional value, and 4) social value. In this study it is proposed that perceived value is motivated by a utilitarian and a hedonic aspect, divided in the four dimensions price, quality, social and emotional value (Sweeny and Soutar, 2001).

(13)

13

perceived costs (Punniyamoorthy and Prasanna Mohan Raj, 2007). This leads to construction of the following hypotheses:

H2b: The higher the level of functional value, the higher the individual’s commitment towards

a certain brand.

H2c: The higher the level of price-worthiness, the higher the individual’s commitment towards

a certain brand.

Hedonic motive. According to Punniyamoorthy and Prasanna Mohan Raj (2007) emotional value refers to the utility derived from the feelings that a product generates. Without a doubt these emotions play a part in every purchase decision. Again in line with Punniyamoorthy and Prasanna Mohan Raj (2007), social value is considered in this study the utility derived from the product’s ability to enhance the social self-concept. In other words, by using the derived utility from the product the individual's perception of the "self", in relation to any characteristic, is enhanced. However, in line with the get and give component of perceived value, in this study social value is enhanced by Khan and Kadir’s (2011) relational value. Relational value is how consumers assess benefits and effectiveness of relationships (Khan and Kadir, 2011).

H2d: The higher the level of emotional value, the higher the individual’s commitment towards

a certain brand.

H2e: The higher the level of social value, the higher the individual’s commitment towards a

certain brand.

Brand commitment is the pledging or binding of a person to his or her brand choice within a product class (Lam et al., 2010). Research shows that perceived value, repurchase intention (Lam et al., 2010) and loyalty (Yang and Peterson, 2004) are positively related. From research (Johnson, Hermann and Huber, 2006) it may be believed that perceived value has a positive direct effect on loyalty intentions, although decreasing over time. This provides ground for the assumption that perceived value also has a positive direct influence on brand loyalty (Dodds, Monroe, and Grewal, 1991; Voss, Parasuraman, and Grewal, 1998; Parasuraman and Grewal, 2000; Reichheld, 1996). Therefore, it is proposed that if the perceived value of a brand is relatively higher than that of the alternative(s) it has a positive influence on brand loyalty. Thus,

H2f: The higher the level of perceived value, the higher the individual’s loyalty towards a

(14)

14 2.4 SATISFACTION

Satisfaction is defined as an evaluation of an (product) experience (Hunt, 1997), based on expectations and prior experience (Lam et al., 2010; Woisetschläger, Hartleb and Blut, 2008; Casaló, Flavián, and Guinalíu, 2008). In this sense one can see immediately that satisfaction occurs only if expectation are less than or met by the product or service. Consumers who are familiar with a brand use existing relationships with that brand as a reference point to evaluate (new) brands (Lam et al., 2010; Keller and Lehmann, 2006). Following Ashmore, Deaux and McLaughlin-Volpe (2004) evaluation refers to the positive or negative attitude that a person has towards [a brand], in which attitude (Eagly and Chaiken, 1993) is defined as a “psychological tendency that is expressed by evaluating a particular entity with some degree of (dis)favour” (p. 1). Consumer satisfaction has been considered a fundamental determinant of long-term consumer behaviour (Oliver, 1980). Literature (Woisetschläger, Hartleb and Blut, 2008; Casaló, Flavián, and Guinalíu, 2008) suggest that consumer satisfaction can be distinguished between 1) transaction-specific, and 2) overall satisfaction (Woisetschläger, Hartleb and Blut, 2008; Fornell, Johnson, Anderson, Cha and Bryant, 1996), consisting of both economic and non-economic satisfaction (Flavián, and Guinalíu, 2008). Overall satisfaction might be better suited considering it is a better predictor of past, current and future firm or brand’s performance (Woisetschläger, Hartleb and Blut, 2008). Fornell, Johnson, Anderson, Cha and Bryant (1996) have introduced the American Consumer Satisfaction Index (ACSI), developed to represent a consumer-based measurement system. In its core essence, the ACSI measures the perceived quality of goods by the consumer taking actual as well as the anticipated total purchase and consumption experience into account. Westbrook and Oliver (1991) argue that consumer satisfaction mediates the concept of learning. By learning from prior experience, consumers are able to alter post-purchase behaviour (i.e. word of mouth, repost-purchase intention). Wang et al. (2001) provide (partial) evidence for Westbrook and Oliver’s (1991) consumer satisfaction concept, as they (Wang et al., 2001) find that satisfaction has significant influence on repurchase intention. From Oliver (1991) and Wang et al. (2001) it may be assumed that if an individual’s satisfaction with the brand increases, an individual’s commitment increases simultaneously. In other words, if an individual satisfaction of a brand increases it is likely that the individual will be more committed to that particular brand. Therefore, it is proposed that if satisfaction with a brand increases it has a positive influence on brand commitment. Thus,

H3a: The higher the level of consumer satisfaction, the higher the individual’s commitment

towards a certain brand.

(15)

15

threshold before influencing loyalty at all, 2) is positive, and 3) has a diminishing effect on increasing loyalty. The quality of relationships between consumers and companies is an important variable for consumer loyalty (Davis-Sramek, 2009). Yang and Peterson (2004) have found evidence that satisfaction has a direct positive effect on brand loyalty. Thus,

H3b: The higher the level of consumer satisfaction, the higher the individual’s loyalty towards

a certain brand.

2.5 PRODUCT INVOLVEMENT

The assumption that the consumer is an intelligent, fully rational thinking, problem-solving and evaluating organism has been outdated for some time now in terms of consumer behaviour (Zaichkowsky, 1985). Instead, consumers behave in terms of a two-fold dichotomy; low-involvement and high involvement behaviour (Zaichkowsky, 1985). The author (Zaichkowsky, 1985) make distinctions in the involvement construct, namely: 1) advertisement involvement – leads to give more counterarguments to the advertisement, 2) product involvement – to lead to greater perception of attribute differences, perception of greater product importance, and greater commitment to brand choice, and 3) purchase involvement - leads one to search for more information and spend more time searching for the right selection, in which all different objects lead to a different response. Moreover, Eagly and Chaiken (1993) argue that attitude might be assessed by behavioural as well as cognitive and affective indicators, making [behavioural] involvement a key element of […] identity (Ashmore, Deaux and McLaughlin-Volpe, 2004). Behavioural involvement is defined as the degree to which the person engages in actions that directly implicate the [collective] identity. According to the literature, involvement is an ongoing unobservable state of motivation, arousal, interest (Punniyamoorthy and Prasanna Mohan Raj, 2007), thoughts, feelings and behavioural response to a product [category] (Gordon, McKeage, and Fox, 1998). This is supported by the findings of Park (1996) and Kim, Scott and Crompton (1997). In their study (Park, 1996; Kim, Scott and Crompton, 1997) the authors have found evidence that involvement is closely related to intentions and behaviours. Knox and Walker (2003) argue that brand involvement is driven by product involvement (and other hedonic formed antecedents on utility) and hedonic formed antecedents on risk. However, considering the assumed t high correlation between brand involvement and brand commitment exist, disabling two fully distinct constructs, brand involvement is considered to be part of commitment. Therefore, brand involvement is not considered as a distinct independent variable for the remainder of this study.

(16)

16

Zaichowsky (1985) and LeClerc and Little (1997), one might argue that if a consumer is highly involved in a product it is likely that the individual will be more committed to the same brand. The underlying relationship might be for example that if a certain individual is fully aware of all the benefits of the products within the category, the individual will be more likely to be committed to that brand which has the most benefits. Thus,

H4a: The higher the level of an individual’s involvement in a product, the higher the individual’s commitment towards a certain brand.

During Park’s (1996) study on leisure activities, Park found that involvement and attitudinal loyalty are highly correlated. Iwasaki and Havits (1998) however have argued that the found correlation between involvement and attitudinal loyalty did not determine that loyalty is formed by involvement. Instead they argued that it is a sequence of different psychological processes. This however has been untested. Ample studies (e.g. Jacoby and Chestnut, 1978; LeClerc. and Little, 1997; Quester and Lim, 2003; Iwasaki and Havits, 1998) have investigated the relationship between product involvement, commitment and loyalty. For instance, LeClerc and Little (1997) argue that brand loyalty interacts with product involvement. Thus,

H4b: The higher the level of an individual’s involvement in a product, the higher the individual’s loyalty towards a certain brand.

2.6 BRAND TRUST

Brand trust may be considered as an intangible part of a brand. Brand intangibles are aspects of the brand image that do not involve physical, tangible, or concrete attributes or benefits (Keller and Lehmann, 2006). It is for instance common practice that these intangibles are used by marketers to differentiate a brand from the competition (Kotler and Keller, 2006). Fournier (1998) argues that brand-relationship quality is multifaceted and consists of six dimensions, for instance the self-concept connection, commitment or nostalgic attachment, and the brand-partner quality.

(17)

17

offline environment the quality of the product prior to purchase, whereas in an online environment physical cues are unavailable (Alba et al., 1997; Moore and Andradi, 1996).

Trust in a person is a feeling of security based on the belief that his/her behaviour is guided and motivated by favourable and positive intentions towards the welfare and interest of his/her partner (Delgado-Ballester and Munuera-Alemán, 1999). To be more specific in regard to this study, brand trust is the willingness of a consumer to rely on the ability of the brand to perform its stated function brand trust (Lam et al., 2010). Delgado-Ballester and Munuera-Alemán (1999) argue that brand trust is a feeling of security perceived by the consumer that his/her expectations are met, built on brand reliability and brand intentions towards the individual. Garbarino and Johnson (1999) have examined the relationship between loyalty and brand trust extensively. Developing positive and favourable attitudes through trust is very important (Garbarino and Johnson, 1999). In line with Garbarino and Johnson (1999) in this study, brand trust is considered the central construct for any long-term relationship. Following Punniyamoorthy and Prasanna Mohan Raj (2007), brand trust may be an important contributor to the kind of emotional commitment that leads to long-term loyalty. This is supported moreover by Morgan and Hunt (1994). According to Morgan and Hunt (1994) trust may be related to commitment and loyalty, because trust is required in order to maintain a sustainable relationship. Concluding, trust is considered an important variable, because trust helps developing positive and favourable attitudes, resulting in commitment to a certain brand (Delgado-Ballester and Munuera-Alemán, 1999). In this study, it is expect that the lesser the doubt an individual has towards the brand’s purposes are questionable, the more likely it is that an individual’s brand commitment influence on brand loyalty will be higher (Delgado-Ballester and Munuera-Alemán, 1999). In short, the higher the feeling of brand trust the more the consumer is committed to it, enabling a higher level of brand loyalty. Thus,

H5: Whereas an increase of brand trust will have a positive impact on the relationship between brand commitment and brand loyalty, a decrease in brand trust will have a negative impact on the same depicted relationship.

2.7 BRAND COMMUNITY

(18)

18

An online brand community, however, is different from any other online community classification (Hagel and Armstrong, 1997; Henri and Pudelko, 2003), considering its unique characteristics. Online brand communities are 1) built around products or services (Hong and Kim, 2002), 2) relatively stable in size and continuity, 3) its members have a strong commitment with common themes and goals (Cova, 1997; Maffesoli, 1996), 4) the community acts as a place for social negotiation where the community reflects mainstream culture and encourages the members’ voluntary interpretation of the brand (Holt, 1997), and 5) community members have a high level of identity and high understanding of the commercial landscape (Muniz and O’Guinn, 2001). Heere, Walker, Yoshida, Jae Ko, Jordan and James (2011) base their definition of brand community on Ashmore, Deaux, and McLaughlin-Volpe (2004) and (Tajfel 1978), in which they argue that brand communities are social groups that share common features with more general groups. A brand community may also be defined as “a specialized, non–geographically bound community, based upon social relationships among admirers of a brand in cyberspace’” (Moorman et al., 1992), in which community members have a shared consciousness, rituals and traditions, and a sense a moral responsibility (Muniz and O’Guinn, 2001), centred around a branded good or service. Muniz and O'Guinn (2001, p. 412) define a brand community as “a specialized, non-geographically bound community, based on a structured set of social relations among admirers of a brand”. For the purpose of this study, online brand communities consist of individuals who possibly never interact face-to-face with other members, still acknowledge membership in the community and engage in social interaction with other members. Therefore, an online brand community is considered to be a social community of brand admirers who acknowledge membership in the community and engage in social relations (Carlson, Suter, and Brown, 2008).

(19)

19

H6: Whereas a FIC type of online brand community will have no impact on the relationship between brand commitment and brand loyalty, a CIC will have a positive impact on the same depicted relationship.

2.8 BRAND COMMUNITY IDENTIFICATION

Now, considering that the topic of discussion has touched upon a brand-consumer relationship, a closer look will be given at which background for this literature review might be best suited. The brand-consumer relationship evolves in the foremost place around ‘identification’ of the brand by the consumer (i.e. Tajfel, 1979). Identification is a person's conception and expression of their individuality or group affiliation (Kuenzel and Halliday, 2010). Identification is ambiguous, namely 1) identifying oneself (through a brand) with a certain group (social identification - Kuenzel and Halliday, 2010) and, 2) an individual’s desire to belong to a particular community (community identification - Heere et al. 2011). Considering the mentioned brand-consumer construct above, social identity theory seems to be useful for understanding consumer behaviour, and to investigate identification with brands, communities and the underlying relationship to brand loyalty (Ashmore, Deaux, and McLaughlin-Volpe 2004; Muniz and O’Guinn 2001; Tajfel 1978).

(20)

20

Much has been written about how consumers perceive brands and about positive outcomes in terms of brand loyalty (i.e. Kuenzel and Halliday, 2010). Literature suggests that the stronger the identification with a brand community is, the more likely it is that the individual will act upon that identification, resulting for instance in higher willingness to pay, higher brand loyalty, increase in purchases, and increases in supportiveness and recommendations (Ashmore et al., 2004; Kuenzel and Halliday, 2010). Carlson, Suter, and Brown (2008) propose that the degree to which individuals identify with 1) the brand and 2) the group who utilize the brand has significant influence on how the community functions. In line with Bergami and Bagozzi (2000) and Carlson, Suter and Brown (2008), this study follows the concept of affective- and cognitive identification.

An important study by Algesheimer et al. (2005) studied social interactions among brand community members. They found that an individuals’ identification with a brand community has a strong impact on the individual’s level of community engagement (Matzler, Pichler, Füller and Mooradian, 2011). Research suggests that commitment plays a critical role in relationship marketing (e.g. Morgan and Hunt, 1994). In this central role of an individual's commitment to a particular brand, identification plays an important role. From research it may be concluded that identification may have a positive effect on (brand) loyalty (Bergami and Bagozzi, 2000; Bhattacharya and Sen, 2003; Kuenzel and Halliday, 2010). For instance, Algesheimer et al. (2005) found that brand community identification has a significant positive impact on an individual's level of community engagement. In this study it is predicted that the relationship between brand commitment and brand loyalty will be influenced by identification: the greater one identifies with a brand, the stronger the influence of one’s brand commitment is on brand loyalty. Carlson, Suter, and Brown (2008) argue that the brand and group identification might have different levels of influence the psychological sense of brand community and identification. In this study however it is not the psychological sense of brand community, but instead that brand commitment is affected by the moderating influence of identification. Thus,

(21)

21

3. RESEARCH DESIGN

The current chapter will elaborate on the research design of this master thesis and its subsequent choices and considerations. In the first part of this chapter an elaboration on the methodology used for analysis is given; how the author proposes to answer the different hypotheses associated with the model (figure 1). The development of measures and the development of the questionnaire are discussed next. Finally, data collection and plan of analysis are discussed.

3.1 METHODOLOGY

A short recapitalization. This study will contribute by investigating the factors that drive online brand loyalty in an online community perspective. During the literature review, it is proposed that 1) perceived value, consumer satisfaction, and product involvement (§2.3 – §2.5) affect the level of an individual’s brand commitment, 2) although brand commitment (§2.2) is the important mediating variable driving online brand loyalty; perceived value, consumer satisfaction and product involvement also have a direct effect (§2.3 – §2.5) on online brand loyalty, and 3) that the relationship brand commitment-online brand loyalty is moderated by brand trust, type of community and brand community identification (§2.6 – §2.8). The conceptual framework guiding this study is presented in Figure 1. Note, the gray frames in the conceptual model are only provided for clarity reasons, and have no other purpose.

(22)

22 3.1.1 RESEARCH METHOD

In order to test the hypotheses proposed in the literature review and in the research model, a type of analysis is required where both the mediating construct as well as the influence of the moderating variables are included. This will be discussed further in detail during the plan of analysis (§3.4) section. Although further discussed in the plan of analysis (§3.4), the multivariate data analysis technique multiple linear regression model is considered most suitable (Malhotra, 2007). In order to generate variables highly trustworthy, per variable multiple questions are used in the questionnaire (Punniyamoorthy and Prasanna Mohan Raj, 2007). The use of this method will help to increase internal consistency and validity. Factor analysis by principal components analysis with VARIMAX rotation (Punniyamoorthy and Prasanna Mohan Raj, 2007) is used, in order to abstract latent dimensions.

3.1.2 DEVELOPMENT OF MEASURES

In order to ensure the extent to which a scale measures a variable of interest, content validity deserves attention. The items selected need to be able to represent the concept from which generalisations are to be made. Without exhaustively forcing the author to focus on determination of the right scales, the selected items are mainly adapted from prior studies to ensure content validity, as most have been previously tested for internal consistency and validity (Heere et al., 2011). Therefore, no additional testing of specific items was considered necessary. However, considering the context of this study versus other existing research, some items have been slightly altered in order to allow for the use of the respectable items. In this tendency, conducted alterations are done in order to improve the items for the specific context, i.e. naming the particular brand. Therefore, pre-testing of the questionnaire was deemed necessary. The questions in the questionnaire are measured accordingly to a seven-point Likert scale.

BRAND LOYALTY

(23)

23

Both attitudinal and behavioural measures (Oliver, 1999; Zeithaml, 2000) might be used to define brand loyalty. An individual’s desire to continue a relationship with a company defines brand loyalty from an attitudinal perspective (Yang and Peterson, 2004). The proportion of times a purchaser chooses the same product or service in a specific category compared to the total number of purchases made by the purchaser in that category defines brand loyalty from a behavioural perspective (Yang and Peterson, 2004; Knox and Walker, 2001). However, this is purely a behavioural construct, meaning that it also takes various different version of spurious loyalty into account, such as brand switching due to price cuts or out of stocks (Day, 1969; Pritchard et al., 1999). This study thus supports other researchers (i.e. Day, 1969; Jacoby and Chestnut, 1978; Bloemer and Kasper, 1994; Knox and Walker, 2001) that and commitment and attitudinal loyalty are essential elements for and of brand loyalty, as it allows to separate true brand loyalty from repeat buying.

Mellens, Dekrimpe and Steenkamp (1995) have discussed the advantages and disadvantages of both attitudinal and behavioural behaviour measures. Attitudinal measures have the advantages that 1) repeat buying is separated from brand loyalty, and 2) wherewith it is easier to pick the right decision unit. However, 1) valid representation of reality cannot be guaranteed, 2) might be incidental, and 3) is harder to collect. Behavioural measures are 1) based on actual behaviour, 2) non-incidental and 3) easy to collect. However, the disadvantages are 1) repeat buying is not distinguished from brand loyalty, and 2) difficult to pick the right decision unit. Separately taken, it can be argued that both miss the field which the other covers. Therefore, Oliver (1999) has proposed a brand loyalty definition combining both aspects. Brand loyalty is “a deeply held commitment to re-buy or re-patronise a preferred product / service consistently in the future, thereby causing repetitive same-brand or same brand-set purchasing, despite situational influences and marketing efforts are having the potential to cause switching behaviour” (e.g. Lam et al., 2010; Oliver, 1999; American Marketing Association, 2011; Aaker, 1991; Kotler et al., 1989).

(24)

24

were deemed unsuitable considering the low degree of purchase frequency of the product at hand. Moreover, for dividing behavioural by attitudinal loyalty in order to derive the loyalty index, for which the measured data needed to be transformed from an ordinal fashion to a ratio scale, a relatively big leap.

Instead, Bloemer and Kasper (1995) are followed. They also argue that attitudinal loyalty together with behavioural loyalty determines the degree of true loyalty. However, Bloemer and Kasper (1995) determine loyalty by the use of an index, but determine the loyalty by using commitment and the chance of buying the same brand next time. This operationalization of brand loyalty (multiplication of the attitudinal loyalty score with the score for repeat purchasing behaviour) is not threatened by the same disadvantages as the earlier indices (Day, 1996; Pritchard et al., 1999).

COMMITMENT

In general, commitment might be considered as ‘someone’s intention to continue a relationship’. There are however different kinds of commitment, namely: attitudinal (Delgado-Ballester and Munuera-Alemán, 1999), affective and calculative (Johnson, Hermann and Huber, 2006). Each different type has a different underlying relationship to an equal other variable, i.e. brand loyalty (Delgado-Ballester and Munuera-Alemán, 1999). In this study only affective commitment is considered. Affective commitment relates to identification and involvement with a brand and/or company (Garbarino and Johnson 1999; Morgan and Hunt 1994), with a direct effect on relationship maintenance and relationship development (Verhoef, 2003). It might be considered as a positive regard for, and attachment to the brand (Delgado-Ballester and Munuera-Alemán, 1999). Prior to investigating the mediating construct measurement of the commitment construct is needed. Fortunately, a valid and reliable measure of commitment has already been developed (Pritchard et al., 1999). Following Punniuamoorthy & Prasanna Mohan Raj (2007) and Pritchard et al. (1999), commitment is best measured in the context of brand loyalty by focusing on the resistance to change scales (Pritchard et al., 1999). There has been chosen to consider four items that represent each a part of the relationship between the consumer and the brand (Punniuamoorthy & Prasanna Mohan Raj, 2007).

PERCEIVED VALUE

(25)

25

purchase attitude and behaviour. Considering the length of the questionnaire Punniuamoorthy & Prasanna Mohan Raj (2007) are followed for the item development of emotional, social and price/worthiness values. For functional value Sweeney and Noutar’s (2001) items are considered to be more appropriate for use in the questionnaire.

SATISFACTION

In order to measure overall satisfaction a multi-item approach is considered (Delgado-Ballester and Munuera-Alemán, 1999). Fornell, Johnson, Anderson, Cha and Bryant (1996) have proposed to use overall satisfaction. According to Spreng, MacKenzie and Olshavsky (1996) overall satisfaction is a summary evaluation of the entire product use experience, consisting of two dimensions: valence (positive and negative) and intensity. The American Consumer Satisfaction Index (ACSI) measures the consumer’s overall evaluation, taking both the actual as well as the anticipated total purchase and consumption experience into account (Fornell, Johnson, Anderson, Cha and Bryant, 1996), based on six latent variables. The ACSI however focuses on aggregated sets of data, such as industry, sector and firm financial results. Considering the limited space in the questionnaire and the level of data, this study follows Helgesen (2006) and Oliver (1999) instead. Satisfaction is measured by using two single variables (Oliver, 1999), capturing 1) consumer’s expressing of fulfilment, and 2) a variable standard for comparison, divided over four items. Consumer’s expressing of fulfilment is measured with two bipolar adjectives "very satisfied/very dissatisfied" and "very pleased/very displeased". The standard for comparison is measured by the two bipolar adjectives "contented/frustrated," and "delighted/terrible".

PRODUCT INVOLVEMENT

(26)

26

BRAND TRUST

The research of Delgado-Ballester and Alemán (1999) and Delgado-Ballester, Munuera-Alemán, and Yagüe-Guillén (2003) have done quite some work in translating trust in general, to brand trust. By following the same researchers (Delgado-Ballester and Munuera-Alemán, 1999) brand trust is considered to be a multi-item type scale, consisting of two different concepts, namely: intentions towards the consumer (i.e. “keep promises”) and reliability. Furthermore, in their study (Delgado-Ballester and Munuera-Alemán, 1999) a six item are employed to represent brand trust, related to reliability of the brand and its intentions towards consumers. The authors focus in general on the feeling of security. Delgado-Ballester, Munuera-Alemán, and Yagüe-Guillén (2003) have developed and validated the brand trust scale further. In their study (Delgado-Ballester, Munuera-Alemán, and Yagüe-Guillén, 2003) the focus is more on reliability and intentions from brand X towards the consumer. From these two dimensions of brand trust eight items were developed (Delgado-Ballester and Munuera-Alemán, 1999).

BRAND COMMUNITY IDENTIFICATION

(27)

Table 2 : SCALE FOR MEASURING LOYALTY

Brand loyalty

A1 I consider myself to be a loyal patron of brand X A4 To me, brand X is the same as other brands

A2 If I had to do it over again, I would use another brand RPB1): Rate the chance of buying brand X the next time you need to buy such a product (percentage ranging

from 0% to 100% (RPB). A3 I try to use brand X because it is the best choice for me.

Commitment

CO1 I have strong preference for brand X CO3 For me, to change my preference from brand X would require major rethinking

CO2 My preference for brand X would not willingly change CO4 Even if close friends recommend another brand, I would not change my preference for brand X Perceived value

F1 Brand X has consistent quality PR3 Brand X is economical

F2 Brand X is well made EM1 Brand X makes me feel good

F3 Brand X has an acceptable standard for quality EM2 Brand X gives me pleasure

F4 Brand X has poor workmanship EM3 Wherever I see brand X, I would use it

F5 Brand X would not last a long time SO1 I will consider brand X as a status symbol

F6 Brand X would perform consistently SO2 I feel proud of being a user of brand X

PR1 Brand X offers value for money SO3 Using brand X will improve the way I am perceived

PR2 Brand X is reasonably priced Satisfaction

S1): Which brand did you bought last time? (open-ended) SF2) Brand X makes me feel..."very displeased/very pleased"

S2): Are you satisfied or dissatisfied with that brand? (Satisfied / Dissatisfied) SC3) Imagine a top brand in the market. Versus a top brand in the market, brand X makes me

feel..."frustrated/contented" S3): How much (dis)satisfied are you in terms of percentage varying from 0% to 100%?

SF1) Brand X makes me feel..."very dissatisfied/very satisfied" SC4) Imagine a top brand in the market. Versus a top brand in the market, brand X makes me feel..."

terrible/delighted"

Product involvement

PIE1): To me, brand X is… “unimportant / important” PIT4): To me, brand X is… “unexciting / exciting”

PIE2): Brand X… “means nothing to me / means a lot to me” PIT5): To me, brand X is… “dull / neat”

PIE3): To me, brand X is… “irrelevant / relevant”. PIT6): To me, brand X is… “boring / interesting”. Brand trust

BT1): Brand X is a brand name that meets my expectations BT5): Brand X’s brand name would be honest and sincere in addressing my concerns

BT2): I have confidence in X’s brand name BT6): I could rely on X’s brand name to solve an occuring problem

BT3): Brand X is a brand name that never disappoints me BT7): Brand name X would make any effort to satisfy me

BT4): Brand X’s brand name guarantees satisfaction BT8): Brand name X would compensate me in some way for the problem with the [product] Brand community identification

IwC1): I am very attached to the community IwC4): How many friends do you have in the brand X’s community? (open ended)

IwC2): I see myself as a part of this brand community IwC5): The friendships that I have with other brand community members mean a lot to me

IwC3): Other brand community members and I share the same objectives IwC6): If brand community members planned something (i.e. attend the launch of brand X’s latest product

(28)

Heere et al. (2011) proposes to use 18 different items for measuring brand community identification, focusing on the underlying dimensions: self-categorization, private and public evaluation, interconnection of the self with the group, sense of interdependence with the group, behavioural involvement, and content and meaning. Considering the already acquired lengthy questionnaire, the ease and comfort for the respondent and the reliability of the items, this study follows Matzler, Pichler, Füller, and Mooradian (2011) in adopting existing scales (Algesheimer, Dholakia and Herrmann, 2005). The completeness and the usability of the adopted items and corresponding scales, did not make it necessary to include items from other research.

COMMUNITY

Online brand communities may be grouped into two major types (Jang et al., 2008): (1) voluntarily initiated and operated by consumers (CIC) and (2) firm initiated and operated community with a firm’s ownership in the brand (FIC). The type of community influences indirectly the consumer’s engagement intentions (Lee et al., 2011) and the community members' attitude toward the community (Muniz and O’Guinn, 2001). For instance, Jang et al. (2008) show in their study that in firm-initiated communities, commitment was just partially affected by the defined antecedents. Following Jang et al. (2008) the same typology is used: the type of community as moderating variable is represented by 1) CIC – an individual (i.e., consumer), or 2) FIC – a company with ownership in the brand.

3.1.3 MODEL

In order to measure loyalty, Bloemer and Kasper (1994) propose to use an index of the loyalty construct. The advantages are 1) it allows investigation of loyalty from a causal perspective, 2) underlying processes can be addressed (Pritchard et al., 1999), and 3) that the measure both takes behavioural as attitudinal perspective of loyalty into account (Oliver, 1999; Zeithaml, 2000; Bloemer and Kasper, 1994). The loyalty index (Bloemer and Kasper, 1994) is calculated by multiplying the change of repurchase ( RPB1) by the mean attitude scores [Ai]. Attitude is scaled so that a high value represents a favourable attitude. Bloemer and Kasper (1994) argue that by multiplying a high change of repurchases by a strong attitude creates index scores close to 28, indicating true loyalty. Multiplying a low change of repurchase by a weak attitude creates scores, however, closer to 4, indicating low loyalty. For instance, if one’s attitude towards the brand is negative (i.e. on a 1 to 7 Likert-scale: 4), but the change of repurchase of the product devoted to brand m is high (i.e.: .75%), the loyalty index scores low (i.e. 3). From this true loyalty is close to 28, whereas no brand loyalty is close to 0. By this simple example, it is easily seen that the loyalty index takes both measures into account, creating a naturally looking and comfortable scale. The loyalty equation, shown below, is used as the basis for this study’s examination on brand loyalty. Brand loyalty (BL) is the function of loyalty behaviour (LB) and loyalty attitude (LA):

∗

(29)

- 29 -

BLmi = the brand loyalty score for the buyer i of brand m,

RPBmi = respondent i’s chance of repurchasing the product of brand m over a set period of time,

LAmi = the mean attitude score of respondent i towards brand m.

Hypotheses involving multiplicative interaction or moderation effects are common in science (Preacher, Curran and Bauer, 2006). relationships between product involvement, satisfaction and perceived value with commitment and brand loyalty are assumed to be linear. This leads to the following proposed model:

α0 α1 α2DE α α α α∗

α ∗ α ∗

(2.1) COM β% β&PI β) OS βPV (2.2)

, -% -&PI -) OS -PV (2.3)

where

COMmi = brand m commitment of respondent i,

PImi = involvement in the product of respondent i towards brand m, OSmi = overall satisfaction of respondent i with brand m,

PVmi = respondent i perceived value of brand m, BTmi = trust of respondent i in brand m,

IDmi = identification of respondent i with brand m community, TCm = type of brand community m,

DEmi = direct effects of independent variables, excluding brand m commitment of respondent i.

3.2 DEVELOPMENT OF QUESTIONNAIRE

(30)

- 30 -

entirely on online communities and its related activities, and that 2) the modus of operandi is communication in the community.

TABLE 3

OPERATIONAL DEFINITIONS AND MEASURES OF THE VARIABLES

Name of variable Operational definition Measurement Source Items

Online brand loyalty

Consistent repurchase of a brand, due to positive affection towards the brand

- Attitudinal loyalty - Behavioural loyalty

- Day, 1969;

- Pritchard et al., 1999; - Bloemer and Kasper, 1994

8

Online brand commitment

Someone’s intention to continue a relationship The resistance to change scales - Delgado-Ballester et al., 1999; - Pritchard et al., 1999; - Punniuamoorthy et al., 2007; 4 Perceived value

Functional value The utility consisting of the product quality and the expected performance of the product

- Pritchard et al., 1999 - Sweeney and Soutar, 2001 - Lindgreen and Wyntstra, 2005 - Punniyamoorthy et al., 2007 - Khan and Kadir, 2011

15 Price-worthiness Consumer’s rational and economic valuation of the

product

Emotional value The utility derived from the feelings that a product generates

Social value Value consisting of the utility derived from the product’s ability to enhance the social self-concept

Satisfaction

Psychological tendency that is expressed by evaluating a particular entity with degree of (dis)favour - Consumer’s expressing of fulfilment, - A variable standard for comparison - Delgado-Ballester et al., 1999; - Fornell et al., 1996; - Oliver, 1999; - Helgesen, 2006;

- Eagly and Chaiken, 1993;

4

Product involvement

An ongoing unobservable state of motivation, arousal, interest thoughts, feelings and

behavioural response to a product

- Importance - Interest

- McQuarrie and Munson, 1992; - Quester and Lim, 2003; - Delgado-Ballester et al., 1999; - Punniyamoorthy et al., 2007; - Gordon et al., 1998;

6

Brand trust

Feeling of security, based on the belief that individual’s behaviour is motivated by positive expectations - Brand reliability - Brand intentions - Delgado-Ballester and Munuera-Alemán, 1999; - Delgado-Ballaster et al., 2003; 8 Brand community identification

A function of the overlap between an individual's self-schema and the schema the for another object

- Self-categorization, - Private evaluation, - Public evaluation, - Interconnection - Interdependence - Involvement, - Content and meaning

- Bergami and Bagozzi, 2000; - Carlson, et al., 2008; - Matzler et al., 2011; - Algesheimer et al., 2005 6 Type of community

Consumer Voluntarily initiated and operated by consumers

- Binary

- Lee et al., 2011; - Jang et al., 2008;

- Muniz and O’Guinn, 2001;

1 Company Firm initiated and operated

(31)

- 31 -

The survey instrument used is hosted by an online commercial web surveying company, namely thesis-tools.nl. In order to ensure good data and decrease the invalid rate of questions answered during the analysis phase of this study, throughout the survey rating scales are used consistent in terms of meaning (i.e. “1= strongly disagree”) and order (i.e. “strongly disagree” is always provided as initial answer). The majority of questions are asked with a neutral tone of voice. On occasion using however a negative or positive formulated question, helps to determine the quality of individual questionnaires. The questionnaire comprises of items related to brand loyalty (attitude and behavioural loyalty), commitment , perceived value (functional value, price/worthiness, emotional, and social value), satisfaction (fulfilment and comparison), involvement (importance and interest ), brand trust (reliability, intentions and control items), type of community, and identification. The first part of the questionnaire consists of variables directly related to the dependent, independent and moderating variables. Last but not least, variables such as age, time period of activity in community and gender (Jang et al., 2008) are asked at the end of the questionnaire.

3.3 DATA COLLECTION

(32)

- 32 - 3.4 PLAN OF ANALYSIS

First, from a theoretical- and practical perspective the rather large amount of different items are grouped together by the use of factor analysis. Second, various simple linear regression and correlation analysis are conducted to identify the relationship between the individual factors. The analysis is conducted in two parts, the first part investigated mediation only, where the second part also includes the moderation effects.

Simple mediation

In order to determine mediation, the four steps proposed by Baron and Kenny (1986) and MacKinnon and Dwyer (1993) were followed. The first three steps are required to investigate whether there is any correlation between the respective variables. The first step is to show that the different number of independent variables is correlated with the dependent variable. By looking at the relationship independent variable (X) on dependent variable (Y), the total effect (c) is determined, depicted in model 1: Y = b1 + c X + e1. The second step is to show that the independent variable is correlated with the mediator (M), depicted in model 2: M = b2 + a X + e2.

FIGURE 2

MEDIATION MODEL EXPLAINED

This estimation (a) is the first part of the indirect effect. The third step is to show that the mediator affects the dependent variable, also recorded as the second part of the indirect effect (b), shown in model 3: Y = b3 + b M + e3. If the previous steps were successful, the last step is to establish the degree of mediation, depicted in model 4: Y = b4 + c' X + b M + e4.

To determine mediation, the effect of the independent variables on the dependent variable, controlling for the mediator (c'), should relatively be lower than the direct ‘independent variables dependent variable’-relationship. For full mediation, the initial independent variables dependent variable relationship should be zero. To determine the mediated effect (also known as the indirect effect) one can choose from two methods. According to Preacher, Rucker & Hayes (2007), the first method (Sobel, 1982) focuses on the product of the two parameters (a * b). The second method looks at the

Independent variable (X) Dependent variable (Y) Mediator (M) c’ a (sa) b (sb) c c' ab

(33)

- 33 -

difference between total and direct effect (c - c') (Judd and Kenny, 1981). The two different methods always present identical outcomes (MacKinnon, Warsi, & Dwyer, 1995). For mediation holds, c – c’ = ab, where both c-c’ and ab ≥ 0.

Moderated Mediation

Firstly, a brief and simple review of moderation will be given. In short, moderation occurs when the strength of a relationship between two undefined variables depend on a third variable (Preacher, Rucker & Hayes, 2007). In moderation, the moderator interacts with the independent variables in predicting the dependent variable. Moderated mediation occurs if the strength of an indirect effect depends on the level of some variable, contingent on the level of the moderator. Moderation is often assessed by using the regression equation:

. /% /& 0 /) / / 0 / 1

Moderated mediation models attempt to explain both how and when a given effect occurs (Preacher, Rucker & Hayes, 2007). This effect is both known as moderated mediation (Baron and Kenny, 1986) as conditional indirect effect (Preacher, Rucker and Hayes, 2007).

FIGURE 3

MEDIATION MODEL EXPLAINED

The expected patterns of mediation varying as a function of the moderator are investigated using linear regression analysis. The steps by Baron and Kenny (1986) are repeated with the independent variable being an interaction with the respective number of moderators, by the use of Preacher, Rucker & Hayes’s (2007) Macro SPSS syntax PROCESS.

Significance testing

The Sobel test (1982) has been by far the most commonly reported in determining significance of the coefficients (MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002). The Sobel test (1982) requires the coefficients a and b, the standard error (SE) of a (sa) and the SE of b (sb) or the t-test

Independent variable (X) Dependent variable (Y) Mediator (M) c’ a (sa) _{b (sb)} c c' ab

Total effect = direct effect + indirect effect

(34)

- 34 -

statistic of both coefficients. In example, sa equals dividing the a coefficient by the t test of coefficient a (ta). The SE of the indirect effect provided by the Sobel test is equal to,

23)₄₅)_/)₄₆)_.

Testing the indirect effect on significance is easy. By treating the ratio (Sobel, 1982), /3

23)₄₅)_/)₄₆)

as a Z test, significance can easily be found of a and b. In example, if the Sobel test statistic is smaller than -1.96 or larger than 1.96 the indirect effect is significant at the .05 level. According to MacKinnon, Warsi, & Dwyer (1995) however, the Sobel test is rather low in power. The test’s low power is due to the positive skewness of the ab sampling distribution. This would not be a problem, if the Sobel test would not presume a symmetric distribution. The test does however, falsely presumes symmetry (MacKinnon, Warsi, & Dwyer, 1995).

In order to gain good answers nevertheless, another test is used, namely bootstrapping (Bollen & Stine, 1990; Shrout & Bolger, 2002). Bootstrapping for standard errors has greater power in small samples (Bollen & Stine, 1990; Shrout & Bolger, 2002). Bootstrapping is a non-parametric method based on sampling with replacement, done many times e.g., 5,000 times (Kenny, 2011). This method for testing the indirect effect, computes the indirect effect and empirically generates a sampling distribution. The advantage of bootstrapping (Kenny, 2011) is that with the distribution a confidence interval, a p-value, and a standard error can be determined. The confidence interval is computed in order to check if zero is in the interval. Without zero being in the interval, one can be fairly sure that the indirect effect is different from zero. Also, determining a p-value is no problem. Standard errors are a different story however. At bootstrapping SE’s suffer the same problem as the Sobel standard errors (Kenny, 2011). Inferences for indirect effects are based on percentile bootstrap confidence intervals (95%). The bootstrap method is included in the analytic tool PROCESS provided by Hayes (2012).

Brand Loyalty of Mobile Operating Systems Factors defining brand loyalty in an online community setting by J. Veldmeijer March, 2012

Brand Loyalty of Mobile Operating Systems

Brand Loyalty of Mobile Operating Systems

PREFACE

MANAGEMENT SUMMARY

TABLE OF CONTENTS

1. INTRODUCTION

2. LITERATURE REVIEW

3. RESEARCH DESIGN