The effect of brand personality associations

(1)

The effect of brand personality associations

Brand personality associations: an understanding of the relationship

between brand personality associations and brand attitude.

(2)

The effect of brand personality associations

Brand personality associations: an understanding of the relationship

between brand personality associations and brand attitude.

Master thesis Noordwolde, 31 October 2011 Master of Business Administration

Marketing Management & Marketing Research

Department of Marketing, Faculty of Economics and Business University of Groningen

Armand Fokko Dekker

Student number: 1733478 E-mail: afdekker@hotmail.com Address: Mauritsweg 28 8391KC Noordwolde (FR) The Netherlands Supervision

(3)

The effect of brand personality associations Management summary

Management summary

Brand personality and brand personality associations have been discussed widely in literature, however the main focus has been on the structure and scaling procedures of brand personality. No parallel research has focused on the influence of brand personality associations on brand attitude. In this research the relationship between brand personality associations and brand attitude is investigated, as well as the extent to which loyalty, brand familiarity, service quality, satisfaction and switching costs affect this relationship. This research contributes to prior literature by providing more insight into the influence of brand personality associations on brand attitude and simultaneously providing more understanding to which extent this relationship is moderated.

In order to investigate the relationship between brand personality associations and brand attitude, and the moderating effects on this relationship, an empirical research for a financial service provider on the Dutch market is conducted. The empirical research starts with a factor analysis of brand personality associations in order to identify a structure among these associations. This structure is applied in an ordinary least square regression analyses to investigate the influence of brand personality associations on brand attitude. Finally, a moderator analyses is performed to examine the moderating effects of loyalty, brand familiarity, service quality, satisfaction and switching costs. The empirical research revealed a structure of five brand personality associations components: social, development, trust, power and relaxed. Social, development and trust have a positive influence on brand attitude, whereas power and relaxed have no influence on brand attitude. Social has the most influence on brand attitude, followed by trust and development. Loyalty, service quality and satisfaction have a positive moderating effect on social. Development is negatively moderated by service quality and satisfaction. The main effects of loyalty, service quality and satisfaction showed a relatively high influence on brand attitude.

(4)

The effect of brand personality associations Preface

Preface

This report represents my final work as a student of the master Marketing within the study Business Administration at the University of Groningen. After graduating my higher education study Marketing Management at Windesheim in 2007, I decided to take the opportunity to increase my

knowledge on marketing by trying to get my master degree in marketing at the University of Groningen. It was a long and interesting journey which had its ups and downs, but I am very glad that I took this opportunity and that I have achieved a double degree in marketing. However, I could

not have done it on my own, therefore I would like to thank all the people who made it possible for me to write this thesis and to graduate from university, but I would like to mention some people in

particular.

First of all, I would like to thank my parents for being supportive and understanding during my study in Groningen. In good and in bad times, their support has always been an encouragement to finish

my study and achieve my master degree.

In addition, I would like to thank both my supervisors prof. dr. Janny Hoekstra and dr. Sonja Gensler for their help and feedback during the period I was writing my master thesis. Their feedback and

recommendations have been a great help in finalizing this report.

Moreover, I would like to thank the Friesland Bank, represented by Sander Baesjou and Sieger Kroes, for giving me the opportunity to conduct a research in the financial service market and their help

during the research.

Finally, I would like to thank all my friends and fellow students for their cooperation, advice and help during my study in Groningen. Most of all, I would like to thank them for all the memorable moments we have experienced at the University and in the city of Groningen. These moments were of great

value during my time in Groningen.

Armand Dekker

(5)

The effect of brand personality associations Table of contents

1. Introduction

In this research the effect of brand personality associations on brand attitude is discussed. In order to clarify the topic choice, this chapter will give a brief introduction on brands, brand associations and the role of brand attitude. Furthermore, the aim of the research will be presented in the following sections.

The introduction starts with relevant background information on this topic in section 1.1. Subsequently, the research objective, problem statement and research questions are presented in section 1.2. In section 1.3, a description of the research design is given, followed by the theoretical and social relevance of this research in section 1.4. The company and the brand applied in this research are presented in section 1.5. Finally, an outline of the structure of this research is given in the last section.

1.1 Background

In the last couple of years financial service brands have changed dramatically as a consequence of the fall of several global financial service providers. Economies suffered from bankruptcies and therefore rigid actions had to be taken to overcome the turbulent times of economic downtime. This resulted in the disappearance of some major brands in the financial sector, as consumers were lacking confidence in the companies behind these brands. Due to the economic crisis, once again service providers became aware of the tremendous value of a strong brand and how consumers experience their brand.

In order to build and maintain a strong brand it is important for a company, and especially for a financial service provider, to have a well-established relationship with the customers in order to earn their trust and loyalty. By focusing on the customer and building long term relationships, companies want to increase the value for the customer and maximize the long term profit of the customer (Reichheld 2003). This is brought together in the concept of customer equity management (Rust et al. 2000, Rust et al. 2004, Vogel et al. 2008). Rust et al. (2000) define customer equity as the expected lifetime value of a firm’s customer base. According to these, and many other authors, customer equity is built up around three key drivers: value equity, brand equity and relationship equity. In the customer equity concept, brand equity is the intangible assessment of a brand beyond its objective perceived value (Vogel et al. 2008). Brand associations, one of the drivers of brand equity (Aaker 1991, Keller 1993), help consumers to determine their overall view and attitude towards a brand.

(8)

equity in relation to services. However, as a result of the increase in the dependency of many modern economies on services in which brands continue to be important (Myers 2003), the development of branding theory is important from both the product as well as the service perspective (Merz et al. 2009).

Brands have a communication function for both companies and consumers. Service providers use brands in order to provide meaning to the different services offered by the firm (Keller 1993). Consumers on the other side, use brands in order to identify the products or services offered by the service provider (Keller 1993). Brand associations play an important role since consumers use these associations in several ways. Consumers use these brand associations to identify the product or service, identify themselves with the brand and to evaluate the product, service or the brand (Keller 2007, Cramer and Koene 2010). Companies instead create associations to enhance the perceived value of their product or service and to differentiate themselves from competition (Aaker and Shansby 1982, Farquhar 1990).

The brand name of the service is often inherent to the company name and therefore failure of brand management could be catastrophic. Managing the brand and their associations are therefore of the utmost importance to companies. Especially when products of service appear to be homogenous, firms can differentiate themselves through solid brand management (Keller 1993). One way to differentiate services is by applying human characteristics or values to a brand. By applying human characteristics or values to a brand, companies try to personalize the brand for the consumers (Aaker 1997, Plummer 1985, Cramer and Koene 2010). Consumers use these and their own personal characteristics and values to identify themselves with the products/services. However, the power of a brand lies in what resides in the minds of the customers (Keller 2007), therefore applying associations which fit the brand are essential when trying to build and maintain a strong brand. Successful service companies are aware of this essence of brand positioning in order to create a brand which has a sustainable competitive advantage or unique selling proposition that gives consumers a compelling reason for buying that particular brand (Aaker and Shansby 1982, Keller 2007).

(9)

1.2 Research objective, problem statement and research questions

The objective of this research is to provide insight into the influence of brand personality associations on brand attitude and the moderators which could affect this relationship. Therefore, the problem statement of this research is stated as follows:

What is the influence of brand personality associations on the brand attitude of customers and to what extent do moderators affect this relationship?

In order to answer the problem statement, the following research questions will be addressed in the literature review and empirical research:

1. What is brand attitude and what are brand personality associations?

2. Which structure of brand personality associations can empirically be identified? 3. How do brand personality associations influence brand attitude?

4. To what extent do moderators affect the relationship between brand personality associations and brand attitude?

1.3 Research design

First, the relevant literature is reviewed, which results in the formulation of hypotheses regarding the relationship between brand personality associations and brand attitude, and additionally the extent to which moderators affect this relationship. Second, the hypotheses are tested with data from a single cross sectional design, supplemented with data from secondary sources. The primary data is collected among a random selection of customers of a financial service provider through an online survey. Out of the 3000 invitations, 247 were returned completely of which 238 remained for analysis. Secondary data is obtained from the customer database of the service provider.

1.4 Academic and managerial relevance

(10)

relationship between brand personality associations and brand attitude will be investigated in this research. These insights contribute to existing literature by providing more understanding in the formation process of consumer brand preferences and the extent to which this relationship is affected by the moderators. Finally, in prior literature mainly product brands are applied in research on brand associations and brand preferences (Aaker 1999, Park and Srinivasan 1994), while in this research a brand of a service provider is used to investigate the influence of brand personality associations on brand attitude.

These contributions to literature give brand managers a better understanding about brands, brand personality associations and the influence of these associations on a brand in the following ways. First, the understanding of the influence of brand personality associations on brand attitude is relevant for a brand manager since it provides insights in what associations change consumers’ brand attitude (Keller 1993). This valuable information can serve as a basis for a brand manager to develop strategies to strengthen their brand (Keller 1993, Krishnan 1996). Second, the impact of the several moderators on brand personality associations provides managers with more insights in circumstances in which the effect of brand personality associations is affected by the moderators, and in what situations the moderators have more or less effect. Finally, the application of a service brand, instead of a product brand, will give brand managers more insights in the influence of brand personality associations and brand preferences from a service provider perspective and is of valuable information to brand managers of service providers.

1.5 Company

The brand applied in this research is the brand name of a typically Dutch service provider, the Friesland Bank. This Dutch service provider operates throughout the whole of the Netherlands and offers a wide range of products and services for both the consumer and the business market. In this research the focus is on the consumer market.

1.6 Structure of the research

(11)

The effect of brand personality associations Theoretical framework

2. Theoretical framework

This chapter contains a theoretical framework which is set up around the central question. The theoretical framework provides an overview of the existing literature on brand attitude, brand associations, brand personality associations and the moderators which could affect the relationship between brand personality associations and brand attitude. This chapter starts with an introduction and presentation of the conceptual model. The conceptual model will be an outline of the theoretical overview and forms the foundation of the empirical research. In the second section, a discussion of brand attitude is given, followed by a discussion of the brand associations in section 2.3. In the following sections, from 2.4 to 2.8, the moderators will be clarified and discussed regarding their influence on the relationship between brand attitude and brand personality associations. The hypotheses underlying the conceptual model are presented in the different sections. Therefore, this chapter will give an answer to research question 1.

2.1 Introduction & Conceptual model

In the field of marketing many research is done on brands and brand equity (Krishnan 1996). Brand equity focuses on the response of a consumer towards the brand and is defined as the brand associations held in memory by consumers (Keller 1993). One type of brand associations, which could be used to differentiate and personalize a brand, is brand personality associations. Brand attitude is conceptualized as the overall evaluation of a brand (Mitchell and Olson 1981, Río et al. 2001) and is important since it often forms the foundation of consumer behaviour (Keller 1993). Therefore, it is important to see how brand personality associations influence brand attitude, which is an indicator of consumer behaviour, in order to get a better understanding of consumer response towards brand personality associations.

(12)

According to Aaker (1991, 2008), brand associations, brand loyalty, brand awareness (familiarity), perceived quality (service quality) and other proprietary brand assets together form brand equity. These five pillars have causal interrelationships (Aaker 1991) and therefore could moderate the effect of brand personality associations on brand attitude. In order to account for these causal interrelationships, loyalty, brand familiarity and service quality are included as moderators in the conceptual model. The other proprietary assets are not included in the model since they relate more to the company than to the customer. These include assets as patents, trademarks and channel relationships (Aaker 1991). In this research the respondents are assumed to be aware of the brand, since the respondents are customers of the Friesland Bank. Therefore, brand familiarity will be applied in this research instead of brand awareness (David and Aaker 2008, Srinivasan et al. 2005). Satisfaction plays an important role in the foundation of customer loyalty (Jones 1995, Lovelock et al. 2008). Therefore, customer satisfaction will also be included as a moderator. One way to increase loyalty and retain customers, even if they are dissatisfied, is by raising switching costs (Aaker 1991, Lovelock et al. 2008). If the switching costs out raise the benefits from switching, customer loyalty will increase (Aaker 1991, Lam 2004). However, this could negatively affect the attitudes towards the brand (Lam 2004, Lovelock et al. 2008), which indicates why switching costs should be taken into account in the conceptual model.

The conceptual model of figure 2.1 describes the aim of this research, which is investigating the influence of brand personality associations on brand attitude and underlying structures, with respect to the proposed moderators. In the following sections the different elements of the conceptual model will be discussed.

2.2 Brand attitude

In the field of marketing considerable research has been done on attitudes, the importance of attitudes in understanding consumer behaviour and consumer response towards brands (Keller 2007). Brand attitude is one of the elements of brand equity, which focuses on the response towards the brand. As mentioned in the introduction of this chapter, brand equity can be defined as the brand associations held in memory by consumers (Keller 1993). In his brand equity concept Keller (1993) classifies three types of associations according to their level of abstraction: attributes, benefits and attitudes. Brand attitude shows the highest level of abstraction among these categories, since the attitude associated to the brand is based on underlying attributes and benefits (Keller 1993).

(13)

(Kotler 2009). Attitude is defined as a person's overall favourable or unfavourable evaluation of a concept, an affective response involving general feelings of liking or favourability (Fishbein and Ajzen 1975, Mitchell and Olson 1981). The attitude concept of interest in this research is brand attitude. In theory many definitions of brand attitude have been given, however most definitions of brand attitude are similar to the definition of Keller (1993, 2003). Keller (1993, 2003) defines brand attitude as what could be described as the overall evaluation that expresses how much we like or dislike a specific brand. These evaluations towards a concept are formed by repeatedly pairing unconditioned stimuli with a positive or negative evaluated association (Mitchell and Olson 1981). This positive or negative association might become related to the brand and might cause the brand attitude to change. According to Fishbein and Ajzen (1975) and Hoyer and MacInnis (2010) attitudes are learned and based on past experiences with the brand. Regarding these definitions, brand attitude can be influenced by the presentation of the association; the better the presentation of the association, the better the attitude towards the brand (Mitchell and Olson 1981).

A good brand attitude is important since it often forms the foundation of consumer behaviour (Keller 1993, Keller 2003, Mitchell and Olson 1981). Therefore, brand attitude can serve as an indicator for consumer behaviour. Agarwal and Rao (1996) and Yoo and Donthu (2001) argue that brand attitude can function as a surrogate for brand equity, since it represents the brand knowledge in consumers’ mind.

Thus, brand attitude is based on all the associations in consumers’ mind and an important indicator for consumer behaviour. Therefore, brand attitude will be defined and applied in this research as the overall evaluation of a brand. In this research we speak of brand attitude as a summary and overall judgment and not as a single association, even though brand attitude could be seen as a single brand association.

2.3 Brand associations and brand personality associations

In this section brand associations and brand personality associations are discussed. First, brand association in general are clarified. Second, a clarification and discussion of brand personality associations is given.

2.3.1 Brand associations

(14)

brand equity (Aaker 1991). These constructs are affected by other characteristics of brand associations and by relationships among brand associations.

In brand equity theory many definitions of brand associations are given, Keller (1993) and Krishnan (1996) define brand associations as informational nodes linked to the brand that contains the meaning of the brand for consumers. These links suggest an association in a consumer’s mind. Keller (1993) defines his definition from his associative network model of brand knowledge. Aaker (1991) and many others (Krishnan 1996, Park and Srinivasan 1994, Yoo and Donthu 2001) define them more simple, as anything linked to a brand in the memory of a consumer. Brand associations are important for the consumers as well as for service providers. They help consumers to process, organize and retrieve information from memory in order to make product evaluations and choices (Aaker 1991, Keller 1993, Van Osselaer, Stijn M. J. and Janiszewski 2001). For companies they are used in order to differentiate, position and extend brands, create positive attitudes and feelings and to create a compelling reason to purchase (Aaker 1991, Kamakura and Russell 1993). Brand associations are created through a consumers learning process. Consumers learn these associations through direct and indirect experiences with a brand (Keller 1993, Krishnan 1996, Van Osselaer, Stijn M. J. and Janiszewski 2001). Direct experiences are consumers’ own experiences and contact with a brand, indirect experiences are communication expressions by some other source of information (Keller 1993, Van Osselaer, Stijn M. J. and Janiszewski 2001). Although both important, direct experiences have a greater impact on the learning process of associations than indirect experiences. Besides that, non-marketer actions are also more credible than market controlled actions (Krishnan 1996). These learned brand associations differ regarding their level of abstraction (Aaker 1991, Keller 1993). Regarding the definition that a brand association is anything linked to a brand in a consumer’s memory (Aaker 1991) it can be concluded that there are several types of brand associations. In their brand equity concepts Aaker en Keller distinguish several types of brand associations. Aaker (1991) defines eleven types of brand associations and Keller (1993) distinguishes three categories of brand associations divided into subcategories. An overview of both classifications of brand associations is depicted in figure 2.2.

2.3.2 Brand personality associations

(15)

Aaker Keller

Product attributes Price

Intangibles Packaging Non product related Customer benefits

Attributes User imagery

Relative price Product related usage imagery Use/application User/customer Functional Celebrity/person Benefits Experiential Lifestyle/personality Product class Symbolic Competitors Attitude Country/geographic area

Figure 2.2 classifications of brand associations according to Aaker and Keller.

(16)

to Biel (1993) brand personality is viewed as a central driver in the establishment of positive attitude and preference for a brand. Brand personality enables consumers to express themselves along several dimensions. Beside the identity function and self-expressive function, they also serve as a symbolic function. Consumers use the brand in line with their own personality traits, marketing programs should therefore be aimed at believing and recognizing of brand personality by consumers (Govers and Schoormans 2005). Nowadays companies can easily copy product attributes, therefore it is difficult for brands to differentiate (Van Rekom et al. 2006). Knowledge about the brand in terms of personality will help to achieve successful differentiation and is very valuable to build brand equity (Van Rekom et al. 2006, Aaker 1991, Keller 1993). Since consumers use brand personality associations to identify themselves with a brand, to express themselves and because marketers imply that brands have certain characteristics, it is likely that consumers map a wide range of brand personality associations.

(17)

previous studies, like the research of Aaker (1997), to derive the 24 brand personality associations and the related classification on the five dimensions (a more detailed overview of the theoretical foundation of their research can be found in section 3.2). Although different terminology is used (human drives instead of human traits and brand appeal instead of brand personality) both researches are comparable and show several similarities. Brand appeal is described as the degree by which a brand touches the fundamental human drives and therefore it could be seen as brand personality and in both researches brand personality associations are used to derive the central dimensions.

Hence, brand personality associations are important for consumers and marketers, since brand personality associations influence consumer behaviour. Especially when they create a positive brand attitude, a marker for consumer behaviour. Since there are many lists of brand personality associations published in theory and since investigating all these lists of brand personality associations would be out of the scope of this research, the research of Cramer and Koene (2010) is adopted in this research. Their research is measured from the consumer perspective and based on several previous studies; therefore it forms a good starting point for this research. These brand personality associations will be used to investigate the influence of brand personality associations on consumer behaviour by meaning of brand attitude. The foregoing is formulated in the following hypothesis:

Hypothesis 1: Brand personality associations positively influence brand attitude.

2.4 Loyalty

One of the key dimensions of brand equity, which have causal interrelationships, is loyalty (Aaker 1991). Consumers, who are satisfied with the brand they have purchased, are likely to show loyal behaviour by repurchasing the same brand. According to Yoo and Donthu (2001) loyalty is the tendency to be loyal to a brand, the intention to purchase the brand as a primary choice. However, according to Reichheld (2003) customer loyalty is much more than repeated purchases; it is the willingness of a customer to make an investment or personal sacrifice in order to strengthen a relationship, which will affect the profitability of a brand. Loyal customers affect profitability since they tend to purchase more over time, devote a larger share of their wallets, reduce acquisition, serving and marketing costs. Loyal customers also make recommendations about brands to friends, family and colleagues. Such recommendations are one of the best indicators of loyalty because customers put their own reputations at risk (Reichheld 2003).

(18)

brand personality associations and brand attitude. Many research has been done on customer loyalty in the field of marketing (Dick and Basu 1994) and also on the positive moderating effect of loyalty, for example in customer retention, customer share development, purchase behaviour and customer referrals have been investigated (Seiders et al. 2005, Verhoef et al. 2002, Verhoef 2003). However, investigating loyalty as a moderator on the relationship between brand personality associations and brand attitude is quite new.

According to Aaker (1991) and Keller (1993) loyalty occurs when favourable attitudes and beliefs result in repeated buying behaviour. However, loyalty can also influence the effect of brand personality associations since loyalty is based on previous purchase experiences. Therefore, it is logical to investigate if loyalty affects brand personality associations since loyal customers already have experiences with the brand and therefore formed their own thoughts about associations. This moderating effect can only be investigated among existing customers of the brand, like in this research, since customers have to be experienced with buying the brand in order to build some form of loyalty.

Thus, since loyalty is the attachment that a consumer has with a brand and loyalty is one of the drivers of brand equity which all have casual interrelationships, it is important to see if loyalty has a positive effect on the relationship between brand personality associations and brand attitude. Therefore, the following hypothesis is suggested:

Hypothesis 2: Loyalty positively enhances the relationship between brand personality associations and brand attitude.

2.5 Brand familiarity

Brand familiarity is an important dimension which can influence consumers’ processing of information (Campbell and Keller 2003). The better consumers know the brand, the better they will be able to process information about a brand and the better they will link associations to the brand since they know how the brand is positioned (Keller 1993, Simonin and Ruth 1998). The degree of familiarity with the brand can be classified as brand familiarity and is also one of the key dimensions of brand equity (Aaker 1991).

(19)

Despite the attention given to brand familiarity and the moderating role of brand familiarity, for example in advertising, price perceptions and brand extensions (Biswas 1992, Campbell and Keller 2003, Kent and Allen 1994, Völckner and Sattler 2006), no research has explored the moderating effect of brand familiarity on brand personality associations in relation to brand attitude.

Brand familiarity is an indicator of the amount of knowledge a consumer has about a brand, whereas brand attitude is a summary judgment and overall evaluation based on the brand related information. Thus, the more familiar consumers are with the brand the better the associations are understood. Therefore, it is reasonable that brand familiarity will have a positive influence on the relationship between brand associations and brand attitude positively. This can be hypothesized as: Hypothesis 3: Brand familiarity positively enhances the relationship between brand personality

associations and brand attitude.

2.6 Service quality

Perceived quality, or in case of a service provider service quality, is like loyalty and brand familiarity (awareness) one of the key dimensions of brand equity (Aaker 1991, Aaker 1996). In theory many definitions from different perspectives of service quality are known (Lovelock et al. 2008). Parasuraman et al. (1988), Aaker (1991), Gupta and Zeithaml (2006) and Keller (2007) define service quality as a consumer’s perception of the overall quality of a service, considering the intended purpose and alternatives or, according to Wakefield and Barnes (1996) and Lovelock et al. (2008), the evaluation of a firm’s service delivery. Despite the number of definitions, the moderating effect of service quality has not been explored.

Although service quality and satisfaction show many similarities and interrelations, in theory and practice they are used interchangeable and apart, in this research they are defined as two separate constructs. Service quality differs from satisfaction since service quality is a global judgment relating to the superiority of the service, an evaluation of value, a comparison with other service providers (Gupta and Zeithaml 2006, Pappu and Quester 2006), whereas satisfaction is related to the service delivery, does it meet or fall short of expectations, a comparison with the expected (Gupta and Zeithaml 2006, Pappu and Quester 2006, Parasuraman et al. 1988). A customer can be satisfied by the service provider in each transaction, however, the perceived service quality could differ over time.

Thus, as perceived quality is one the pillars of brand equity and is a measure of value, it is reasonable to say that an increase in perceived quality will have a positive effect on a consumer and therefore the following hypothesis is formulated:

(20)

2.7 Satisfaction

Attracting new customers is very expensive, therefore retaining customers can be a profitable strategy. One key component in retaining customers is the satisfaction of customer needs. Satisfaction is a diagnostic measure of loyalty and very useful in understanding how customers feel about a brand (Aaker 1991). Many definitions of satisfaction have been given, most definitions are defined as the overall evaluation, or judgment, of the performance of the brand based on prior purchase and consumption experiences (Anderson et al. 1994, Jones et al. 2000, Oliver 2010, Reichheld 2003). According to Gupta and Zeithaml (2006) and Pappu and Ravi (2006) these definitions show many similarities and in essence can be defined as: “the consumer’s judgment that a product or service meets or falls short of expectations, a comparison with the expected”. This could result in different outcomes. In the case that the expectations are met at the expected level, the customer will be satisfied, which could be named as simple confirmation. In case the real experience exceeds the customer’s expectations, the customer will be overwhelmed, a positive disconfirmation, and in the opposite case if the real experience is not fulfilling the customer’s expectations, the customer will be disappointed, a negative disconfirmation (Oliver 2010). An increase in customer satisfaction enhances the overall reputation of the firm, which indirect can influence consumer evaluations (Anderson et al. 1994). Although attitude and satisfaction both are evaluations/judgments of a brand, they are not identical. An attitude is relatively enduring and can be formed without purchase experience while satisfaction is based on purchase experience but an alteration in customer satisfaction can influence the attitude towards a brand. Despite the attention given to satisfaction (Taylor and Baker 1994) and the moderating role of satisfaction in service quality and purchase intention (Jones and Suh 2000, Taylor and Baker 1994) only one study to date (Pappu and Quester 2006) has examined the effect of satisfaction on brand associations. According to Pappu and Quester (2006) there is a positive relationship between satisfaction and brand associations.

Thus, as an increase in satisfaction can lead to a higher attitude and because of the positive relationship between satisfaction and brand associations, one can expect that customer satisfaction will have a positive effect on the relationship of brand associations and brand attitude. Therefore, the following hypothesis is proposed:

(21)

2.8 Switching costs

As mentioned in the previous section retaining customers can be a very profitable strategy since attracting new customers can be very expensive. One way to retain customers is the development of switching costs (Aaker 1991). These switching costs help to retain customers when fluctuations on service quality and satisfaction on the short term otherwise would lead to customer defection (Jones et al. 2000). According to Colgate and Lang (2001), Jones et al. (2000), Klemperer (1987) and Ping and Ping Jr. (1993) switching costs are the costs in terms of time, money, effort and psychological costs consumers face when changing from service brand. They define three different types of switching costs: transaction costs, learning costs and artificial costs. Transaction costs are the costs necessary to make the service possible, including search costs. Learning costs can be defined in terms of brand knowledge, a result of customizing the service, and therefore only applicable to the brand in use. Artificial costs are the costs made by the service provider to retain customers; these are loyalty programs, quantum discounts, etc. Besides that, there are also behavioural and psychological costs in terms of time and effort which will diminish the tendency of switching. According to Colgate and Lang (2001) and Ping and Ping Jr. (1993) customers tend to be loyal if the switching costs are high, the customers may remain loyal despite their dissatisfaction due to the perception that the switching costs outweigh the switching benefits. This could be seen as a form of obligatory loyalty and therefore have a negative effect on the influence of brand personality associations on brand attitude.

Out of the moderators in this research, switching costs are the most explored moderating effect, however with mixed results. Patterson and Smith (2003), Lam (2004), Bell et al. (2005) and Burnham et al. (2003) all investigated the moderating effect of switching costs however without success. Jones et al. (2000) and Yang and Peterson (2004) did find a moderating effect of switching costs, they found a moderating effect of switching costs on the relationship between satisfaction and loyalty. Yang and Peterson (2004) also found a moderating effect of switching costs on the relationship between perceived value and loyalty. The moderating effect of switching costs on the satisfaction-loyalty relationship was found to be negative (Jones et al. 2000, Yang and Peterson 2004), so an increase of switching costs would diminish the effect of satisfaction. The application of switching costs as a moderator on the relationship between brand personality associations and brand attitude is new. In line with Jones et al. (2000) and Yang and Peterson (2004), one would expect that switching costs diminish the positive effect of brand associations on brand attitude. The foregoing results in the following hypothesis:

(22)

The effect of brand personality associations Research design

3. Research design

This chapter gives an overview of the data collection and research methods applied in this research. The chapter starts with the process of data collection in section 3.1. In section 3.2, an overview of the survey measures and scaling procedures is given, followed by the scale evaluation in section 3.3. The chapter ends with the methods of analysis in section 3.4.

3.1 Data collection

Data was collected on the Dutch market through an online questionnaire. The choice for an internet based survey has several advantages (Malhotra and Birks 2007). The main advantage of the internet is the ability to acquire information fast and flexible. Moreover, it offers a convenient and user-friendly way of administrating a survey. Furthermore respondents can fill in the survey at their own time in an experienced environment. The easiness to complete the survey will contribute to more complete and reliable answers. The respondents were invited through e-mail, with a direct link to the survey. For each respondent the survey was hosted on a unique web address, in order to secure that only the selected customers of the service provider could fill in the survey. The survey starts with the evaluation of the brand personality associations, the independent variables, followed by the measurement of brand attitude, the dependent variable, and ends with measurement of the different moderators. Secondary data was collected from the database of the service provider of interest. The secondary data concerns data like gender, age and province and type of service concept. This data is used in order to give descriptive information of the research sample.

(23)

The consistency check of the data resulted in the deletion of 9 respondents, which led to a defined dataset of 238 respondents which is used for further analysis.

3.2 Survey measures and scaling procedures

In order to measure the constructs, scales are developed based on existing literature. An overview of the measurement items for the brand personality associations, brand attitude, switching costs, brand familiarity, service quality, loyalty and satisfaction can be found in table 3.1. In this table the measurement item(s), source, measurement scale(s) and Cronbach’s Alpha, if applicable, are given for each variable.

(24)

Variable Measurement item(s) Source Measurement Scale Cronbach’s Alpha

Brand Personality Associations

When you think of the Friesland Bank, how do the following aspects match the Friesland Bank?

Cramer and Koene (2010) (based on a number of studies, a more detailed description of these studies is given in this section) 1 = Extremely unlikely, 5 = Extremely likely n/a

AS01. Ambition AS07. Creative AS13. Independent AS19. Proud

AS02. Appreciation AS08. Decisive AS14. Influential AS20. Reliable

AS03. Capable AS09. Elegant AS15. Innovative AS21. Risk averse

AS04. Care free AS10. Energetic AS16. Passionately AS22. Status

AS05. Caring AS11. Healthy AS17. Playful AS23. Transparency

AS06. Connected AS12. Idealism AS18. Profit driven AS24. Unique

Brand Attitude

Please express your attitude towards the Friesland Bank. BA1. I think the Friesland Bank is ____.

BA2. I ____ the Friesland Bank.

BA3. I am ____ towards the Friesland Bank. BA4. I think the Friesland Bank is ____.

Gill et al. (1988)

1 = Very bad, 7 = Very good

1 = Very much dislike, 7 = Very much like 1 = Very unfavorable, 7 = Very favorable 1 = Very worthless, 7 = Very valuable

α = 0.934

Switching Costs

When you think of switching of your financial service provider how much do you agree/disagree with the following statements?

SC1. Switching banks is too much bother in terms of time and effort.

SC2. When switching banks I am concerned about the negative financial outcomes. SC3. When switching banks I feel locked in because of the products I have with my current bank.

Colgate and Lang (2001), Ping and Jr. (1993) 1 = Strongly disagree, 7 = Strongly agree α = 0.629 Brand Familiarity

Please express your familiarity towards the Friesland Bank. BF1. How familiar are you with the services of the Friesland Bank? BF2. How experienced are you with the services of the Friesland Bank? BF3. How knowledgeable are you with the services of the Friesland Bank?

Kent and Allen (1994), Machieit et al. (1993) 1 = Unfamiliar, 7 =Familiar. 1 = Inexperienced, 7 = Experienced. 1 = Not knowledgeable, 7 = Knowledgeable α = 0.856 Service Quality

Overall, what is the level of service quality you receive from the Friesland Bank? SQ1. The service quality offered by the Friesland Bank is?

SQ2. The service quality offered by the Friesland Bank is? SQ3. The service quality offered by the Friesland Bank is?

Wakefield and

Barnes (1996) 1 = Terrible, 7 = Great.

1 = Much worse than I expected, 7 = Much better than I expected. 1 = Not at all what it should be, 7 = Just what it should be.

α = 0.897

Loyalty L1. How likely is it that you would recommend the Friesland Bank to a friend or colleague?

(25)

construct, except for the brand personality associations, satisfaction and loyalty, a seven point scale was used in order to keep the survey consistent and convenient for the respondents. A pre-test among approximately 10 respondents has been conducted in order to test question content, wording, form & lay-out, question difficulty and instructions. Regarding the pre-test unexpected errors were identified and corrected, however no critical problems or comments on the survey were mentioned and therefore face validity for the survey items appeared to be appropriate. The original questionnaire, in Dutch, can be found in appendix B. Since the questionnaire mainly consists of questions with Likert scales, the data gathered by the questionnaire will be treated as interval data (Cooper and Schindler 2006).

3.3 Scale evaluation

(26)

the cumulative variance explained by all the factors should be greater 60% (Hair 2010, Malhotra and Birks 2007). The variance explained per factor is 17.86%, which exceeds the critical value by far; the cumulative variance explained by the three factors is 72.55%, which exceeds the critical value of 60%. Also the eigenvalue criteria are met, since the eigenvalue for the three factors are all greater than one. Another indication of a stable solution are the communalities of the measurement items which need to be greater than 0.5 (Hair 2010). The nine measurement items all meet this requirement. The measurement items, which should represent an individual construct, indeed show a high loading on a separate factor (see appendix C for the rotated factor loadings). This shows that the measurement items indeed represent the different constructs.

The constructs for brand familiarity, service quality, switching costs and for the dependent variable (brand attitude) are constructed through summated scales. These summated scales are calculated by measuring the average score on the items of each construct. The constructs will increase the interpretation, since the multi-item constructs are represented by one variable, and reduce the measurement error.

3.4 Methods of analysis

In order to test the proposed hypotheses several analyses will be performed. First several descriptive statistics will be given in order to give a description of the data. After the descriptive statistics, an explanatory factor analysis on the brand personality associations will be performed in order to identify if there are any underlying structures (latent constructs) among these brand associations. This will reduce the number of variables for the model in this research, since analysing the effect of the full set of 24 brand personality associations on brand attitude including the different moderators would be too time-consuming, the number of parameters to specify would explode and the interpretation of the different associations and moderators would decrease. As mentioned in the previous section the sample size should exceed a minimum number of 5 observations per variable, where 10 is recommended, in order being able to apply a factor analysis (Hair 2010). The minimum number is met by far and the recommended number is almost met; the number of observations is 238 while there are 24 brand associations to be analysed, indicating that there are almost 10 observations per variable. In order to give a better interpretation of the brand personality associations, summated scales are constructed for the factor outcomes, by calculating the average score of the brand personality associations of each factor. Moreover, this factor analysis will give an answer to the second research question.

(27)

associations on brand attitude will be tested. This gives an answer to the third research question. The influence of the brand personality associations on brand attitude can estimated by the following formula and will test hypothesis 1.

Where: BA = Brand attitude.

BPAC = Brand personality associations component.

C = Total number of components derived from the applied factor solution,

starting at one.

In order to test if the relationship is moderated by satisfaction, loyalty, service quality, brand knowledge and/or switching costs the previous formula has to be extended. The effect of these moderators will be determined by including these moderators as an interaction in the regression analysis. In order to determine the interaction between the brand personality associations and the moderators five additional models have to be estimated. These models will test hypotheses 2 to 6 and will give an answer to the fourth research question. The extended model to be estimated for each moderator of interest will be formulated as follows:

Where: BA = Brand attitude.

BPAC = Brand personality associations component.

C = Total number of components derived from the applied factor solution,

starting at one.

Moderator = Moderator of interest.

(28)

The effect of brand personality associations Results

4. Results

This chapter will give a presentation of the results of the performed analyses. The chapter starts with some descriptive information of the respondents and the data in section 4.1. In section 4.2, the results of the factor analysis on the brand personality associations are given. In section 4.3, the results of the ordinary least square regression (OLS) analysis are presented. Finally, the results of the moderator analyses are shown in section 4.4.

4.1 Descriptive statistics

In this research 247 respondents participated, after a consistency check of the data 238 remained for the analysis. Of the 238 respondents 150 (63%) are males and 88 (37%) are females. The average age of the participants is 40 (39.65) years, most of the respondents are relatively old (see table 4.1). Most of the respondents are from the province of Friesland, which can be explained by the origin of the brand. The distribution of the respondents among the service models, classified by the service provider, also gives a good representation of the participants. In table 4.2 the service models are displayed, retail is the biggest service model among the participants, followed by advice and top advice. These findings are in line with the research population.

In table 4.3 the means and standard deviations of the measurement items are given, these numbers give an indication of the distribution in the measurement items. The means of the brand personality associations are mostly centred around 3.5. The standard deviations for the personality associations are relatively high for a 5-point Likert scale. Loyalty and satisfaction have a high mean and relatively normal standard deviation regarding the 11-point Likert scale applied for these items. These items also show some differentiation, however with relatively high answers. The means of the items for brand familiarity, service quality and switching costs are relatively low, they vary between 3.5 and 5.0, however have a relatively high standard deviation for the 7-point Likert scale applied for these items. Brand attitude has a slightly higher mean, but slightly lower standard deviation. Thus, some of the items are being evaluated relatively low, where others are evaluated relatively high. Based on the means and standard deviations one can concluded that there is enough degree of differentiation in the individual answers.

Age-category Frequency Percentage Service model Frequency Percentage

20 - 30 years 34 14.30% Retail 169 71.00%

31 - 40 years 76 31.90% Advice 56 23.30%

41 - 50 years 128 53.80% Top advice 13 5.50%

Total 238 100% Total 238 100%

(29)

Variable Mean Std. Deviation Variable Mean Std. Deviation

Ambitious 3.42 0.816 Independent 3.72 0.817

Appreciation 3.67 0.743 Influencable 2.93 0.732

Capable 3.66 0.757 Innovative 2.97 0.881

Carefree 3.07 1.162 Passionately 3.32 0.810

Caring 3.49 0.836 Playful 2.39 0.973

Connected 3.47 0.913 Profit driven 3.33 0.689

Creative 3.03 0.876 Proud 3.81 0.818

Decisive 3.39 0.897 Reliable 3.95 0.792

Elegant 3.16 0.723 Risk averse 3.65 0.643

Energetic 3.31 0.823 Status 3.34 0.800 Healthy 3.72 0.705 Transparency 3.48 0.783 Idealism 3.15 0.770 Unique 3.37 0.957 Attitude 1 4.94 1.099 Satisfaction 7.78 1.857 Attitude 2 4.86 1.116

Attitude 3 5.00 1.243 Service Quality 1 4.88 1.295 Attitude 4 4.86 1.107 Service Quality 2 4.27 1.213

Service Quality 3 4.17 1.002

Brand Familiarity 1 4.45 1.407

Brand Familiarity 2 4.36 1.413 Switching costs 1 4.11 1.793 Brand Familiarity 3 3.93 1.337 Switching costs 2 3.55 1.875

Switching costs 3 3.71 1.888

Loyalty 7.28 2.231

Table 4.3 means and standard deviations of the measurement items

4.2 Factor analysis

In order to reduce the total number of 24 brand personality associations applied in this research, an explanatory factor analysis is performed. The outcome of this factor analysis will give an indication of the underlying structure of the brand personality associations. This structure will be used as input for the subsequent OLS regression and moderators analyses. Summated scales will be applied in order to represent the underlying structure of the brand personality associations. Reducing the number of brand personality associations will increase the interpretation of OLS regression models. Principal component analysis is applied as factor method, since the main goal of the factor analysis is to reduce the number of brand personality associations. Principal components analysis is the most suitable method in this case, since it will determine the minimum number of factors that will explain the most variance in the data for the use in the subsequent OLS regressions models (Hair 2010). An orthogonal rotation method will be applied since the outcomes of the factor analysis will be used in subsequent analyses. Applying an orthogonal method will result in an uncorrelated factor solution, which is desirable for the subsequent OLS regression models (Hair 2010).

(30)

results imply that factor analysis is appropriate. The sample size requirements, already discussed in section 3.3, are met by far.

In order to determine the number of factors the following criteria are used: the eigenvalue of the factors should be greater than 1.0, the percentage of variance explained per factor should be greater than 5.0% and the cumulative variance explained by all the factors should be greater 60% (Malhotra and Birks 2007). The table of the total variance explained, necessary in order to check these criteria, can be found in appendix D1. The eigenvalue of the fifth component is 1.111 and exceeds 1.000, this indicates a factor solution of 5 components. The percentage of variance explained per factor indicates a component factor solution, since the variance explained per factor is 5.67% for a 3-component factor solution. The fourth and fifth factor do not meet this criterion, they are slightly below 5.00% (4.80% and 4.62%). The cumulative variance criterion also indicates a 5-component factor solution, similar as the eigenvalue criterion. For a five factor solution the cumulative variance explained by all the factors (63.35%) exceeds 60%. Regarding these criteria a 5-component factor solution will be applied in this research.

In order to have a stable solution the communalities of these brand personality associations need to be bigger than 0.5 (Hair 2010). All the 24 brand personality associations meet this requirement except for the brand personality association healthy, this brand personality association has a communality of 0.383, and overall these communalities show a stable solution. Interpretation of the rotated factor component matrix (see appendix D1) indicates that the brand personality association healthy, has besides a low communality, also a low factor loading. Together with ambition, these associations do not have a significant loading on one of the factors. Another problem occurred due to cross loadings of the brand personality associations reliable and energetically. Different rotation methods have been applied in order to solve these problems, however without any improvements. In order to solve these problems these four brand personality associations will be eliminated. Regarding the elimination of these four associations, a new factor model is specified based on the remaining 20 brand personality associations.

(31)

component factor solution, since the variance explained per factor is 5.36% for a 5 factor solution. The cumulative variance criterion also indicates a 5-component factor solution, for a five factor solution the cumulative variance explained by all the factors exceeds 65.23% which exceeds the 60%. Based on the new factor model, a 5-component factor solution will be applied in this research. The communalities of the new factor model are all above 0.5, which shows that the new solution is more stable than the initial solution. Interpretation of the rotated factor component matrix (see appendix D2) shows that all the brand personality associations have a significant loading on one of the factors, except for passion. Again, different rotation methods have been applied in order to solve this problem, however without any results. Therefore, passion will also be eliminated in order to solve this issue. Because of the elimination of this association, a final factor model is specified based on the remaining 19 brand personality associations.

For the final factor model also the same measures and criteria like previous two factor models are used to examine the final factor model and to indicate the improvements of the final factor model with respect to the previous factor models (see appendix D3 for the total variance explained table). The final factor model has a KMO measure of 0.908, which is slightly lower than the previous two models but still far above the lower limit of acceptability 0.6 (Hair 2010). The Bartlett test of sphericity of the final factor model tested significant (p = 0.000), both results imply that factor analysis is still appropriate. The eigenvalue of the fifth component is 1.067 and exceeds 1.000, indicating a factor solution of 5 components. The variance explained per factor and cumulative variance explained both slightly increased. The variance explained per factor increased to 5.61% and the cumulative variance explained increased to 65.80%, for a 5-component factor solution. Based on the examination of the final factor model, a 5-component factor solution will be applied in this research. The communalities of the final factor model are all above 0.5, which demonstrates that the final solution is a more stable one than the initial solution. Interpretation of the rotated factor component matrix (see appendix D3) shows that all the brand personality associations have a significant loading on one of the factors without any cross loadings. Finally, these significant loadings are used to label the five components. In table 4.4 an overview of the principal components, their labels and their corresponding loadings is given.

(32)

associations which give a description of the power related to a brand. The last component is about the playfulness and the relaxation of a brand, and can be described as the relaxing factor of brand. As for the multi-item constructs in section 3.3, the Cronbach’s alpha statistic is measured in order to check the reliability, or internal consistency, of these components. Table 4.5 contains an overview of the brand personality associations and the Cronbach’s alpha for each brand personality associations components.

The Cronbach’s alpha for each of the brand personality associations components exceeds the lower limit of acceptability of 0.6 (Malhotra and Birks 2007) indicating that these components are internal

Component

1) Social 2) Development 3) Trust 4) Power 5) Relaxed

Connected 0.814 0.250 0.198 0.063 0.080 Energetic 0.538 0.435 0.039 0.345 0.258 Idealism 0.577 0.206 0.023 0.187 0.390 Transparency 0.794 0.159 0.156 0.119 0.067 Capable 0.568 0.222 0.474 0.227 -0.096 Caring 0.801 0.169 0.183 0.082 0.198 Creative 0.369 0.681 0.170 0.206 0.087 Elegant 0.252 0.505 0.050 0.438 0.248 Unique 0.200 0.542 0.481 0.063 0.199 Innovative 0.260 0.792 0.153 0.145 0.095 Appreciation 0.400 0.161 0.554 0.491 -0.054 Independent -0.010 0.454 0.665 0.066 -0.026 Risk averse 0.339 -0.179 0.640 0.158 0.088 Proud 0.130 0.267 0.657 -0.033 0.298 Influential 0.287 0.250 -0.049 0.661 0.136 Status 0.197 0.101 0.299 0.676 0.081 Profit driven -0.053 0.054 0.041 0.710 0.149 Playful 0.208 0.194 -0.068 0.251 0.740 Carefree 0.099 0.053 0.360 0.120 0.775

Table 4.4 Factor solution of the final factor model.

Component Brand personality associations Cronbach’s Alpha Social AS03. Capable AS10. Energetic

AS05. Caring AS12. Idealism AS06. Connected AS23. Transparency

α = 0.879

Development AS07. Creative AS15. Innovative AS09. Elegant AS24. Unique

α = 0.790 Trust AS02. Appreciation AS19. Proud

AS13. Independent AS21. Risk averse

α = 0.716 Power AS14. Influential AS18. Profit driven

AS22. Status

α = 0.610 Relaxed AS04. Care free

AS17. Playful

α = 0.624

(33)

checked if deleting of the items would increase or decrease the Cronbach’s alpha, none of the items is deleted, since deleting one of the items did not increase the Cronbach’s alpha of the brand personality associations components. Summated scales are constructed for the five components, by calculating the average score of the brand personality associations of each component. These summated scales will be used as input for the brand attitude model in the subsequent analyses.

4.3 Ordinary Least Square Regression Analysis

Ordinary Least Square Regression Analysis (OLS regression analysis) is used to determine the influence of the brand personality associations components on brand attitude and to see if one of the moderators influences this relationship (section 4.4). In order to perform the ordinary least square regression analysis, first the assumptions of the multiple regression analysis have to be tested. Secondly, the brand attitude model will be estimated in order to test the first hypothesis. In section 4.4 the interaction effects of the moderators are included in the model, in order to test hypotheses two to six.

4.3.1 Testing the OLS assumptions

The analysis of the assumptions regarding the OLS regression analysis involves testing the following assumptions: linearity of the relationship, constant variance of the error terms, independence of the error terms and normality of the error term distribution. In addition, also multicollinearity will be discussed. Testing these assumptions and addressing the violations in the best way possible, will give a better representation of the brand attitude model. First, the normality assumption is tested since violation of this assumption many times contributes to violations of other assumptions. Therefore, addressing possible violations of normality may assist in meeting the other assumptions of the OLS regression (Hair 2010).

4.3.2 Normality of the error term distribution

The assessment of the normality assumption will be done by a visual and statistical check on the residuals.