Brand Extensions Challenge or Support Attitudinal Loyalty

(1)

Brand Extensions Challenge or Support

Attitudinal Loyalty

Master thesis, Msc Business Administration, Marketing

University of Groningen

Faculty of Economics and Business

June, 2013

Eline Minou Janssen Jozef Israelsplein3a 9718 EN Groningen Tel. +0031 (0) 6 137 537 12 E-mail: e.m. janssen@student.rug.nl

Student number: 1814796

Supervisor University Dr. K.J. Alsem

(2)

Abstract

(3)

1. Table of Contents

Abstract ... 2

1. Introduction ... 5

1.1 Background ... 5

1.2 Relevance of the research ... 9

1.2 Research Questions ... 11

1.4 Research Structure ... 12

2. Theoretical Framework ... 13

2.1 Perceived fit ... 14

2.2 Attitudinal Loyalty ... 18

2.3 Effect of Perceived Fit on Attitudinal Loyalty ... 22

2.4 Moderator 1: Brand Trust ... 25

2.5 Effect of Brand Trust on the relationship between Perceived Fit and Attitudinal Loyalty ... 26

2.6 Moderator 2: Self-‐Congruity ... 28

2.7 Effect of Self Congruity on the relationship between Perceived Fit and Attitudinal Loyalty ... 29

3. Method ... 31

3.1 Participants ... 31

3.2 Research Design ... 32

3.3 Stimuli ... 33

3.4 Variables and Scaling ... 33

3.5 Procedure ... 36 3.6 Analysis ... 39 4. Results ... 40 4.1 Manipulation check ... 40 4.2 Descriptive Statistics ... 41 4.3 Correlations ... 41

4.4 The regression analysis ... 42

5. Discussion ... 49

H1: Main effect Perceived Fit on Attitudinal Loyalty ... 49

H2: Moderator effect Brand Trust ... 50

H3: Moderator effect Self-‐congruity ... 51

(4)

(5)

1. Introduction

Today's consumers want variety and choice (Nijssen, 1999), which has increased the opportunity for (line or brand) extensions. Extensions have become a popular and challenging part of a brand strategy as it should keep existing customers and attract new ones to (re)purchase. In this context; where would Apple be without the I-pad, I-phone and I-Tunes? In an increasingly competitive and cluttered retail market, today’s innovations become more and more important: the brand needs them, the customer wants them and the trade requires them.

In most literature an extension strategy refers to ‘brand extensions’, yet implicitly brand and line are meant. These constructs are regularly, and wrongly, used interchangeably but should really be seen as distinct concepts. A line extension is defined as a new product variant in an existing product category. A brand extension is a product launch in a new category, thereby using the current brand name (Aaker and Keller, 1990). The latter is the strategy that will be investigated in this research.

1.1 Background

A great deal has been written and said about the use of brand extensions. Marketers have long recognized that this strategy provides considerable advantages compared to the creation of an entirely new brand. Brand extension is supposed to be cheaper, quicker and less risky than creating a new brand. (Martinez and De Chernatony, 2004). This is probably the reason why in recent decades brand extensions tend to be one of the most profitable growth strategies for companies (Martinez and Pina, 2003), which is illustrated by the fact that, on average, eight out of every ten new products are introduced as an extension of existing brands (Ourusoff et al., 1992).

(6)

consumers because consumers transfer positive attitudes or affect from the parent brand to its extension. Yet, this transfer may not be automatically dependent on the fit or similarity that consumers perceive between a parent brand and its extension (Aaker and Keller, 1990; Boush and Loken, 1991). The extension strategy also has a number of pitfalls. Marketers for example fear that inappropriate extensions might flop, or worse, weaken the existing value of the brand name (Aaker and Biel, 1993). Extensions that fail to meet consumer expectations carry the risk of diluting what the brand name means to consumers, especially in the case of extensions that are inconsistent with the brand's image (John et al., 1998). Perceiving or not perceiving fit is determined in the minds’ of the consumers, via the S-O-R processing method (Mehrabian and Russell, 1974). Perceived fit is a concept that has been broadly discussed in literature, which has been graphically presented in figure 1.1 and 1.2.

Author Fit perception

(Category/Image/Both)

Research area; Fit on Extensions

Aaker and Keller (1990)

Category How consumers form attitudes toward brand extensions.

Park et al. (1991) Both; product feature

similarity and brand concept consistency.

Most favorable reactions when brand extensions are consistent. prestige brand with concept consistency, functional brand with feature similarity

Aaker and Keller (1992) Category Evaluations of extensions

characterized by concrete, tangible attribute associations

Broniarczyk and Alba (1994) Both; brand affect and

category similarity.

Importance of brand specific associations.

Bhat and Reddy (1997) Both; parent brand attribute

beliefs and affect Importance of an extension's fit with the parent's image and category similarity

Bridges (2000) Both; dominant attribute and

dominant non-‐attribute associations

Importance of consistency and transfer parent brand's dominant association to the extension

Bhat and Reddy (2001) Both; product category fit+

brand image fit Role of parent brand affect and attribute associations play in extension evaluation.

Mao and Krishnan (2006) Both; brand prototype fit and

product exemplar Consumers will favorably evaluate a brand extension when (a) it has high fit with the brand and (b) the brand has positive evaluations.

(7)

Author Fit Perception Research area; Fit on Parent Brand

Romeo (1991) Category Effect of unsuccessful

extensions on a family brand's image, dilute family brand beliefs and create negative affect.

Loken and John (1993) Both; brand affect and product

category similarity

Brand extension failures can cause "dilution" of specific attribute beliefs about an established brand name. Park et al. (1993)

Both; brand category incongruity with functional and prestige brand concept

Poor fitting brand extensions cause dilution effects but dilution effects can also occur for high fitting brand extensions. Gürhan-‐Canli and Maheswaran

(1998)

1.2 Relevance of the research

The above literature contemplates on constructs that contribute to the suggestion to empirically research the relation between brand image (associations) and customers (Alsem and Kostelijk, 2008). The following section illustrates how this research adds to existing research because, despite the recent interest in the evaluation of fit, several research gaps still exist.

Firstly, instead of approaching fit from the traditional product category angle (Aaker & Keller, 1990; Boush & Loken, 1991), this study perceives fit from the brand image perspective. This implies that the core interest and relevance of this study concerns the non-product-related associations of a brand. This choice expounds on the call for research initiatives to examine the relative role of brand image in brand level fit (Czellar, 2003). Since brand image appears to be a strategic tool for developing brand loyalty (Yoo et. al., 2000) it is rather surprising that this perspective on fit has not yet received much attention (Milberg et al., 1997).

Secondly, although consumer evaluations of extension success have been the focus of a great amount of studies (Aaker and Keller, 1990; Boush and Loken, 1991; Keller and Aaker, 1992; Bottomley and Holden, 2001) evaluations related to the parent brand have received much less attention. The studies that did examine the effect of extensions on the parent brand, merely focused on reciprocity effects i.e. enhancement and dilution (e.g. Loken and Roedder John, 1993;

Perceived

Fit

Figure 1.3 Previous research on fit

(10)

Martinez and Pina, 2003; Martinez and de Chernatony, 2004). These studies are very much related to change in the magnitude of (parent brand) attitudinal loyalty, as reaction to certain brand extensions. This thesis research, however, measures differences in attitudinal loyalty between two groups confronted with dissimilar brand extensions. Figure 1.1 and 1.2 illustrate the fields that have mostly been researched in previous studies.

Furthermore, the studies that did focus on the parent brand primarily emphasized on parent brand name (Gürhan-Canli and Maheswaran, 1998), brand image (e.g. Martínez Salinas and Pina Pérez (2009), Martinez and De Chernatony (2004)) or parent brand believes (Loken and John, 1993). However, there is hardly any empirical research about the relation between the strengths and uniqueness of associations and performance measures such as brand loyalty (Alsem and Kostelijk, 2008). Therefore the perception of fit in this study be directed towards the rather uncharted area of attitudinal loyalty as most studies of loyalty have already been approached from a behavioral perspective (e.g. Mc Mullan (2005), Yim and Kannan (1999). This is supported by Malhotra (2005) who states that affect, besides cognition, has emerged as an important construct in consumer decision making. This research will fill the gap to understand the nature of attitudinal loyalty and how the construct as a whole is affected by different image association’s i.e. different levels of fit.

(11)

Attention for the second moderator, self-congruity, is derived from the trade and academic literature (e.g. Aaker 1997). It stated brand personality to be an important component to be leveraged on when introducing an extension (e.g. Lau & Phau, 2007). Its relevance can further be derived from a suggestion to devote more empirical research to the relation between brand image (associations) and customers (Alsem and Kostelijk, 2008). For example, it is assumed that customers prefer brands that have images compatible with their perceptions of the self (Zinkhan and Hong, 1991). There are some studies in this field (Sirgy, 1982; Sirgy et al., 1997; Helgeson and Supphellen, 2004) but they come to different conclusions. In an attempt to find out more about this construct, it would be interesting to see what the role of self-congruity would be in the relation between perceived fit (from image perspective) and attitudinal loyalty and in what direction it might impact the effect and to what extent.

From a managerial perspective, it is imperative to know which of the potentially relevant determinants of attitudinal loyalty should receive the most attention. When indeed different levels of fit (i.e. high vs low) result in varying levels of attitudinal loyalty, brand managers must take up the results the way they are and figure how the outcomes could be best incorporated into their branding strategy. Hence, it is eminent to have relevant and timely research to find and monitor the answers to key questions, driving customer loyalty.

1.2 Research Questions

The research question for this study will be:

How does perceived fit affect attitudinal loyalty and how is this relationship influenced by brand trust and self-congruity?

The sub-questions supportive to the main research question are:

Would extensions, differing on the fit dimension, influence attitudinal loyalty differently?

How does brand trust influence the relationship between perceived fit and attitudinal loyalty?

And similarly,

(12)

1.4 Research Structure

(13)

2. Theoretical Framework

A decision to extend a brand into new product categories is definitely not a clear, straightforward one as it might impact the parent brand in various ways. A first goal of this study is to addresses a significant gap in the current literature by developing and testing a model that explores the antecedents of brand image fit and its relationship with attitudinal loyalty. In order to understand the consequences of brand extensions on the parent brand, the possible underlying motivations first need to be explored. From theory it has been derived that the relationship between perceived fit and attitudinal loyalty could be influenced by brand trust and self-congruity. The relevance to incorporate brand trust in the conceptual model relies in the emerging theory of brand commitment (Fournier, 1998; Gundlach et. al., 1995). The inclusion of self-congruity is derived from the fact that so far, studies of self-congruity have not established consensus over its influence (Sirgy et al., 1997; Helgeson and Supphellen, 2004). Additionally, self-congruity elaborates on the suggestion to investigate more into the relation between brand image (associations) and customers (Alsem and Kostelijk, 2008).

Variance between a consistent and an inconsistent brand extension has been defined as a difference in ‘perceived fit’. Although literature approaches fit from altered perspectives and relevancies, the overarching idea is commonly accepted: perceived fit examines the similarity between an extension and the parent brand on a continuum ranging from high to low similarity respectively high to low perceived fit.

(14)

Attitudinal loyalty Perceived Fit Brand Trust Self - Congruity Bra H1 (+) _{H3 (+)} H2 (+)

Figure 2.1: Conceptual model

The conceptual model (see figure 2.1), included all possible relations that will be examined in this research. By discussing the relevant literature, a theoretical basis for the answer to the problem statement will be made.

2.1 Perceived fit

Drivers of brand extension success have been extensively researched and Volckner and Sattler (2006), amongst others, found ‘fit’ to be the best determinant. The pioneers in the area of brand extensions and perceived fit, Aaker and Keller (1990), defined perceived fit as ‘the extent to which a consumer perceives the new item to be consistent with the parent brand or as Arslan & Altuna (2010) state ‘the degree to which they are related’. Although it is generally agreed that fit is vitally important, there is considerable conflict about its underlying dimensions (Grime et. al., 2002). According to the literature, fit comprises of a number of dimensions, including similarity, typicality, relatedness and brand concept consistency.

(15)

Fit approached via the product category fit perspective states that judgments of fit are dependent on the relatedness between the extension category and the brand (cf. Morrin 1999; Smith and Park 1992). Consumers view a brand extension as an exemplar of the parent brand category (e.g. Park et al., 1991; Loken and John, 1993) and are likely to base their assumptions about the new information on its category membership (Simmons and Lynch, 1991). For example, a bicycle or motorcycle is classified as ‘an object with two wheels’. However, if consumers were told the object also had ‘pedals’, they would most likely categorize it as a bicycle, and dedicate associations such as “good for exercise” to it (Sheinin, 1998).

Perceived fit is, however, not restricted to product category similarity (Arslan and Altuna, 2010). Perceptions of fit may also be grounded in its relevance to abstract brand benefits (Broniarczyk and Alba, 1994). These abstract meanings or brand concepts position products in the minds of consumers and differentiate them from other brands in the same product category (Park et al., 1986). This in is line with the “associative network theory”, which defines a brand image as a mental scheme formed by a network of concepts and interconnected by linkages or associations (Morrin, 1999). For example, consider the product category "toys" as a possible extension of the brand McDonald's. At the level of product features (category), toys are clearly dissimilar to hamburgers. However, in light of McDonald's new advertising campaign and slogan, "food, folks, and fun," toys may be perceived as an appropriate extension for McDonald's as McDonald's brand concept (image) would be consistent with the "fun" that is associated with toys (Park et al., 1991).

(16)

Bra nd trust Bra nd trust

Figure 2.2: Sequential S-O-R model

Organism

S

_R

Figure 2.3: Dynamic S-O-R model

S

1 ₂ 3 4 5 6

O

R

7

Figure 1.1 and 1.2 illustrate that fit is often determined by both described perspectives. Romeo (1991) and Boush and Loken (1991) approach fit via the category dimension, whereas Bhat and Reddy (2001) and Batra et. al (2010) state image to be essential.

Hence, an answer should be provided to the reasoning behind the determination of ‘fit’. This triggered the underlying research to this thesis where image similarity is related to brand extensions and their perceived fit. A framework that can be used to describe the information processing flows required to establish fit, is the Stimulus-Organism-Response (S-O-R) framework (Mehrabian and Russell, 1974) which will be elaborated on next. This model (figure 2.2) emphasizes on factors, internal to the individual and elaborates on the simplistic Input-Output (I-O) model (Malhotra, 2005).

(17)

Jacoby (2002) has portrayed the S-O-R constructs as overlapping circles to do justice to the psychological and dynamic phenomena that underlie perceived fit (figure 2.3). It clarifies to a greater extent why every individual relies on his or her own perceptions to verify or categorize the incoming information. Related to ‘perceived fit’, the S-O-R model posits that the consumer is the recipient of all the information that is communicated by the brand, intended or unconsciously.

This model is often applied in purchase situations where atmosphere embodies the role of stimulus (Eroglu et al., 2001). When translated to the era of brand extensions, the extension (S) affects customers (O) that evaluate the information provided to them which affects their response (R) i.e. judgment of fit. This model helps to investigate how a stimulus, affects a consumer's cognitive structure for a brand and how this affects the possible transfer of associations to other product categories. It implies whether or not consumers perceive the extension to fit with their previous brand beliefs. As long as the customer does not perceive the brand message the way it is intended, responses to the extension will be very diverse. In this manner the same extension, either consistent or inconsistent, can lead to different perceptions of fit across customers.

(18)

When the company launches an extension coherent with the pre-existing associations, the brand scheme will assimilate the new concept with no significant alterations (Martínez and Pina Pérez, 2009) and this will contribute to the establishment of a single, overarching image across a brands’ portfolio (Boush and Loken, 1991). Hence, high perceived fit results when consumers identify explanatory links that make the brand category "hang together" and permit it to remain cohesive when an extension is introduced (Bridges et. al., 2000). However, despite the considerable effort brand managers dedicate to avoid serious brand image incongruity, the parent brand concept and the extension are not always perceived to be consistent with one another (Meyvis and Janiszewski, 2004). Inconsistencies across the explanatory links in sector 4, may lead to low fit perceptions, either because the dominant brand associations are unnoticed or because they are simply not meaningful in the extension context (Bridges et al, 2000). This discrepancy in the minds’ of the consumers can lead to a low perceived image fit.

In this research brand image fit, or a lack of it, will be operationalized with the scale developed by Bhat & Reddy (2001). Their conceptualization refers to consumers' perceptions of the similarity of the extension's initial image with that of the parent brand. It must be noted that image fit, as conceived by Bhat & Reddy (2001), is broader than the concept consistency approach of Park et al.'s (1991), as they also include the brands’ quality. The underlying assumption therefore is that the beliefs or attitudes regarding the original brand will be transmitted to the extension, and a greater perceived quality in the original brand will have a positive effect on acceptance of the extension (Milewicz and Herbig, 1994).

2.2 Attitudinal Loyalty

(19)

and therefore an overview of definitions is provided by figure 2.4. The next section will elaborate further the issue of loyalty and in particular on attitudinal loyalty.

Figure 2.4 different approaches to the loyalty concept

(20)

Figure 2.5 loyalty models respectively Dick and Basu (left) and Oliver (right)

A deeper understanding on the construct of attitudinal loyalty is established by elaborating on its three components, respectively affective-cognitive and conative. Dick and Basu (1994) theorize that the formation of overall loyalty consists of three attitudinal stages, cognition-affection-conation, which in the end will lead to a behavioral manifestation, referred to as the ‘repeat patronage’ phase (figure 2.5). They state that attitude is formed by incorporating cognition, affection and conation into one step: “each may play a role in defining the nature of the attitude and consequently its relationship with patronage behavior”. This view, amongst others (cf. Crites et. al., 1994; Eagly et. al., 1994), has led to endless debates about whether affect and cognition should be seen as independent or, as distinct but highly correlated components. Oliver (1999) for example agrees on the constructs proposed by Dick and Basu (1994), but notes that consumers become ‘loyal’ according to a specified sequence, as described in figure 2.5. His framework has ever since served as a basis for several research studies on loyalty (e.g. Chaudhuri and Holbrook, (2001); Mc Mullan, (2005)) and is consistent with psychological models of individual behavior (Ajzen and Fishbein, 1991).

(21)

that and why the attitudinal component is crucial to the loyalty concept. It therefore seems appropriate to provide a more in-depth view on people’s attitudes towards a brand after a brand extension has been launched. The construct of attitudinal loyalty will thus be adopted as dependent variable and its components will be elaborated on next.

Cognitive component

Some authors take a narrow view of attitudes and define them as evaluative judgments rather than emotional states (Bagozzi et al., 1999). The cognitive stage is based on brand belief only: prior knowledge or recent experienced-based information. What this implies is that, for a consumer to become and remain cognitively loyal, he or she must have faith in the brand that its products remain to be the best choice alternative (East and Hammond, 1996). Such evaluative judgments (e.g., by good-bad reactions) are often seen as indications of satisfaction (or dissatisfaction). Loyalists are usually, correctly, labeled as satisfied customers and many empirical studies have indeed shown that satisfaction is an antecedent of brand attitude, brand intention, and attitudinal brand loyalty (e.g. Oliver, 1980). The cognitive or evaluative component can therefore be stated to represent the intentional aspect of attitudinal loyalty (Russell-Bennett et. al., 2007).

Affective component

(22)

Conative component

According to Oliver’s framework the conative component of attitudinal loyalty could be seen as an even deeper level of commitment, compared to the previously discussed affective component. It considers the brand-specific commitment to purchase, which can therefore be compared to a certain degree of brand love. Brand love is defined as the degree of intimacy, passion and commitment a consumer has for a specific brand and it has the strong potential to result in positive word of mouth (Sternberg, 1986). In this stage, the company’s (or brands’) offerings seem to fit exceptionally well with the consumers’ needs (Jones & Sasser, 1995), which, not surprisingly, result in a drive to continue interaction with the product or brand in the (nearby)

In this research attitudinal loyalty is defined an attitudinal predisposition towards a brand, consisting of commitment, recommendation and intention to repurchase the brand ( Russell-Bennett et. al., 2007) as all aspects (commitment, recommendation and intention) have shown to be relevant in de configuration of attitudinal loyalty.

2.3 Effect of Perceived Fit on Attitudinal Loyalty

The main interest in this study is how extensions, varying on the fit continuum, affect consumers’ attitudinal loyalty of the core brand. Rundle-Thiele and Bennett (2001) state that a branding strategy can capitalize on loyalty and more specifically in the domain of prestige brands: symbolic associations play an important role in nurturing consumer loyalty (Anisimova (2007).

(23)

In line with the bookkeeping model, information about an attribute that is inconsistent with the family brand will lead to belief dilution, regardless of whether that information is presented in the context of a low or moderately typical extension (Weber and Crocker, 1983). This is likely to occur when a new product launch is perceived to be inconsistent with consumers' expectations of the core brand (John et al., 1998). For example, if Neutrogena, a brand name associated with "mildness" introduces a new product perceived by consumers to be "harsh" under the same family brand name (e.g. a "strong" Neutrogena shampoo), this introduction should dilute the belief that "Neutrogena products are mild" (Loken and John, 1993). Such incongruent brand extensions are evaluated less favorably and can lead to negative perceptions that may be difficult to reverse (Aaker, 1996).

A lack of consistency or lack of ‘fit’ between the core brand and the extension may be harmful for a brand as it might weaken or dilute feelings and beliefs consumers’ hold about the parent brand (Ries and Trout, 1986). Dilution in this study is defined as ’a negative change in consumer beliefs’ (John et al., 1998) and occurs when specific extension associations (e.g. an extension of BMW which shows a lack of technological development) are inconsistent with parent brand beliefs (Loken and John, 1993). This is likely to confuse (potential) customers leading to negative feelings and beliefs towards with a parent brand. Consequently, the affect towards the parent brand should be negatively impacted as well.

Even though the associations for the new product may be positive, negative associations might be created that adversely affect its associations (Pitta and Katsanis, 1995). For example, if consumers register something out of the ordinary, such as a brand extension, they will often experience some degree of surprise which is followed by a positive or negative emotion corresponding to how pleasant ‘the experience’ is (Lindgreen and Vanhamme, 2003) In this context, if people experience anger, frustration, or irritation due to an inconsistent brand extension, they are likely to spread negative messages that slander the firm (Wetzer et. al., 2007) or brand. In other words, they will act as negative referrals.

(24)

brands’ intended message might lead to a complete loss of reputation. This is unfortunate as a good brand reputation elicits a greater perceptual enhancement of the brands in consumers’ mind. The better a company’s reputation, the higher its chances are of getting a favorable first hearing for a new product (Herbig and Milewicz, 1997) and this positive affect is likely to translate to the parent brand. The more stable a brands’ reputation, the more credible the message is and finally, the more likely it is that positive and influential word of mouth between a brands customers will take place.

Furthermore, Launching close extensions to the parent brand can be a way of strengthening current associations (Bhat and Reddy, 2001). Consistent associations will facilitate recognition and recall of the brand (Nandan, 2005) and similarly enable better understanding of a brands‘ image (Milewicz and Herbig, 1994). Based on the fact that brand image is affected by an extension strategy, Pitta and Katsanis (1995) suggest that an extension may reinforce the core product’s brand image instead of weakening it as consumers’ perceptions may favor the brand image after the extension (Martinez and De Chernatony, 2004).

Although literature has not explicitly assessed the effect of fit on attitudinal loyalty and most studies measuring parent brand affect have stated that poor image fit will lead to dilution effects (e.g. Martinez and Pina (2003); Martinez and de Chernatony (2004); Lau & Phau (2007)), the interaction between extension and parent brand is clearly not a one-sighted construct.

Elaborating on the different findings in this field of research, it is expected that higher levels of perceived fit will result in higher levels of attitudinal loyalty compared to an extension with low perceived fit. This hypothesize is derived from the reasoning that perceptions about a brand image are affected by an extension strategy (Martinez, E., & De Chernatony, 2004), and image is compiled by associations which in turn consist of feelings and beliefs. And beliefs and feelings, as described in the two approached loyalty models, are said to be determinants of attitudinal loyalty. Therefore the following hypothesis has been conducted:

H1; Extensions with high perceived fit will result in higher attitudinal loyalty towards the parent brand compared to extensions with low perceived fit.

(25)

2.4 Moderator 1: Brand Trust

Early investigation of Rempel et al. (1985) had pointed out that trust is one of the most desired qualities in any close relationship. However for years the concept of trust in customer-brand relationships, has not received any theoretical or empirical attention. But, due to the relational orientation emerging in marketing activities, brand trust has recently won a more central role in marketing literature and practice (Morgan and Hunt, 1994). Along in this line of reasoning Delgado-Ballester et al. (2003) suggest brand trust to be ‘the most important attribute any brand can have’.

As brand trust is derived from both psychology and marketing fields there does not seem to be a consensus on its exact definition. However, it has long been clear that trust can only occur in a familiar world; it needs history as a reliable background (Luhmann, 2001). A more careful review on the existing literature clarifies that most perspectives do in fact agree on two critical components of trust: confident expectations and risk, of which the latter is also known as a feeling of security (Delgado-Ballester et al, 2003).

These originated central elements of trust, familiarity, confidence and risk, are also determined in Luhmann’s (1979) ‘social theory of trust’, and presented in Figure 2.6. The model visualizes the underlying mechanism that brand trust develops through previous interaction and experience with the brand (Ravald and Grönroos, 1996). Trust becomes more central in purchase situations with increased risk and hence, less risk is perceived when the brand becomes more familiar to the consumer due to increased interactions. In this case trust reverts to confidence (Elliott and Yannopoulou, 2007).

(26)

In line with this model, trust is defined as “the extent to which a person is confident in, and willing to act on the basis of the words, actions, decisions of others” (McAllister, 1995) and, uniquely in the consumer-brand domain, “the willingness of the average consumer to rely on the ability of the brand to perform its stated function” (Chaudhuri and Holbrook, 2001).

This last definition is used in this study. Brand trust, as defined here, focuses on the perceived brand performance: a certain reliance on performing activities and delivering expected outcomes. This dependability should not be taken advantage of as that could damage the customer-brand relationship to a great extent. The view of Chaudhuri and Holbrook (2001) is also in concordance with Moorman et al. (1992) as both studies have incorporated the behavioral intention of ‘willingness’ into their definition. This underscores a sense of predictability that the brand satisfies customers’ needs in a constant and positive manner (Deighton, 1992).

2.5 Effect of Brand Trust on the relationship between Perceived Fit and

Attitudinal Loyalty

“The ultimate goal of marketing is to generate an intense bond between the consumer and the brand, and the main ingredient of this bond is trust” (Hiscock, 2001). Therefore, it seems

essential to understand consumer-brand relations in the evolution of marketing and consumer behavior theories (Delgado-Ballester et al, 2003). The relevance of the brand trust concept relies in the emerging theory of brand commitment (Fournier, 1998; Gundlach et. al., 1995).

(27)

Nonetheless, little is known about the moderator brand trust as a factor intervening in the relationship between perceived fit and attitudinal loyalty. What we do know is that a trustworthy brand is viewed as a promise of future performance and a feeling of security, held by the customer, that the brand will meet his/her consumption expectations (Deighton, 1992) and that familiarity is said to have a positive effect on the general perceived brand image (Czeller, 2003).

People can only build relationships with companies they “know” (Alsem and Kostelijk, 2008). This is also reflected in Keller's definition of Customer Based Brand Equity: the effect that customers react more favorably to the marketing instruments of a brand they know than to the marketing instruments of a brand they do not know (Keller, 2003). Moreover, trustworthily customers believe that the brand will not promote a flawed product; they will forsake all other alternatives and rely on that single brand which they expect not to let them down. An extension, with either high or low perceived fit, may therefore benefit from customers faith in the brands’ good intentions. This expectation is perhaps even better understood by comparing it to trust-relations of human beings. Trusting a person implicitly means that there is a reasonably high chance that his or her actions will result in positive, or at least non-negative, outcomes (Delgado-Ballester et al, 2003). Translating it back; a trustworthily customer will be more willing to rely on a partner in whom one has confidence.

Thus, when a customer trusts a brand due to previously satisfying experiences, an extension expanded to other product categories will be perceived with less risk (Keller and Aaker, 1992) and hence, supplier selection for existing and new products stems from trust (Sichtmann, 2007). A positive extension evaluation could be regarded as an extra dimension to the brand that in turn leads to positive parent brand feelings (Delgado-Ballester and Munuera-Alemán, 2001) Moreover, a satisfying extension is also more likely to result in positive word of mouth behavior (Bolton and Lemon 1999), which, as clarified by the previously discussed loyalty models, is positively associated with attitudinal loyalty. Based on the above literature, it is expected that higher levels of brand trust will positively influence the relationship between perceived fit and attitudinal loyalty. Therefore the following hypothesis has been conducted:

H2: Higher levels of brand trust will enhance the positive relationship between perceived fit and attitudinal loyalty.

(28)

2.6 Moderator 2: Self-Congruity

In today’s highly competitive business environment, meanings of products play a critical role in consumers’ choices. The way a brand is positioned in terms of brand images is extremely important (Bhat and Reddy, 1998). Brands (and their associated images) enable consumers to express who they are, what they are, where they are, and how they want to be viewed (Greaff, 1996). This matching process between human and a brand personality is labeled ‘self-congruity’ and provides a useful research method for consumer studies. Its dimensions and relevance will be discussed in the following sections.

As already touched upon, much research has uncovered the fact that consumers purchase goods that act as a way to express their identity. Sirgy (1985) argued that brands have personal image attributes; just as people do (e.g. a brand can have a youthful and friendly image). More specifically, brands are considered to have a “personality” (Kressmann, 2006) or brand-user image. There have been great amounts of practitioners’ efforts invested in identifying “what brand personality is” and “how it works.” In fact, brand personality, generally refers to the human characteristics associated with a brand (Aaker, 1997), is about (some) non-utilitarian brand attributions (Keller, 1993). It concerns the personality or product images that can be described in terms of a set of attributes such as friendly, modern, youthful and traditional. By providing such descriptive explanations for abstract brand dimensions; brand personality creates a more meaningful and contemporary brand meaning (Geuens et. al., 2009).

(29)

with consumers' self-concept is referred to as self-congruity. Self-congruity plays an important role in purchase motivation and brand loyalty (Sirgy, 1985).

Self-congruity is guided by concept motives such as the need for esteem and self-consistency (Aaker, 1997). The need for self-esteem is referred to as the ideal self-image. This ideal state consists of behavior that allows individuals to reduce discrepancies between their actual and ideal self which serves to boost self-esteem. Conversely, the need for self-consistency motivates people to behave in ways consistent with how they see themselves, in other words: consistent with their actual self (Kressmann, 2006). Putting both lines of reasoning into practice: a brand may be used to express the self because the brand is congruent with the self: I buy a jeep because I view myself as tough. However a brand may also be used to display the ideal self to others: I buy a Jeep to show others that I am tough.

According to Baumeister and Leary (1995) “the desire for personal attachment is one of the most far-reaching and integrative concepts available to understand human nature.” Using interpersonal attachment as an analogy to product or brand attachment, it seems reasonable to assume that people indeed build and maintain relationships with brands on an emotional level. In this perspective, individuals use personal or subjective criteria such as taste, pride, desire for adventure, and desire for expressing themselves in their consumption decisions (Bhat and Reddy, 1998). Consumers can thus use brands to express, display, and better understand, their selves (Swaminathan et al., 2007).

2.7 Effect of Self Congruity on the relationship between Perceived Fit and

Attitudinal Loyalty

(30)

The idea that consumers prefer brands that have images similar to their own self-image is referred to as the image congruence hypothesis (Greaff, 1996). Conversely, low self-congruity typically occurs when the product–user image does not match the consumer’s perceived self-image (Cowart, 2008). As people act in ways that maintain and enhance their self-concept (Graeff, 1996) an extension incompatible with the parents brand image will harm individuals with high on self-congruity, as their self-concept is affected. A logical inference is that the feelings and beliefs about the parent brand will be diluted. This is supported by (Heath and Scott, 1998) who state that brand related information, inconsistent with the customers’ self-concept, is unlikely to gain customers’ attention, acceptance, and retention (Heath and Scott, 1998). Counter wise, higher self-congruity is experienced when the product–user image matches consumers’ own self-image. Brand consumption in the image-consistency case would allow consumers to express their identities with this new extension, as the extension image is perceived to be similar to their own self-images (Aaker, 1999). Hogg et al. (2000) provide support for a strong relationship between self-image congruency and brand preference. Consumer researchers confirm this as they state interaction between images results in self-image/product image congruity, which in turn affects the consumer’s product preference and purchase intention (Thorbjørnsen et al., 2007). This would provide evidence for a relation between self-congruity and attitudinal loyalty (Zinkhan and Hong,1991).

Apparently, self-congruity appears to be an important factor in determining the quality of the relationship between brand and individual and concerns the individuals’ beliefs about their identity, values, lifestyles, preferences, and habits. And, according to Kressman (2006), ‘both similarity and positive dissimilarity in interpersonal relationships can enhance the quality of a relationship, which translates to consumers' relationships with products and brands’. Based on the above literature, it is expected that higher levels of self-congruity type of processing is more likely to result in positive brand affect when the self-image matches the extension image. Therefore the following hypothesis is purposed;

(31)

3. Method

This section describes the research methods that will be used to generate an answer to the hypotheses outlined in the previous chapter. The central question that will be operationalized is:

how does perceived image fit influence attitudinal loyalty towards the parent brand and do the moderating effects of brand trust and self-congruity hold true for both types of extensions? The

expected hypotheses are visualized in the theoretical framework (figure 2.1). Two pretests were conducted to select a familiar brand that was symbolic in nature, and to develop extension stimuli for the main study. The subsequent section elaborates on the design and procedure of the study of which the results will be reported in chapter 4.

3.1 Participants

In total 311 participants took part in the experiment, of which 78 had to be omitted (table 3.1). Explanation is unlikely to be found in ‘fatigue effects’ as the survey on average took 5-10 minutes to complete. Investigation of the lacking responses has neither shown a structural pattern for the moment the participants had stopped. The reasons for exclusion were primarily due to participants that have accidently overlooked how to properly end the questionnaire (n=62). Other reasons could be derived from a lack of interest, familiarly or experience with the focal brand. The final analysis is done with 233 participants of which 135 (58%) were men and 98 (42%) women. According to Malhotra (2007), the target sample size for a study like this is 100 respondents, because two conditions are tested and each condition should contain at least 50 respondents. Due to the fact that Ben & Jerry’s targets ‘health-conscious young adults and affluent, pleasure seeking adults who are in general brand conscious’1. It is likely to assume that participant’s aged between 18-30 are familiar with the brand and thereby the students who participated (n=203, 87 %) form a representative sample for this research. The results can be generalized across the Netherlands as almost half of the sample (42.50%) lives throughout the Netherlands, but outside of Groningen. The participants of the pretest (20) contained the same demographic characteristics as the sample population. Data were collected in May 2013.

(32)

Deleted Respondents

Reason for deletion No. of Respondents Percentage

Total participants 311 100 %

Not completed questionnaires 62 19.94 %

Not familiar with Ben & Jerry’s 2 .64%

No experience with Ben & Jerry 8 2.57 %

Did not take the survey seriously 6 1.93 %

Total valid respondents 233 74.92 %

Table 3.1 respondents

Demographic Variable Sub-Constructs Number Percentage of sample

Gender Male 135 57,90%

Female 98 42,10%

Extension Chocolate chip 114 48,90%

Salad 119 51,10% Age (M = 33.6) 20-30 203 87,00% 30-40 2 0,85% 40-50 4 1,72% 50+ 24 10,30% City Groningen 124 53,20% Other in Holland 99 42,50% Abroad 10 4,30% Total respondents - 233 100,00%

Table 3.2 Demographics of the sample

3.2 Research Design

(33)

a between subjects design which aims to discover whether the participants assigned to the high fit group (n=114) differed in their responses compared to the low fit group (n=119). An advantage of the random allocation to groups is that the subjects within the groups, on average, have the same profiles (Aaker, 1997) which justify comparison between them.

3.3 Stimuli

Before the final survey was distributed certain requirements for brand choice had to be met. In order to prevent the research from being influenced by the type of parent brand, its strength and stretchablility, this research will use one particular brand. An ice cream brand, Ben & Jerry’s, has been selected for this experiment for the following reasons. First, respondents had to be familiar with the brand as that enables participants to draw upon their prior knowledge which helps to retrieve specific brand associations (Low and Lamb, 2000) which can be used to determine the fit dimension, our independent variable. Another benefit of familiarity is that real brands exploit the richness of the ‘personality’ associated with them (Sheinin and Schmitt, 1998) which is required for the examination of self-congruity in this research. Second, exploratory results from the second pretest suggested that Ben & Jerry’s was a narrow brand, known for a single-category. This was important as previous research suggests that consumers evaluate extensions of broad brands with alternative referent points (Mao and Krishnan, 2006). A narrow brand also ensures more precise measures of the third variables. Furthermore, the predominant symbolic image of Ben & Jerry’s was a prerequisite for appropriate judgment of perceived image fit.

3.4 Variables and Scaling

(34)

reliable. Table 3.3 provides an overview of the variables and scaling, table 3.4 provides some examples of how the questions were formulated.

Author Constructs

Sub-Constructs

Cronbach’s Alfa’s

No. of items

Bhat & Reddy (2001) Perceived Fit (PFIT) - 0,93 2

Russell-Bennett et al. (2007) Attitudinal Loyalty (AttLoyal)

- 0,87 5

Chaudhuri & Holbrook (2001) Brand Trust (BT) - 0,79 4 Kressmann (2006) Self-Congruity (SC) Brand

personality

0,77 15

Actual self 0,78 15

Ideal self 0,92 15

Bhat and Reddy (1998) Symbolic brand (pre-test)

- - 20

Table 3.3 Variables and Scaling

Independent variable: Perceived image fit (PFIT)

Brand image fit was measured by using a scale developed by Bhat and Reddy (2001). The two items enquired the participants on whether the extension and parent brand exposed similar images and whether the extension brand conveyed the same impression as its parent brand. These questions served two purposes: firstly as measurement for the independent variable perceived fit and secondly as manipulation tool for perceived fit, measured in the pre-test. Six-point scales were used with 1 indicating ‘strongly disagree’ and 6 indicating ‘strongly agree’. These scales have a Cronbach’s Alpha reliability of 0.93.

(35)

statements on a six-point likert scale, varying in the elements that embody the response continuum. The five questions were internally reliable with a cronbach’s alpha of 0.87.

Moderator 1: Brand Trust (BT)

Brand trust was measured with a four-item scale adopted from Chaudhuri and Holbrook (2001). Each of the four statements had to be rated on a six-point Likert scale, ranging from 1 ‘strongly disagree’ to 6 ‘strongly agree’. Coefficient alpha for this four-item index of brand trust was 0.79.

Moderator 2: Self-Congruity (SC) The scale measuring self-congruity is based on the difference scores between Aakers’ (1997) brand personality dimensions and both magnitudes of self-congruity, the actual and ideal self (Kressmann, 2006). This choice has been made because these personalities allow describing brands, alongside people. By this it is possible to draw a comparison between people and a brand. Aaker's (1997) brand personality measure entails 15 brand personality items that enable a brand to be characterized as a person. This in turn allows for the comparison between brands and people. The participants were asked to rate the 15 statements on a six-point likert scale ranging from 1 ‘strongly disagree’ to 6 ‘strongly agree’. The difference scores were able to vary between 0 (perfect SC) and 75 (No SC). This test has to be performed three times, once for the personality dimensions of Ben and Jerry (I think Ben & Jerry’s is…), once for the actual self (to what extent do the stated personality attributes apply to you?) and once for the ideal self (to what extent the following personality attributes apply to how you like to be). To minimize potential memory effects the order of the ‘actual self’ items has been reversed. The three tests all had high item-to-total correlations exceeding 0.6 (respectively 0.77, 0.78, 0.92) and thereby, high levels of internal reliability are established.

Control Variable Gender

(36)

Construct Author Item

Perceived fit Bhat & Reddy (2001)

Chocolate chip cookies fit with the idea/image you have of Ben & Jerry’s

Chocolate chip cookies evoke the same impression as Ben & Jerry’s

Attitudinal loyalty Russell-Bennett et al.

(2007)

I would recommend Ben & Jerry’s to others Purchasing my Ben&Jerry’s would be favorable

Brand Trust Chaudhuri &

Holbrook (2001)

I trust this brand I rely on this brand

Self-congruity Kressmann (2006) Brand personality

I think Ben & Jerry’s is spirited I think Ben & Jerry’s is upper-class Actual self

I am outdoorsy I am honest Ideal self

I would like to be imaginative I would like to be charming

Symbolic brand (Pre-test) Bhat&Reddy (1998)

Using Ben&Jerry’s says something about the kind of person you are.

Table 3.4 Operationalization variables

3.5 Procedure

Prior to the actual experiment a focus group, two pre-tests and a pilot study have been performed; respondents were comparable to those in the final sample. The focus group consisted of 10 potential users (n=10) and collected potential brand extensions for the first pre-test (see appendix).

First pre-test

(37)

were opposed to 15 hypothetical brand extensions (see appendix) and had to rate them on a continuum ranging from 1 ‘not very similar’ to 10 ‘very similar’. A choice has been made to exclusively include extensions of the food & beverage category as brand extensions that are a-typical to a greater extent, such as a car, will not dilute the parent brand because associations from the parent brand to the extension will not be transferred. This would make it difficult to generate effects about attitudinal loyalty. In line with Keller and Aaker (1992), food and beverages were chosen as it was believed that subjects would enjoy purchase and usage experience. Men and women were equally assessed to avoid gender bias in the results. Furthermore, to avoid influences of earlier mentioned extensions on latter ones, the order of the products varied for each participant. The two most differing brand extensions on the fit continuum appeared to be chocolate chip cookies (Mean= 9.4) and salad (Mean= 1.5). These were used as stimuli throughout the final research.

Second Pre-test

In line with this study’s main objective of exploring the issue of brand image, the objective of this second pre-test (n=20) was three-fold. It had to verify that Ben & Jerry’s was a well-known, single-category and prestige oriented or symbolic brand. Functional brands satisfy immediate and practical needs. Symbolic brands satisfy symbolic needs such as those for self-expression and prestige, and their practical usage is only incidental (Bhat and Reddy, 2001).

(38)

ratings (P<.05). Furthermore the respondents were asked to answer the following two questions: ‘are you familiar with Ben & Jerry’s? and ‘What is/are the product categories you associate with Ben & Jerry’s’? Despite one outlier on the latter issue, these quantitative and qualitative results imply that Ben & Jerry’s is primarily a symbolic brand, known for its primary product category: ice-cream.

After the two pre-tests, the first stage of data collection involved a pilot study to test the survey instrument. The aim was to control whether everything was set up right and if the experiment actually tested what it was intended to measure (construct validity). During debriefing it was found that respondents had interpreted the meaning of some traits differently (wholesome, during and spirited). For these concepts a synonym has been provided to ensure common understanding and higher reliability. In addition, some adjustments have been made to the lay-out as some visual stimuli appeared to be confusing. The second stage of data collection involved the distribution of the survey tool to a larger sample. Participants were approached via (personal) mail or using social media. The questionnaire captured data on the focal constructs of the model. The items used to measure these constructs were based on past marketing studies.

The final questionnaire was organized in ‘five blocks’ that related to the constructs of the conceptual model. The brands' logo was printed on the questionnaire to facilitate participants' recognitions of the brand, as suggested by Yeung and Wyer Jr. (2005). Beforehand, one half of the participants was directed to the ‘high fit’ testing condition (chocolate chip cookies), the other half was bound to the ‘low fit’ condition (salad). To guarantee that the process was as randomized as possible, the Qualtrics system allocated participants to one of the two questionnaries on an individual and random basis. Both questionnaires were alike except for the provided manipulation.

(39)

of Ben and Jerry’s on a six-point Likert scale and consequently had to classify their actual (who you are) and ideal self (who you would like to be) with the same statements.

The fourth and fifth block presented the participants with respectively five questions about attitudinal loyalty and two items relating to perceived fit, both on a six point likert type scale ranging from 1(strongly disagree) to 6 (strongly agree). These two “easy-to-answer” constructs were placed towards the end as this avoids the problem of response bias from fatigue (Kressmann, 2006). Before the questions were answered the participants were opposed to the extension which they were asked to view for a couple of seconds. The printed extension, used for each group, differed in its fit with the brand. No further information was provided because it was thought that providing additional extension information would possibly bias the participants in ways that might defeat the measure of attitudinal loyalty. Afterwards the respondents were thanked for their participation and told that the information provided in the research was fictive and made up by the researcher. Lastly, it must be noted that the questionnaire contained three additional items. The first was to ensure that people read the questions carefully and the second conveyed whether the respondents met the predefined requirements of familiarity (requirement for stimuli) and experience (underlying dimension of brand trust). If these three questions resulted in an undesirable response, the respondent was deleted from the data set.

3.6 Analysis

(40)

0 1 2 3 4 5 Cookies Salad

Perceived Fit (PFIT)

Mean

4. Results

This next chapter will present the results of the regression analyses which will lead to the acceptation or rejection of the hypotheses, as presented in the theoretical framework (figure 2.1). All independent variables correlate with the dependent variable attitudinal loyalty, but the VIF-scores of all variables were below 1.1 with a Tolerance >.10, which implies that multicollinearity is not an issue (Mason and Perreault, 1991).

Before the regression analyses were performed, preliminary analyses were conducted to ensure no violation of the assumptions of normality, linearity, multicollinearity and homoscedasticity (Osborne and Waters, 2002) (see appendix). The plots confirmed that regression was the appropriate statistical analysis. Upon completion of the experiment, the following section will present the manipulation check, descriptive statistics and correlations which evidently will lead to the final regression analyses.

4.1 Manipulation check

In order to verify that the manipulation indeed resulted in different scores on the fit continuum, a t-test for unstandardized betas was conducted. To check for the manipulation, the two scenarios

(cookies & salad) were compared on the two six-item questions of perceived fit. The group opposed to cookies recalled higher perceived fit (M = 4.13, SD = 1.25) than the group that saw the salad manipulation (M = 1.61, SD = 0.78). This difference was significant, (188.09) = -18.251, p < .001. Thus, participants in the cookies condition perceived a greater fit between the extension and the parent brand compared to the participants in the salad condition. This indicates that the differences between the condition means are not likely due to chance but can probably be directed to the manipulation of the independent variable. It can be concluded that the priming of perceived fit succeeded and that the indication obtained from the pre-test, is confirmed.

(41)

4.2 Descriptive Statistics

Table 4.1 lists the means and standard deviations of the variables used in the study. All measures were obtained on a 6-point scale with 6 being highest and 1 being lowest. Attitudinal loyalty (M= 3.81) and brand trust are rated above average (M= 4.31). Considering the 6-point scale on which answers had to be given, the standard deviations of the variables are quite high; this implies that answers relating to PFIT loyalty vary quite a lot.

Descriptive Statistics

N Scale Minimum Maximum Mean SD

PFIT 233 Scale 1-6 1,00 6,00 2,84 1,629

BT 233 Scale 1-6 2,00 6,00 4,31 ,788

SC 233 Scale 0-75 6,50 38,00 16,36 5,456

AttLoyal 233 Scale 1-6 1,00 6,00 3,81 ,987

Valid N 233 - - - - -

Table: 4.1 Descriptive statistics

4.3 Correlations

As a first step in the analysis correlations between the model variables, and the control variable gender, have been calculated Correlations are suitable in demonstrating linear relationships between the variables. Table 4.2 shows many significant and relevant outcomes, in particular the relationships with the dependent variable attitudinal loyalty. None of the variables extended .7; therefore all variables were retained (Pallant, 2010). Attitudinal loyalty correlates significantly (p<.05) with all other independent variables in the model, respectively perceived fit (r=.399,

p<0.001), brand trust (r=.345, p<0.001) and to a lesser extent with self-congruity (r=.184, p=.005). Besides the previously mentioned significant relationship with attitudinal loyalty, brand

trust also correlates positively with perceived fit (r= .132, p=.045) and self-congruity (r= .248, p < .001). Self-congruity and perceived fit do not show any strong relation (r=.014), neither a significant one (p =.833). The variable gender is not part of the conceptual model but because it correlates significantly with attitudinal loyalty (r=.275, p<.001) it has been incorporated in the regression analysis, to control for possible intervening effects.

(42)

Gender PFIT BT SC AttLoyal Gender _Pearson Correlation 1 PFIT _Pearson Correlation ,099* 1 Sig. (2-tailed) ,131 BT _Pearson Correlation ,142* ,132* 1 Sig. (2-tailed) ,030 ,045 SC _Pearson Correlation ,083 ,014 ,248* 1 Sig. (2-tailed) ,207 ,833 ,000 AttLoyal _Pearson Correlation ,275* ,399* ,345* ,184* 1 Sig. (2-tailed) ,000 ,000 ,000 ,005 Table 4.2 Correlations between all variables

4.4 The regression analysis

Upon completion of the experiment, data was analyzed using a hierarchical regression analyses. After controlling for gender, the analysis started with a simple linear regression to test for evidence on H1. Subsequently the moderator variables and interaction terms were entered into the equation and with use of multiple regression models, H2 and H3 were tested.