1
Development of a Scale for Customer Brand Engagement
Master Thesis, Final Version Tim van der Wal
5815430
29th of January, 2015
MSc. Business Studies – Marketing
Amsterdam Business School, University of Amsterdam Supervisor: Dr. Karin Venetis
2 Table of Contents Abstract ________________________________________________________________________ 4 Introduction _____________________________________________________________________ 5 Research Objectives _____________________________________________________________ 6 Research Method _______________________________________________________________ 7 Chapter 1: Literature Review ________________________________________________________ 8 1.1 Engagement and its dimensions _________________________________________________ 8 1.2 Customer Engagement ________________________________________________________ 9 1.3 Customer Brand Engagement __________________________________________________ 13 1.4 Dimensions of Customer Brand Engagement ______________________________________ 14 1.5 Relational constructs surrounding Customer Brand Engagement ______________________ 15 Chapter 2: Scale development and validation _________________________________________ 21 2.1 Initial scale development _____________________________________________________ 21 Chapter 3: Results _______________________________________________________________ 24 3.1 Pretest ____________________________________________________________________ 24 3.2 Survey procedure and descriptive statistics _______________________________________ 24 3.3 Testing for Reliability ________________________________________________________ 25 3.4 Confirmatory Factor Analysis __________________________________________________ 26 3.5 Testing for validity ___________________________________________________________ 28 3.6 Common Method Bias ________________________________________________________ 32 Chapter 4: Discussion _____________________________________________________________ 33 4.1 Research objectives 1 and 2 ___________________________________________________ 33 4.2 Discussion of the results of the scale development__________________________________ 34 4.3 Limitations _________________________________________________________________ 36 4.4 Avenues for further research __________________________________________________ 37 Chapter 5: Conclusion ____________________________________________________________ 39 Reference list ___________________________________________________________________ 41 Appendices _____________________________________________________________________ 45 Appendix 1: Initial scale items _____________________________________________________ 45 Appendix 2: Pretest _____________________________________________________________ 47 Appendix 3: Survey _____________________________________________________________ 50
3
List of figures and tables
Table 1.1 - Key constructs related to CBE 17
Table 3.1 - Model fit summary 27
Figure 3.1 - Final CFA Model for CBE including factor loadings and covariances 28 Table 3.2 - Means, SD’s, Factor Loadings, Factor Alphas and t-values for CBE dimensions 29 Table 3.3 - Means, SD’s, Correlations and validity measures for CBE scale 30 Table 3.4 - Validity Measures for CBE dimensions and Involvement 31
Table 3.5 - Common Method Bias differences for CBE model 32
Acknowledgements
First of all I would like to thank my supervisor Dr. Karin Venetis for her invaluable feedback and her endless patience. I would also like to give my thanks to all my friends and my family for helping with feedback, motivation and support. Especially I would like to thank Rianne and my parents for their constant love and support in finishing this thesis.
4
Abstract
Brands are engaging more and more with their customers. Different from simply interacting (buying and selling), customers are creating deeper connections with brands through campaigns or social media. This, in short, is known as Customer Brand Engagement or CBE. Despite substantial interest in the topic, consensus about CBE, its dimensions and a quantitative scale for measurement still eludes the academic literature.
In this research the construct of CBE is provided with a scale that allows for quantitative research. Firstly the need is established for the construct of CBE in previous academic literature and the disagreement among authors regarding its dimensionality is assessed. Building on previous qualitative research by Hollebeek (2011a, b; 2013) 40 items were tested in a survey among 112 respondents which was conducted online as well as offline. Confirmatory Factor Analysis was used to come to a 12-item scale for CBE which consists of three dimensions. The scale withstood all
measures of reliability and validity. However, some limitations appear as sampling error and common method bias.
This thesis contributes to academic knowledge by providing a statistical basis for the three-dimensional concept of CBE and a scale with which this phenomenon can be tested in a statistical model. The managerial applications are, among others, to see if a brand’s campaigns are engaging with its customer and if so, on which dimension are they most prolific?
5
Introduction
In today’s fast and rapid changing world it is increasingly difficult to differentiate yourself from your competitors. Therefore, businesses are looking for new ways to find out what their customers are looking for in their offering. When a relationship can be created in which your customer is satisfied, they will remain your customers and may even spread the word to their friends and colleagues and so expanding the client base of your business. One way of accomplishing this is by using what is called engagement and its marketing-discipline counterpart customer engagement.
The Marketing Science Institute (MSI) has made customer engagement one of their research priorities in 2006-2012 (MSI, 2010). This was done to generate more academic interest as practitioners are embracing the concept, but academic theory and underlying concepts are lagging behind which makes for a limited scope of CE (Verhoef et al., 2010).
Businesses are adopting the notion of engaging with their customers. Where they previously saw their customers as passively receiving marketing incentives, the current view fixates on customers proactively participating in the value-creation processes of the organization (Hollebeek, 2013), even if there is no transaction involved. This is illustrated by Lay’s Great Britain in their contest to develop a new flavor for their potato chips. The winner received a cash prize and one percent of the turnover of the new flavor introduced (Verhoef et al, 2010). Even though the customers did not need to purchase a bag of potato chips, Lay’s received tons of submissions and new product ideas and because of the success of the campaign it was also launched in Australia and the Netherlands (Marketingonline.com). Firms are adopting customer engagement (CE) to increase profitability and awareness since measures as customer satisfaction do not seem to be the only way to influence brand loyalty (Bennett & Rundle-Thiele, 2004) which results in more sales.
Even though there seems to be academic and practitioner congruence about the phenomenon of CE, how it can be measured or what the underlying concepts are is still somewhat of a mystery. There are several issues among authors that make for muddy waters in CE research. One
6
of these incongruences is whether or not CE is a one-dimensional, two-dimensional or multi-dimensional concept and if it is multimulti-dimensional, which dimensions make up CBE?
The broader concept of engagement is predominantly found to be multi-dimensional which consists out of cognitive, emotional and behavioral dimensions. This has been found in social sciences as student- and employee-engagement (Schaufeli, Salanova et al., 2002; May et al., 2004) and reshaped into the multi-dimensional construct of customer-brand engagement by Hollebeek (2011a). As an emerging area in the academic marketing literature (Hollebeek, 2013), customer-brand engagement (or CBE) has been found to be a sub-type of CE. Where CE pertains to the interaction between customer and organization (Vivek, 2012), CBE is further narrowed to the relationship between the customer and a brand.
Even though there has been substantial interest in the topic, a scale to measure CE has not reached consensus and a scale for CBE has still to be developed to provide generalizable results (Hollebeek, 2011a, 2011b; Brodie et al., 2012). The focus of this thesis is to address this gap and further develop a scale to quantitatively test for CBE based on qualitative research done by Hollebeek (2011b).
ResearchObjectives
In this thesis, the research objectives are:
1. To establish the need for the construct of customer brand engagement.
In the current literature (customer brand) engagement is gaining attention and therefore the need for a general understanding of the construct needs to be established because the current debate on the subject remains nebulous.
2. To determine its dimensions.
Due to different opinions of authors in the literature, the use of dimensions is cause of debate. In theoretically supporting the notion of dimensions in customer engagement, clarity may be provided towards developing a universal scale and understanding.
7
3. To develop a scale of customer brand engagement.
Research concerning customer engagement has, with the exception of Appelbaum (2001) and Vivek (2009), been theoretical or qualitative in nature. The lack of research in the field of customer brand engagement also requires a quantitative measure to make generalization possible among results.
Research Method
To come to a scale first the literature has been reviewed to gain an overview of the extant literature concerning engagement (CE) and CBE. This also consists of the theoretical underpinnings and linkages with other marketing concepts. Building on this overview the dimensions are determined that make up CBE.
From these dimensions and the literature a scale is constructed and pretested. This scale is adapted from extant literature taken from qualitative research done by Hollebeek (2011b) and from the quantitative scale on consumer engagement by Vivek (2009). The scale contains items, both positively and negatively valenced, on a 5-point Likert scale ranging from ‘strongly agree’ to ‘strongly disagree’
The scale is administered through the internet surveytool Qualtrics. In this survey respondents self-select a brand with which they feel highly engaged. In the introduction of the survey a short description of engagement is given to avoid confusion amongst respondents.
The first chapter provides a literature overview of engagement, customer engagement and customer brand engagement. Chapter two will comprise the dimensionality, possible antecedents and consequences of CBE that are found among the various articles. Chapter three will comprise the initial scale development and the validation of the scale. Chapter four will discuss the results of the analyses and chapter five will provide the discussion of the results, the research’s limitations and possible avenues for further research. This is followed by the conclusion. The references and appendices can be found at the end of this thesis.
8
Chapter 1: Literature Review
Many authors and the Marketing Science Institute (MSI, 2010) have called for contributions concerning engagement and its further sub-forms such as customer engagement and customer brand engagement but so far the level of new contributions remains low due to several issues. These are the conceptual nature, the dimensionality and providing a measure (Hollebeek, 2011b). Because each one is inextricably linked to the other this literary review will provide an overview of the different views on and dimensionality in engagement and its sub-forms to arrive at a sound conceptual nature for CBE and its dimensions before arriving at a scale for marketing purposes.
1.1 Engagement and its dimensions
To come to an understanding of customer brand engagement it is necessary to describe the overarching concepts of engagement and customer engagement. In this section a literary review of engagement will be given, which is the broadest type in engagement literature.
In reviewing the academic literature there were several articles found which describe engagement, mostly in the fields of psychology, sociology, political science and organizational behavior (Brodie et al., 2011). By looking at the construct of engagement through different lenses, different approaches and highlights are presented which makes for a large variety and little consensus among researchers about the true nature of engagement.
The first mention of engagement in a business related context is found in an article by Kahn (1990) in which he presents personnel engagement in the workplace. This employee engagement is about expressing themselves physically, cognitively and emotionally in their work roles (Kahn, 1990). In defining the construct as such this is a clearly multi-dimensional phenomenon. Apart from Kuhn’s article, employee engagement has been further researched in organizational behavior literature (Saks, 2006; Frank, Richard & Taylor, 2004). Civic engagement (Jennings and Stoker, 2004; Mondak et al, 2004) has been researched in the sociology discipline, state engagement (Resnick, 2001) in political science, student engagement (Bryson & Hand, 2007; Hu 2010) in educational psychology
9
and in the field of psychology social (Huo et al., 2009), task (Matthews et al., 2010) and occupational engagement (Bejerholm and Eklund, 2007) have been researched.
So what constitutes engagement? Brodie et al. (2011, p.254) state that: “At the meta-level, ‘‘engagement,’’ as a form of social, interactive behavior, has been characterized as a transient state occurring within broader relevant engagement processes developing over time.”. In this global definition two things stand out; firstly that engagement has a social and interactive nature which implies that the engager and the engaged have a reciprocal relationship and secondly that this relationship evolves over time. This is underlined by Bejerholm and Eklund (2007,p.21): “Occupational engagement describes (…) the ability to move around in society and interact socially, implying that occupational engagement occurs over time”.
A point of interest is the level of dimensionality among the articles. Dimensionality is important because one dimension is straightforward and easy to apply but it fails to encompass the larger scope of the concept of engagement (Brodie et al., 2011). Therefore it is vital for completely understanding engagement and its sub-forms to provide dimensions. Some articles have a one-dimensional perspective on engagement; Achterberg et al. (2003) describe social engagement and Resnick (2001) state engagement as a behavioral concept. By far the largest part of articles researched see engagement as multi-dimensional and use three dimensions (cognitive, emotional and behavioral) to explain engagement.
1.2 Customer Engagement
The concept of engagement is further condensed into customer engagement, which is a concept found in the marketing literature. Two avenues of research in marketing are very much intertwined with this concept. The first one is relationship marketing (RM). RM is “attracting, maintaining and –in multi-service organizations- enhancing customer relationships”(Berry 1983, p.21). This is further developed by Morgan and Hunt (1994) into relational exchanges because not all relationships have a customer, such as partners in a strategic alliance or internal marketing within a firm.
10
As pointed out by Vivek et al. (2012) customer engagement can further help relationship marketing by focusing on the attracting, maintaining and enhancing of the customer relationships as RM literature has primarily focused on the retaining of an organization’s existing customers where possible and potential customers of the organization have slid to the background. The difference between the latter two is that possible customers are still interacting with the corporation and are not in the decision-making stage and potential customers are considering to do business.
Vivek et al. (2012) mention several ways how customer engagement is helpful in understanding and transforming these possible and potential customers into existing customers.
First, businesses try to expand by spending billions on getting prospective customers to engage with their products even if they are not immediate prospective clients. The second follows on the first; many campaigns are not designed to bring new customers in but rather to create awareness and engagement with the organization among interested parties. Third, existing, possible and potential customers exchange experiences among themselves without interference by the organization. For example, they can engage with the organization’s offerings at a friend’s house or look at reviews online.
Through the lens of RM, these types of doing business would not seem like relationship marketing. Customer engagement rather invests in the long term and in doing so creates positive connections with the organization resulting in prospective customers.
The other avenue of theory that laid the conceptual roots for customer engagement is the service-dominant logic as developed by Vargo and Lusch (2004). Opposite the long time accepted goods-dominant view which looks at the exchange of tangible goods, the service-dominant view is about exchanging intangibles, processes, skills and knowledge (Vargo and Lusch, 2004). This other kind of viewing logic is better integrated with services because services are not strictly goods or intangibles but an combination of processes, goods etcetera combined in an offering which creates value and “the focus to services is shifted from the means and the producer perspective to the utilization and the customer perspective” (Gummesson, 1995, pp.250-51).
11
The S-D logic is laid out in a set of ten foundational premises regarding marketing exchanges between customers’ experiences with stakeholders (Brodie et al. 2011). According to Brodie et al. (2011), four of the ten premises (Vargo and Lusch, 2008, p. 7) are relevant to customer engagement. Premise 6, “The customer is always a co-creator of value” relates to customer engagement in that the customer is a part of the value creation processes for the firm and this indicates an interactive relationship.
Premise 8 is “A service-centered view is inherently customer-oriented and relational”. Services have to have customer benefits and they are co-created so the focus for the firm has to be on the customer and his or her relationship with the firm.
Premise 9, “All social and economic actors are resource integrators” sees resource integrators as networks and since every customer is part of a network, value creation happens within a network. The three ways how customer engagement is helpful mentioned above also stipulate that networks are the context of customer engagement.
The final premise, number 10, reads “Value is always uniquely and phenomenologically determined by the beneficiary”. For every customer their value is unique through their own set of beliefs, experiences and context. Therefore value is very subjective and contextual when engaging with a product or service.
After evaluating these two avenues, relationship marketing theory and the service-dominant logic believe customers are proactive in engaging with the firm instead of passively receiving the information brought forward by the firm (Grönroos, 1997a; Vargo and Lusch 2008).
These two avenues provide the basis for a definition of customer engagement. Even though there are several other definitions (Hollebeek, 2011b; Vivek et al. 2012) the definition by Brodie et al. is the most exhaustive:
“Customer engagement is a psychological state that occurs by virtue of interactive, cocreative customer experiences with a focal agent/object (e.g. a brand) in focal service relationships. It occurs under a specific set of context-dependent conditions generating differing CE
12
levels; and exists as a dynamic, iterative process within service relationships that cocreate value. CE plays a central role in a nomological network governing service relationships in which other relational concepts (e.g., involvement, loyalty) are antecedents and/or consequences in iterative CE processes. It is a multidimensional concept subject to a context- and/or stakeholder-specific expression of relevant cognitive, emotional and/or behavioral dimensions.”(Brodie et al., 2011, p. 260)
In this definition the crucial elements are highlighted. Customer engagement is about interactive and co-creating experiences with an agent or object such as a firm or brand, that form a process to create value for both the customer and the firm. This is what sets customer engagement apart from other similar constructs as ‘involvement’ or ‘participation’ because these do not capture the level of interaction and co-creation as does customer engagement (Brodie et al, 2011). Verhoef and Lemon (2013) agree: “customer engagement behaviors go beyond transactions and specifically reflect customers’ behavioral manifestations that have a brand or firm focus, beyond purchase, resulting from motivational drivers” (Verhoef & Lemon, 2013, p. 12).
Customer engagement plays a central role in the network of relationships where other relational concepts are antecedents and consequences. Because customer engagement is by definition a relational concept it is linked to other constructs. Being in the center of processes makes it even more important to understand as measurements may be off if customer engagement is not considered. Furthermore, as mentioned earlier, the relationships of the customer itself are not strictly dyadic but they are engaged with numerous types of stakeholders and in doing so they create their own network of interactions and co-created value.
The question of who is engaged is broader than one might think. Vivek et al.(2012) describe the focus of engagement as being on the whole process of moments of interaction. If this is illustrated by the activity of buying a car, people who have the intent of interacting with Mercedes Benz are included in the focus, even when they have no means of buying a car. The process goes on to taking a test-drive, buying the car and subsequent post-purchase activities as check-ups. Note
13
that this focus is not only present in customer engagement relationships but in every one of the engagement-types named in the previous section.
The multidimensionality of customer engagement is not surprising as it also was mentioned with engagement above. The definition has added that these dimensions are important because of differences among context and stakeholders. In this way it is easier to distinguish and appropriate the right course of action for the specific customer. For example, imagine you’ve lost your job and you are sitting at home alone wondering what to do and you decide to drink a beer to take some pressure off. Your engagement with the beer will be different when you’re enjoying the same beer along with the music of your favorite band at a music festival, surrounded by your friends. The levels of customer engagement may be placed on a continuum ranging from ‘actively disengaged’ to ‘fully engaged’ (Bryson and Hand, 2007).
1.3 Customer Brand Engagement
Where customer engagement sees the object of the relationship to be an organization, product or brand (Brodie et al, 2011), customer brand engagement merely focuses on the customer-brand relationship. As Hollebeek is the only author to specify the object (e.g. the customer-brand) there is little known about theoretical linkages between customer engagement and the brand, this avenue for research is still wide open. Certain issues surrounding the concept have also emerged which will be discussed later.
The concept of customer brand engagement is defined by Hollebeek (2011a) as; ‘the level of an individual customer’s motivational, brand-related and context-dependent state of mind characterised by specific levels of cognitive, emotional and behavioral activity in direct brand interactions’. This definition highlights several topics that are consistent with the previous definitions regarding the customers’ psychological state (taken at a specific point in time), the context-dependent nature of engagement, multidimensionality and interaction.
What distinguishes customer brand engagement from other forms of engagement such as customer engagement (Vivek, 2009; Brodie et al., 2012) is the fact that engagement only pertains to
14
a customer’s brand interactions (Hollebeek, 2011a). As opposed to interactions with the firm or its offerings, customer brand engagement only focuses on the contact the customer has with the brand such as accessing the website of the brand or using the brand.
Furthermore, where engagement can be between all stakeholders, the focus of customer (brand) engagement is with the state of mind of the individual customer and his or her relationship with the brand and therefore it is seen from a Customer to Business (C2B) perspective. After all, brands are by definition a business tool and this helps business professionals understand the ways in which customers interact with the business and so they can align their business model and in providing the best service in their brands.
It is important to note that in the definition it is implicit that relationships form over time and are not reciprocal one-time exchanges but rather a series of them. The state of mind of a customer is made up over a series of interactions with the brand. CBE is also viewed to be positively valenced (Hollebeek,2011a) which in turn is thought to affect positive customer loyalty outcomes (Bowden, 2009; Patterson et al., 2006).
1.4 Dimensions of Customer Brand Engagement
Across all levels of engagement, there is a lack of consensus that it is a multidimensional construct. In several literary reviews regarding (customer) engagement in selected academic disciplines other than marketing (Brodie et al., 2011; Vivek et al., 2012; Hollebeek, 2011b), a wide range of different dimensionalities is observed. Ranging from one-dimensional (Resnick, 2001; Achterberg et al, 2003) to dimensional (Saks, 2006; Bryson and Hand, 2007). Within the multi-dimensional articles there is also large variation. Where Bejerholm and Eklund (2007) describe occupational engagement as a cognitive and behavioral construct, Huo et al. (2009) believe social engagement is an emotional and behavioral construct. Furthermore, London et al. (2007) explain educational engagement along cognitive and emotional lines. However, the three-dimensional view has gotten the largest following (Bakker et al., 2007; Bakker et al., 2008; Handelsman et al.,2005; Salanova et al., 2005; Schaufeli, Martinez et al.,2002).
15
The unclarity surrounding the dimensionality perceived in other disciplines is also found in the marketing literature. In his article regarding advertisement engagement, Heath (2007) describes engagement as being an emotional, one-dimensional construct. The general view however, is that engagement is seen as a three-dimensional concept consisting of cognitive, emotional and behavioral engagement (Bowden, 2009; Mollen and Wilson, 2010; Vivek et al., 2012). The dimensions may vary due to specific contexts or actors (Brodie et al, 2011) as is illustrated by Patterson et al.’s (2006) dimensions absorption (cognitive), dedication (emotional) and vigor/interaction (behavioral) regarding customer engagement.
1.5 Relational constructs surrounding Customer Brand Engagement
In the relationship marketing literature the concept of customer brand engagement is related to several other marketing constructs but is thought to be conceptually different (Bowden, 2009; Hollebeek, 2011a; Patterson et al., 2006).
Relationship marketing is based on the premise that customer-brand relationships are positive influencers of business outcomes such as customer retention and profitability (Palmatier et al, 2006). To arrive at a nomological network that is comprehensive in explaining these business outcomes, literary research was done to comprise a list of constructs mentioned in relation to customer engagement or customer brand engagement. In several articles (Bowden, 2009; Brodie et al., 2011, Hollebeek, 2011b; Vivek 2012) constructs or lists or constructs were found containing possible antecedents and consequences of customer (brand) engagement. Table 1.1 shows an overview of the key constructs, their definitions and the relationship to CBE found by Hollebeek (2011a, p.793-794). These constructs were found in multiple articles and are discussed below in more detail.
Involvement is the first construct and is seen as an antecedent to CBE. This seems logical because before engagement can be reached, some level of involvement is needed. CBE is more comprehensive in reflecting the nature of engagement; namely the interactive, co-creative experiences (Brodie et al., 2011). Involvement pertains to need satisfying abilities and therefore only
16
a customer’s cognitive abilities (Saks, 2006) where CBE also includes emotional and behavioral dimensions.
Interactivity is also a prerequisite for CBE but is distinctly different because interacting with a brand is behavioral. A few other authors use the name participation for the construct of interactivity (Brodie et al., 2011; Vivek et al., 2012).
Flow is closely related to the cognitive dimension or as Patterson et al. (2006) describe it, absorption. These concepts both rely on cognitive involvement where time passes quickly. However, May et al. (2004) see flow as a heightened, “peak” state of cognitive involvement where CBE is broader than just the cognitive dimension and also involves lower states of the cognitive dimension. Therefore flow can be seen as a possible antecedent of CBE.
Rapport is a potential CBE consequence in human/brand interactions. Rapport is establishing a true connection with the object, be it another human or a brand or organization. Rapport can be a consequence for customers because engaging with the brand will result in a better rapport in certain contexts (Brodie et al., 2011). For existing customers rapport may be an antecedent because a higher rapport will increase engagement levels and create a continuous loop over time.
Co-created value can serve, as with rapport, as a potential CBE consequence when human customers interact with the brand. Brodie et al. (2011) describe CBE states as a precursor for co-creating value through dynamic, iterative processes and will in turn provide co-created value.
Co-created value, interactivity and rapport are concepts that are of greater importance in service and/or Web 2.0 settings which are characterized by human interaction (Hollebeek, 2011a). Albert Heijn for example can use interactivity and rapport to form higher levels of engagement through service encounters with staff when doing grocery shopping. This will result in co-created value. However, these concepts are only interesting for brands where human interaction is part of the shopping routine of the customer. A brand with a lack of human interaction with the customer may find these constructs irrelevant or it may be made up by interaction in Web 2.0 settings as Facebook, such as Heineken does (Facebook.com/Heineken).
17
18
The construct of brand experience is a possible consequence of CBE. As Brakus et al. (2009) state, brand experience includes a behavioral state that is evoked by brand-related stimuli. However, it is lacking a motivational or emotional state. Because CBE is based on consumer’s needs and values this motivates consumers to a specific brand and this is where it differs from CBE. Brand experience is also reactive of nature, where CBE allows for proactive behavior which may result in co-created value. (Hollebeek, 2011a)
Another possible consequence of CBE is what is called perceived quality. This construct looks at superiority of a product from the consumer’s viewpoint (Zeithaml, 1988). Perceived quality can be particularly prominent in service- or co-created value situations. As perceived quality is influenced by lower level attributes which range from functional to emotional benefits (Zeithaml, 1988), this aligns with the concept of CBE. Gaining benefits due to brand engagement will likely result in higher perceived quality.
19
The constructs customer satisfaction, trust and commitment form the overarching concept of relationship quality (Dorsch et al., 1998; Ulaga & Eggert, 2006). These three constructs are seen as possible positive consequences of CBE. Even though there is no research found that links CBE to relationship quality, several other linkages have been found. When researching employee engagement, Saks (2006) found a positive relationship with job satisfaction and organizational commitment. Translating these findings to the current research, customer brand engagement may positively be related to customer satisfaction and commitment. Ganesan & Hess (1997) and Morgan & Hunt (1994) have found a positive relationship between consumer trust and commitment. Because of this relationship trust is also believed to be a consequence of CBE. Higher levels of engagement may, through the various dimensions of CBE, cause for increased relationship quality. The relationships above may also be reversed for existing customers. Customers with earlier established levels of the dimensions of relationship quality may, as described above, function as an antecedent for customer brand engagement (Hollebeek, 2011a).
The construct of customer value is also seen as a possible consequence of CBE. It has strong similarities with perceived quality as it is also based on the perceptions of customers (Zeithaml, 1988) with the difference being that value assesses the utility for the customer itself where quality is an evaluation of the product or service. This means that low quality products still can have high value, depending on the consumer’s needs. When CBE increases the utility of the customer, it is likely that customer value is also increased.
The final possible CBE consequence is brand loyalty. As discussed by Day (1969), brand loyalty needs to be viewed along two dimensions; behavioral and attitudinal loyalty. This aligns with some of the dimensions of CBE. Bowden (2009) believes that ‘true’ brand loyalty is reached when engagement provides strong bonds, as opposed to convenient repurchases that can be lost to competitors who come with a better offering which is called ‘spurious loyalty’ by Day (1969). CBE would present a better predictor of ‘true’ brand loyalty.
20
Because of the strong indications that the constructs of involvement is a consequence of CBE, this construct is taken into the survey. This is done to account for construct validity. Construct validity consists of convergent validity (are constructs somewhat similar) and discriminant validity (are constructs sufficiently different). Involvement has already been tested in a CE-scale (Vivek, 2009) and is believed to garner evidence for discriminant validity. The concepts of validity and the constructs mentioned above will be further explained in the research methodology.
21
Chapter 2: Scale development and validation 2.1 Initial scale development
The research methodology concerning scale development of customer brand engagement or CBE follows standardized guidelines. Scale development is an extensive process which has been thoroughly examined in the literature (Churchill, 1979; Gerbing & Anderson, 1988; Nunnally & Bernstein, 1994). In quantitative research, measuring a construct or attribute is done by using multiple-item measures because of limitations with single-item measures (Churchill, 1979).
Nunnally and Bernstein (1994, p.66-67) list 5 of these limitiations; firstly an individual item correlates poorly to the attribute. Second, each item tends to relate to other attributes than the one that is measured. Third, each item has a degree of specificity in the sense of not correlating with any general attribute. Fourth, an individual item is unreliable because people tend to rerate stimuli. This is averaged out when individual scores are summed up. Finally, attributes or constructs need a finer differentiation than can be accomplished by a single item. Such a multiple-item measure is called a scale (Gerbing & Anderson, 1988).
Scale development consists of several steps. Along the lines of Churchilll (1979) the three initial steps will be discussed briefly. The first step includes specifying the domain of the construct to come to an understanding what the construct is and is not about. By looking at previous definitions, researchers find what the construct entails and if they want to change the definition as it is known a good reason is needed. To come to such a definition extensive literature search is required.
After the domain has been specified a sample of items will be generated. This is done through qualitative research as literature search, interviews or focus groups. This research has to be done according to previously defined dimensions from the literature. It is imperative to use open discussion to promote more items to be generated. After an initial item pool is created it should be edited for double-barreled statements and wording before collecting the first data.
These initial qualitative steps have been taken by Hollebeek (2011b). An initial pool of items has been generated through interviews and focus groups and the current research converts the
22
initial items into a quantitative scale for customer brand engagement. Hollebeek (2011b) has come to three different dimensions: Immersion, passion and activation. After studying the literature these dimensions seem to overlap with the cognitive, emotional and behavioral dimensions found in the engagement literature. The items were stated in several keywords and descriptions and these have been adapted for survey research on a 5-point Likert scale ranging from completely disagree to completely agree. (the initial survey can be found in Appendix 1).
2.2 Method for CBE scale development and validation
The next step Churchill (1979) recommends step is further purifying the measure because a large part of measurement error comes from non-relevant items. Hinged on the assumption that all items will have some common ground, the first step here is performing a coefficient alpha calculation. A large alpha indicates that the item correlates well with the construct and low scores need to be eliminated. This will result in fewer items. The items will also be judged on wording and flow and corrected where necessary through a pretest sample. This likely will result in the dropping of some items and thus purifying the scale. This leaves us with several items which can be further developed into a scale for CBE.
The full survey will contain the remaining items for CBE and 5 bipolar items for involvement that are taken from Mittal’s (1995) Personal Involvement Inventory (MPII) (Vivek, 2009). New data will be collected online through Qualtrics, an online survey tool. The survey will also contain a small introduction where several examples of brands and engagement are mentioned to provide the people taking the survey with a complete comprehension of what CBE is. Several demographic questions will be asked about age, education, etc. After the data is collected, the sample will be viewed for missing data and normality.
Exploratory factor analysis (EFA) (Sprott et al. (2009); Vivek (2009)) will not be used because it is assumed that CBE will have three factors or dimensions. The items are checked for reliability.
23
Reliability describes whether the scale measures the same every time and validity pertains to measuring what the scale is supposed to measure (Nunnally & Bernstein, 1994).
The sample data will be used for confirmatory factor analysis (CFA) (Sprott et al. (2009); Vivek (2009); Strizhakova et al. (2008)) whereby validity of the scale is measured along two types of validity; discriminant and convergent validity. Discriminant validity assesses if the scale is different among factors or other scales and convergent validity assumes that theoretically linked scales are in fact related to each other (Vivek, 2009). CFA is also used to confirm the assumed three-dimensional model of CBE and internal consistency is checked (Churchill (1979); Delgado-Ballester et al. (2003).
Finally construct validity is also checked with the construct of involvement. As described above involvement is seen as an important antecedent of CBE. To be able to create levels of CBE, a level of customer interest (i.e. a behavioral aspect) is needed before specific customer-brand relationship levels can be attained (Hollebeek, 2011a; Mittal, 1995; Zaichkowsky, 1985) with respect to focal brands. It is expected to have overlap with CBE (convergent validity) but it also is sufficiently different to account for discriminant validity.
24
Chapter 3: Results 3.1 Pretest
In gathering enough responses of the survey for CBE it was vital for the construct to be clear for everyone participating in the sample. As the sample was designed as open for every consumer to take part, a pretest was necessary. After composing the initial pretest (see Appendix 2) this was submitted to 12 non-students to check for clarity, flow, spelling and grammar.
The questions themselves seemed straightforward as none were mentioned as unclear. This could not be said for the overall concept of CBE as there was much confusion about the concept. More explanation was needed and a table with examples was included. Also the flow was improved through a native English speaker who works a translator.
3.2 Survey procedure and descriptive statistics
The survey was conducted online using Qualtrics. A link was provided through mailing, social media and word-of-mouth. The full survey can be found in Appendix 3. The respondent was
presented with a cover page explaining the concept of CBE. This was followed by which brand they felt engaged with, which industry the brand belonged to and how long they’ve been engaged with the brand. Next were 12 items concerning cognition, 18 items about emotion and 10 items about behavior, totaling at a number of 40 items to be answered. These 40 items were measured on a 5-point Likert scale (Strongly disagree (1); Disagree (2); Neutral (3); Agree (4); Strongly agree (5)). Some items were negatively positioned to prevent data pollution. To measure involvement 5 items were included in the survey, also on a 5-point Likert scale. The survey was concluded with 5 questions about demographics, namely age, gender, education, occupation and nationality. 129 people started the survey online, of which 47 did not finish or were missing data and were excluded, leaving 82 responses.
25
As this was not sufficient for researching a scale for CBE, a printed version was filled in by 30 respondents with no missing data. This resulted in a total of 112 responses to be used for further research. After recoding and resolving missing data, no further deletions were made.
The sample consists of relatively young and highly educated people. In this respect the sample is not representative of the population as 75 percent was between 18 and 39 years and every respondent completed high school and more than 75 percent have a Bachelor’s degree or higher. The sample was predominantly Dutch (93.8%) and 50.9 percent of the sample was male, indicating a balanced sample on gender. Regarding CBE, nearly 90 percent of respondents were engaged with their brand for 3 years or more.
Apart from the sampling error discussed above, there may also be a response error
(Malhotra and Birks, 2003). Due to the difficult concept, it may be troublesome for people to think of a brand with which they actively engage because this is not asked of people on a normal basis. It may be hard to find the sort of information which is asked in the survey, resulting in a possible measurement error.
The full sample was checked for normality and no issues were found. After testing for reliability, a Confirmatory Factor Analysis was performed to reduce the number of items and to confirm the assumed dimensionality. After that convergent and discriminant validity is tested.
3.3 Testing for Reliability
As there were assumed to be three dimensions, reliability was checked for these dimensions to clean up the data before starting with the Confirmatory Factor Analysis. In IBM SPSS 20 a
reliability analysis was run for each dimension.
The Cognitive dimension had 12 items. The item-to-total correlations, the Cronbach’s Alpha and Alpha if item deleted were examined. The Cronbach’s Alpha was .821 for the twelve items, indicating a reliable scale as it was above 0.7 (Nunnally & Bernstein, 1994). The difference in the Alpha if item deleted column was never 0.2 or greater giving no reason for deletion.
26
Carmines and Zeller (1974) mention another way to assess reliability. When at least 50 percent of item-to-total correlations are between 0.30 and 0.70, the scale is assumed to be reliable. 10 of the twelve items were in this range, with C7 (__ is the first brand I think of in its category) at .198 and C11 (I pay a lot of attention to anything about __) at .734. Therefore no item was deleted on the Cognitive dimension.
The Emotional dimension consisted out of 18 items. The Alpha was .881 and no Alpha if item deleted was greater than this number giving no reason for deletion. 16 of the items on this
dimension had item-to-total correlations between 0.3 and 0.7 where E10 (I engage with __ because I have to) had .282 and E14 (I am heavily into __) was only slightly higher at .701 which called for all items to be retained.
The third dimension, Behavioral, had 10 items. The Alpha was .762 and one item had an Alpha if item deleted greater than Alpha plus 0.2. Therefore the item B6 (I won’t engage with another brand similar to __) was deleted, also having an item-to-total correlation of only .154. Only B7 (Engaging with ___ feels boring) fell outside of the item-to-total correlations out of the remaining 9 items at .205. The 9 item dimension had a Cronbach’s Alpha of .777 and was deemed reliable.
3.4 Confirmatory Factor Analysis
To perform the Confirmatory Factor Analysis or CFA the statistical program IBM SPSS AMOS Graphics 22.0 was used. Starting with 39 items the analysis was run many times and each time an item was deleted based on several indicators to come to a good model fit. These were factor loadings, modification indices and standardized residual covariances. Factor loadings below 0.6 and with an average above 0.7, modification indices above 10 and standardized residual covariances above 0.4 were taken as thresholds (Hair et al, 2010) resulting in the deletion of 27 items. The final model along with its factor loadings can be found in figure 3.1 on the following page.
The remaining 12 items were divided amongst the dimensions as follows: The Cognitive dimension had 4 items; C8 (I like to know more about __), C10 (I notice information related to __),
27
C11 (I pay a lot of attention to anything about __) and C12 (I keep up with things related to __). This dimension pertains to the gathering and processing of information to engage with a brand in a more informed way.
The Emotional dimension consists of 5 items; E1 (I am passionate about __), E2 (I love __), E5 (I am proud when I engage with __), E7 (I identify myself with __) and E8 (I get enthusiastic when I engage with __). The dimension consists of items relating to feelings and emotions felt when interacting and engaging with a brand.
The third and final dimension is the Behavioral dimension which is made up out of 3 items; B4 (I want to share my opinions about ____ with others), B5 (I will defend ___) and B8 (I love talking about__ with my friends). The overarching concept in this dimension is acting on the brand engagement with other people. This dimension has a clearly social theme.
Table 3.1 Model fit summary
In table 3.1 a summary is given of the model fit along with thresholds (Hu and Bentler, 1999). The Root Mean Error of Approximation is in the moderate range (0.05-0.10) as is the Goodness of Fit index where 0.90 is seen as acceptable but not great. The model therefore indicates a good to average model fit.
Model Thres-
hold1 Threshold Yes/No?
Degrees of Freedom 49
Chi-square Minimum Value 77.67
Chi-square Minimum Value Divided by DoF 1.59 < 3 Yes Root Mean Square Error of Apporoximation (RMSEA) 0.07 < 0.05 Moderate
PCLOSE for RMSEA 0.12 > 0.05 Yes
Standardized RMR 0.05 < 0.09 Yes
Goodness of Fit Index (GFI) 0.90 > 0.95 Moderate Adjusted Goodness of Fit Index (AGFI) 0.85 > 0.80 Yes
Comparative Fit Index (CFI) 0.95 > 0.95 Yes
Note: Thresholds from Hu and Bentler (1999)
28
The covariances between the three factors or dimensions are relatively high (0.57, 0.68 and 0.68) but this was likely to happen as the three dimensions are all part of the same greater construct of Customer Brand Engagement.
Figure 3.1 Final CFA Model for CBE including factor loadings and covariances
3.5 Testing for validity
Different types of validity were tested in several ways. First convergent validity was assessed The standardized loadings and the respective t-values are shown in table 3.2. The factor loadings range from 0.64 to .85 with all dimensions averaging above 0.7. The loadings indicate convergent validity. All t-values above 1.96 are significant and this also indicates good convergent validity (Vivek, 2009). The reliability of the dimensions is also above the threshold of 0.7 (Cognitive, α=0.81; Emotional, α=0.83 and Behavioral, α=0.76) indicating good reliability.
29
Table 3.2 Means, SD’s, Factor Loadings, Factor Alphas and t-values for CBE dimensions
Table 3.3 on the following page shows the correlations, means, standard deviations of the CBE scale along with Composite Reliabilities (CR), Average Variance Extracted (AVE) and the square root of the AVE. The composite reliabilities for the factors show that the factors are reliable. Cognitive is above the threshold of .7 (Hair et al., 2010) and has a CR of 0.81. Emotional has a CR of 0.83 and Behavioral has a CR of 0.77.
Convergent validity is also examined by the Average Variance Explained or AVE. As threshold the AVE has to be above 0.5 which is the case for all three dimensions. Cognitive (AVE=0.52), Emotional (AVE=0.50) and Behavioral (AVE=0.52) all are just above the threshold so the items in these factors seem to have convergent validity. However, the small margin raises some concern regarding generalizability because in a similar research it might just fail to be above 0.5.
To assess discriminant validity the square root of the AVE is taken to see if any correlations are higher. If this is not the case there is discriminant validity (Hair et al., 2010). All correlations are below 0.6 where the lowest √AVE is 0.71 giving evidence to discriminant validity.
Dimension / Items* Cognitive (α=.81)
C 8 I like to know more about ___ 3.16 0.94 0.64 35.40
C 10 I notice information related to ___ 3.82 0.84 0.65 48.11 C 11 I pay a lot of attention to anything about ___ 3.17 0.99 0.85 34.04 C 12 I keep up with things related to ___ 3.27 1.04 0.73 33.27
Emotional (α=.83)
E 1 I am passionate about ___ 3.43 1.00 0.69 36.21
E 2 I love ___ 3.40 1.01 0.64 35.69
E 5 I am proud when I engage with ___ 3.25 0.99 0.74 34.71
E 7 I identify myself with ___ 3.13 1.04 0.71 32.03
E 8 I get enthusiastic when I engage with ___ 3.64 0.85 0.75 45.49
Behavioral (α=.76)
B 4 I want to share my opinions about ___ with others 3.36 1.01 0.81 35.10
B 5 I will defend ___ 3.33 0.91 0.67 38.55
B 8 I love talking about ___ with my friends 2.71 0.95 0.68 30.27
Mean SD LoadingFactor T-value
30
Table 3.3 Means, SDs, Correlations and validity measures for CBE scale
Mean SD 1 2 3 4 5 6 7 8 9 10 11 12
Factor 1: Cognitive AVE .52 CR .81 √AVE .72
Cogn 1 3.16 0.94 1.00
Cogn 2 3.82 0.84 0.46 1.00
Cogn 3 3.17 0.99 0.52 0.55 1.00
Cogn 4 3.27 1.04 0.52 0.43 0.63 1.00
Factor 2: Emotional AVE .50 CR .83 √AVE .71
Emot 1 3.43 1.00 0.34 0.27 0.49 0.23 1.00
Emot 2 3.40 1.01 0.22 0.29 0.38 0.21 0.57 1.00
Emot 3 3.25 0.99 0.12 0.21 0.36 0.21 0.41 0.44 1.00
Emot 4 3.13 1.04 0.23 0.12 0.44 0.28 0.47 0.41 0.60 1.00
Emot 5 3.64 0.85 0.27 0.19 0.39 0.24 0.54 0.45 0.58 0.53 1.00
Factor 3: Behavioral AVE .52 CR .77 √AVE .72
Beha 1 3.36 1.01 0.34 0.45 0.44 0.38 0.39 0.31 0.47 0.32 0.39 1.00
Beha 2 3.33 0.91 0.25 0.38 0.29 0.37 0.22 0.35 0.46 0.39 0.24 0.56 1.00
Beha 3 2.71 0.95 0.22 0.34 0.45 0.39 0.30 0.29 0.38 0.33 0.38 0.54 0.44 1.00
Correlation Matrix of CBE Scale Items
31
Discriminant validity was also measured by comparing CBE with Involvement. These two measures are closely related but still different as mentioned earlier in the research (Mittal, 1995; Zaichkowsky, 1985). The constructs are tested by computing the correlation coefficients of the dimensions and Involvement. The results of these computations can be found in table 3.4.
Table 3.4 Validity Measures for CBE dimensions and Involvement
Using AMOS, a model was made to include Involvement. This model showed good fit (χ2/df = 1.31, GFI= 0.87, AGFI = 0.83, CFI = 0.96, RMR = 0.05, RMSEA = 0.05 and PCLOSE = 0.41) allowing to go further with this analysis. The correlation coefficients are all lower than the square root of the AVEs which are placed on the diagonal. This suggests that the construct of involvement is distinct from all the dimensions of CE and show discriminant validity (Fornell and Larcker 1981).
Convergent validity in this model has similar evidence concerning AVEs as the final model mentioned above. All factors are above the threshold of 0.5, including Involvement (AVE=0.57), indicating convergent validity (Fornell and Larcker, 1981).
No. of
Items Mean SD CR AVE Cogni- tive
Emoti- onal Inter- action Involve- ment Cognitive 4 3.35 0.76 0.81 0.52 0.72 Emotional 5 3.37 0.76 0.83 0.50 0.57 0.71 Behavioral 3 3.13 0.79 0.77 0.52 0.68 0.68 0.72 Involvemen 5 3.56 0.68 0.87 0.57 0.47 0.61 0.49 0.76
Correlation Matrix of CBE Dimensions and Involvement
Note: SD=Standard Deviation, CR= Composite Reliability, AVE=Average Variance Extracted, N=112 On the diagonal the square root of the AVE is mentioned. Correlations above r= |.16| are significant at p <.05 (one-tailed)
32 3.6 Common Method Bias
The model was also tested for Common Method Bias or CMB. CMB refers to a bias in which an external factor may be responsible for biased outcomes due to for instance a single method of collecting data, i.e. in an online survey (Podsakoff et al., 2003). Even though this particular case is not applicable to this study, CMB was evaluated through two tests.
The first test is Harman’s single factor test whereby an EFA is run with one factor and no rotation. If CMB is an issue the variance explained by one factor is above 50 percent. In our case the variance explained was 42.78 percent and no issue was found.
The second way of testing in with the common latent factor method. This approach connects a common latent factor to all items in the model and compares the standardized regression weights of that test with those without the common latent factor. When there are differences above 0.2 there is Common Method Bias. In table 3.5 the difference applied to this model are shown.
Table 3.5 Common Method Bias differences for CBE model
Especially the Emotional dimension (4 out of 5 items) and one item on the Cognitive dimension have Common Method Bias. To overcome this in further research is to retain a common latent factor in analyses with a structural model.
Item C 8 C 10 C 11 C 12 E 1 E 2 E 5 E 7 E 8 B 4 B 5 B 8
Difference 0.11 0.07 0.23 0.00 0.60 0.39 -0.08 0.21 0.32 0.12 -0.03 0.10
33
Chapter 4: Discussion
This chapter will discuss the first two research objectives found in the introduction to see how these have been answered. Then the third research objective, concerning the scale
development, is discussed by looking at the results and this will be followed by the limitations of the research and possible avenues for further research.
4.1 Research objectives 1 and 2
The first research objective that was formed was to see if there was a need for the construct of Customer Brand Engagement. The Marketing Science Institute made ‘Customer Engagement’ one of their top priorities for 2006-2012 (MSI, 2010). There has been research into several similar concepts to CBE, such as employee engagement (Saks, 2006; Frank, Richard & Taylor, 2004), student engagement (Bryson & Hand, 2007; Hu 2010) and occupational engagement (Bejerholm and Eklund, 2007).
Even though the year is 2015, many researchers still call for a further deepening of the concept of customer engagement and customer brand engagement as a truly separate concept, especially in the field of marketing (Bowden, 2009; Hollebeek, 2011a; Patterson et al., 2006). After these reading these arguments it is clear that there is a need for CBE.
The second research objective was to determine the dimensions of CBE. In the literature about engagement and similar concepts there is no consensus about the dimensions. Some researchers see (customer) engagement (Achterberg et al., 2003); Resnick, 2001) and CBE (Heath, 2007) as a one-dimensional concept. However, the greater part of researchers see (customer) engagement (Bowden, 2009; Mollen and Wilson, 2010; Vivek et al., 2012) and CBE (Hollebeek, 2011b, 2013) as a concept with three different dimensions.
This greater part of research is in unison about a three-dimensional model however there is confusion among researchers how to name these dimensions even though they pertain to the same type of contents. Following Hollebeek (2011a, 2011b; 2013) the dimensions are given names that
34
capture these different contents best. Note that these dimension all happen in a motivational, context-dependant, interactive state with the brand.
The first dimension is called the cognitive dimension. This dimension looks at the processing and gathering of information in the mind of the consumer. He or she than uses this information when engaging with the brand of choice. Some examples of this dimension are learning or thinking about the brand or actively searching information about the brand. This dimension contains “objective” and introvert characteristics.
The second dimension is the emotional dimension. In this dimension feelings about the brand are found. Examples of this dimension are love and passion for the brand, pride when
engaging with the brand and not being able to live without the brand. These characteristics are more subjective in nature but still introvert in the sense that it is purely personal to the consumer
engaging with the brand.
The third and final dimension is the behavioral dimension. This dimension is about acting on your engagement with the brand. The main things in this dimension are spending time and effort to engage with the brand but also a social aspect in defending the brand and using it with others. Therefore this dimension has subjective and objective parts and is truly extrovert in nature.
The results of the scale development in the following section prove that CBE is a three-dimensional concept.
4.2 Discussion of the results of the scale development
The third and most important research objective of this thesis is the development of a scale for customer brand engagement. In the literature a quantitative scale for customer engagement was elusive with the exceptions of the research of Appelbaum (2001) and Vivek (2009). A scale for CBE was not compiled thus far. Based on the qualitative research by Hollebeek (2011b), this thesis goes on to develop a quantitative scale for CBE.
35
After a pretest among non-students, an online survey was conducted with an online response rate of 64%. This was supplemented with offline surveys to come to an overall sample of 112 responses. As mentioned before, due to the difficulty of the concept, it may be troublesome for people to think of a brand with which they actively engage because this is not asked of people on a normal basis. It may be hard for respondents to find the sort of information which is asked in the survey, resulting in a possible measurement error. The descriptive statistics also indicate a sampling error, as the sample is relatively young and highly educated.
The data was then tested for reliability whereby one item was deleted, leaving 39 remaining. The next step was performing a Confirmatory Factor Analysis using IBM AMOS. By looking at factor loadings below 0.6 and with an average above 0.7, modification indices above 10 and standardized residual covariances above 0.4 (Hair et al, 2010) 27 items were deleted.
The 12 remaining items resulted in three reliable dimensions with moderate to good model fit. The Cognitive dimension had 4 items (α=0.81), the Emotional dimension had 5 items (α=0.83) and the Behavioral dimension had 3 items (α=0.76). The individual items are listed in the previous chapter. It has to be noted that the three items in the Behavioral dimension all have a social component. Therefore it is also an option to name this dimension ‘Social’ instead of ‘Behavioral’. Reliability, convergent and discriminant validity were tested by looking at numerous indicators. Composite reliabilities, Cronbach’s Alpha, t-values, Average Variance Extracted, √AVE and correlation coefficients were all found to support either reliability or validity. The AVE’s were above the
threshold of 0.5 but only barely (0.52, 0.50 and 0.52). It will be interesting to see if these indicators will be higher or lower with a different and bigger sample.
The three dimensions also had high covariances (between .57 and .68) but this seems logical given the nature of the dimensions. They are three parts of the greater concept of CBE and therefore high covariances are to be expected.
The construct of involvement was also examined to see if there was sufficient construct validity. A model with Involvement as a fourth factor was constructed and presented good model fit.
36
All indicators supported reliability and validity and therefore involvement is seen to be significantly different from CBE.
A final test was done to check the model for Common Method Bias. CMB refers to a bias in the data where an external factor may be responsible for biased outcomes. This is due to for instance a single method of collecting data, i.e. in an online survey (Podsakoff et al., 2003). When a study has CMB, the majority of the variance is explained by a single factor.
After testing for this phenomenon 5 items out of 12 have Common Method Bias. One item on the Cognitive dimension, just above the threshold of 0.2, and 4 out of 5 on the Emotional dimension. A possible reason for this bias is the subjective nature of the questions in the Emotional dimension. Respondents may find it difficult to attribute feelings towards brands.
The scale developed in this thesis has good model fit and no reliability and validity concerns. With this research an answer has been provided to the call for a quantitative scale for Customer Brand Engagement.
4.3 Limitations
This thesis has several limitations. The first and the biggest limitation is the way in which respondents seem to lack a grasp of the concept of CBE or their ability to attribute characteristics to their brand of choice. The online response rate was 64%, indicating that the concept of CBE is not clear enough and has to be explained shorter and better. The online tool Qualtrics showed that of the people that did not complete the survey, the majority stopped after the first page. People might be scared off by a large piece of text and therefore will not proceed with the survey.
The second limitation in this survey is the sample that was used in this research. The respon-dents that participated are highly educated and relatively young, indicating sampling error. The sample is not representative of the population and does not give a true image of that population, leaving the possibility that a representative sample will yield a different outcome.
37
The final limitation of the research is the Common Method Bias. When using the items in a structural model a test for CMB must be done to prevent from using biased data. When the data indeed presents with CMB, the factors need to be imputed with a common latent factor (Hair et al., 2010). However, it is to be seen if this problem still arises with a large, representative sample.
4.4 Avenues for further research
Further research should aim to test the scale for CBE on generalizability. This is the only way in which the strength of the scale can be tested objectively. A larger, representative sample will indicate if the scale will stand up to scrutiny.
It would be interesting to see if there is difference between different demographic groups. This could be differences between countries, gender, income and several other examples could easily thought of. Another possible avenue is to look at different moments in time and to see if levels of engagement strengthen or weaken over time. This could be a natural process or after a certain marketing campaign, which would be a research avenue for managers.
As hedonic and utilitarian products differ very much in a customer’s experience, it is
expected that utilitarian products score higher on the cognitive dimension and hedonic products on the emotional dimension. If this is indeed the case, a whole range of new research possibilities would open up. Not only for academic literature but also for most brands that want to see how to engage better with their customers.
Another distinction that can be made is that between normal hedonic products and products such as alcohol or cigarettes. Certain items in the survey, that however have been dropped, such as “I engage with __ because I have to” and “__ is part of my life” get a significantly different meaning when viewing it from these different angles. For social sciences research regarding addiction or firms as Heineken (which is very active in engaging its customers) it may be that for these categories a new scale needs to be developed.
38
This chapter has discussed the research objectives and how they have been met in this research. This was followed by several limitations. Finally several avenues for further research were presented, both for academic and managerial applications. In the next chapter a conclusion for the research is given.
39
Chapter 5: Conclusion
In this chapter a summary is given of the previous sections. First the research objectives are discussed concisely, after which the literature review will be summarized. After that the scale development and method are discussed shortly. Then the results are discussed and the limitations and avenues for further research are summarized.
In the literature review the need for customer brand engagement was thoroughly examined. The review started with the academic history and origins of the concept of engagement and through customer engagement further narrowing towards the concept of customer brand engagement. It became clear that there is a still growing need for exploration of these constructs and for CBE, as research has been lacking in marketing between customer engagement and brands in particular.
The construct of CBE also seems to have numerous antecedents and consequences such as involvement, relationship quality and brand loyalty. These other constructs may be better
understood by taking CBE into account which further establishes the need for customer brand engagement.
Now that the need for the construct of CBE is established, further investigation was done into the dimensionality of the construct of CBE and its overarching concepts of engagement and customer engagement. In the academic literature many different forms were found. Some saw one dimension as best fitting, while other researchers took three dimensions (cognitive, emotional and behavioral or similar named dimensions) as best describing the model. It was also noted that often two of the preceding dimensions were used. After careful analysis the outcome seems to favor the three-dimensional approach.
Measuring CBE would be troublesome as research has failed to provide a scale. Therefore this research took initial qualitative research done by Hollebeek (2011 a, b; 2013) and this resulted in an online and offline survey which supplied data of 112 respondents to form a quantitative scale. Using SPSS and AMOS, a confirmatory factor analysis provided a moderate to good fitting model
40
which stood up to scrutiny concerning reliability and construct validity. This model contained 12 items divided over three dimensions. However, due to the small sample in this research and newly formed survey, sampling and measurement errors may be an issue.
The scale provided is open to numerous options for further research both academically as well as from a managerial standpoint. Further testing would strengthen the scale and managers could better evaluate their campaigns and the way they interact with their customers.
