Student: Joshua Picauly Student number: 10018115
Research Question: How does the stigmatization of an industry affect the consumer evaluation of legitimate brands entering this stigmatized industry with a brand extension?
Keywords: Stigma | Entry Strategy | Branding | Gambling | Media Conglomerates
Master Thesis
Is It Worth The Risk? A Study On The Effects Of
Entering A Stigmatized Industry As A Legitimate
Organization.
STATEMENT OF ORIGINALITY
This document is written by Joshua Picauly who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
A
BSTRACT
This study examines how the stigmatization of an industry affects the consumer evaluation of legitimate brands entering this stigmatized industry with a brand extension. The study was conducted for TV broadcaster RTL Nederland who is considering to enter the (stigmatized) gambling industry. The reason for this is that broadcasters are looking for alternative revenue models that can bring more stability than the traditional advertising model can (Evans & Donders, 2013). RTL management is considering gambling as an interesting opportunity but they are cautious to enter due to the assumed negative feedback effects. In order to test this assumption, RTL wants to know that when they extend one of their brands into this stigmatized industry, what the effect on the initial parent brand would be.
In order to answer the research question several hypotheses were formulated. Data was collected through an experiment with four treatments, associating an RTL brand with a gambling game (Bingo, Slots, Casino, Control). The research shows that associating a brand with a gambling game is damaging to the brand image. Furthermore, an indirect effect was found between the degree of stigmatization and parent brand image post extension via the brand extension attitude. No evidence was found for a direct effect of the stigma on the parent brand nor an indirect effect via the perceived fit.
To conclude, managers need to think careful about extending a brand and they should always be aware of the damage it can do to the parent brand. The brand image is one of the most valuable assets an organization has and taking a risk with this can be very damaging to the company.
F
OREWORD
Before you lies the product of six months of dedication and hard work. It is the final effort of my entire education. After this, a new journey will start without the guidance and assistance of teachers and professors. Thanks to their combined effort that I feel I am ready to start.
I would like to thank Bjorn Fuchs for providing me such an interesting topic, Joris Ebbers for his feedback, all the respondents to my questionnaire, and, of course, my boyfriend for all the moral support.
Bedankt. Joshua Picauly
T
ABLE
OF
CONTENT
1. Introduction 6 2. Literature review 8 2.1. Brands 8 2.2. Stigma 17 2.3. Hypothesis 20 3. Method 25 3.1. Empirical setting 25 3.2. Research design 26 3.3. Survey flow 27 3.4. Materials 27 3.5. Sample 29 4. Results 30 4.1. Reliability analysis 304.2. Difference between treatments 31
4.3. Hypothesis testing 33
5. Conclusion & Discussion 37
6. References 40
7. Appendix 44
7.1. Online Survey 44
7.2. Development of variables 50
1. I
NTRODUCTION
Traditional television broadcasters operate in volatile media landscape with many difficult obstacles to overcome (Evans, 2010). One of the big problems commercial broadcasters need to deal with is that their performance is highly dependent on the macroeconomic cycle (EAO, 2012). In prosperous periods, selling advertisements is relatively easy and high conversions can be achieved. However, in times of recession, decline in advertising revenues and subsequently decline in overall performance of the broadcasters will occur. Hence, broadcasters depend on advertisers making them vulnerable to macroeconomic declines. The TV industry is aware of these challenges and obstacles and is therefore looking for alternative revenue models that can bring more stability than the traditional advertising model (Evans & Donders, 2013). Instead of looking at advertisers for revenue, broadcasters are trying to earn money directly from the consumers (Doyle, 2010). For instance, through propositions such as streaming platforms, broadcasters can increase their revenues by charging the consumer with a pay-per-view- and/or subscription-model (Doyle, 2010). Furthermore, broadcasters are also investing in start-ups through their own venture vehicles. German broadcaster ProsiebenSat.1 was part owner of the online retailer Zalando and still has equity in travel comparison website TriVaGo. Dutch broadcaster RTL has an equity-stake in companies such as Pepper, Couverts and Squla. After a successful growth period, equity is sold and additional revenue is created. These are just a few examples of TV broadcasters looking for new opportunities to create additional revenues. One of the most promising opportunities is the expected legalization of the Dutch gambling industry in 2016. In the current situation, only four parties are allowed to exploit gambling propositions with very strict governmental regulation. However, due to the internet, Dutch consumers are no longer bound to these four parties. Hence, a major amount of betting is done illegally, through foreign sites such Pokerstars, BWIN and Ladbrokes. The government wants to control this illegal gambling in order to protect consumers for gambling addictions, to create a fair competitive market for Dutch and international parties, and, most importantly, to receive tax money over the illegal betting. A key part of proposed legislation is allowing more parties to offer gambling propositions and distributing this proposition online.
Media companies are all eager to enter because of the very high margins this industry has to offer. RTL, Sanoma, SBS, Talpa and TMG are all preparing their own gambling proposition, to be launched on the very first day legalization is accepted. However, entering the gambling industry is not without dangers.
Gambling is just like private military contracting (Baum and McGahan, 2013), big oil (Levy and Egan, 2003), big box retailing (Yue, Rao, and Ingram, 2013), and tobacco (Galvin, Ventresca, and Hudson, 2004), seen as a stigmatized industry. What these industries have in common is that their industry belief systems are in conflict with those of certain social audiences (Galvin, 2002). The core attributes of these industries, such as their products, their output or their customers, are perceived as violating the social norms of certain audiences (Hudson, 2008). Being part of a stigmatized industry has many negative effects for business. Devers et al (2009) describes two main effects of stigmatization on firms; (1) deindividuation and (2) discredit to the organization. The effects of stigmas on organizations can therefore have severe implications for strategy, marketing and other business conduct. Nonetheless, the profit margins in the gambling industry are so high (see table Table 2), companies are willing to take the risk.
Although studies describe the negative effects of stigmatized industries on industry actors, none of the research has focused on describing the effects of entering a stigmatized industry as a legitimate organization. Less is even known about the effects of stigmas on brand management. Therefore, research has to be conducted in order to help legitimate companies such as RTL, make a well informed decision about entering a stigmatized industry.
In order to investigate this issue, help companies make a well informed decision, and to contribute to the existing literature, this paper will examine the effect of stigmas on market entries and consumer evaluations of both the parent brands and brand extension. This results in the following research question:
‘How does the stigmatization of an industry affect the consumer evaluation of legitimate brands entering this stigmatized industry with a brand extension?’
The purpose of this paper is to gain insight into the effects of stigmas on market entry. Specifically, based on theoretical perspectives from the branding- and sociology literature, the author proposes and tests several relationships involving the parent brand image, degree of stigma, brand extension attitude and perceived fit. In order to give an answer to the above mentioned research question, this paper will start by reviewing the available literature on brand, brand extensions and stigmas. Then a methodological chapter will provide the needed explanation on how the research is conducted. Next, a results section will be provided and finally the conclusion and discussion.
2. L
ITERATURE
REVIEW
The following section provides a review of the current state of literature about the concept of brands, brand building, extension feedback and stigmas. These will provide the necessary theoretical background for the further analysis in this paper.
2.1. B
RANDSA brand is a complex phenomenon (Maurya & Mishra, 2012). Many different scholars have defined a brand according to their own perspectives, which has led to countless definitions with many nuance differences (Kapferer, 2004). One of the most often used definitions was proposed by Keller (1991). He defined a brand as “a name, term, sign, symbol, or design, or combination of them which is intended to identify the goods and services of one seller or group of sellers and to differentiate them from those of competitors” (p. 442). Although widely used and accepted (e.g. American Marketing Association), Keller’s definition also received criticism focusing on the disregard for intangible assets (Bennett, 1988; Dibb et al. 1997; Levy 1999) and for lack of consumer's perspective (Crainer, 1995, Arnold 1992). A definition that takes into account these insufficiencies is the one from author Seth Godin (2003). He states: ‘A brand is the set of expectations, memories, stories and relationships that, taken together, account for a consumer’s decision to choose one product or service over another. If the consumer (whether it’s a business, a buyer, a voter or a donor) doesn’t pay a premium, make a selection or spread the word, then no brand value exists for that consumer’ (Godin, 2003). The consumer is centered in Godin’s perspective and issues like legalities are not incorporated. Other examples of definitions describe a brand as a legal instrument (Crainer, 1995; Broadbent and cooper, 1987), as a company (Varadaranjan et al., 2006) as an Identity system (Kapferer, 1992; Balmer, 1995; Aaker, 1996) or brands as an image in the consumer's mind (Keller, 1993; Keeble, 1991; Gardner and Levy, 1995; Park et al., 1986).
Since there is no conformity in the academic world about the definition, this paper will use the definition that is currently used by the AMA and is an adaptation of Keller’s 1991 definition. It states:
A brand is a "Name, term, design, symbol, or any other feature that identifies one seller's good or service as distinct from those of other sellers." AMA (2007)
A brand is something different than a product. According to Philip Kotler (2003) a product is a thing that can be offered to a market to satisfy a want or need”. Thus a product can be a good, a service, a shop, a person,
an organization, a place or an idea. A brand then adds some dimensions to the product to distinguish it from the rest. It provides the basis upon which consumers can identify with the product and distinguish it from other similar products (Weilbacher, 1995). Through its name, term, design, symbol, or any other feature that identifies its good or service, brands can create perceptions and feelings with the consumer (Keller, 2003). Brands fulfill different functions for the consumer. According to Alsem (2005), brands make products and services recognizable, identifiable, they create trust, they provide an indication of the quality and they have a symbolic function. Brands identify the source of the product or service and thereby create meaning for the consumer (Alsem, 2005). These meanings are bundled in a network of associations in the memory of consumers and behold all the consumer knowledge about products and services (Alsem, 2005).
The process of strategically managing brands and associations with the purpose of creating commercial value is known as branding (Kapferer, 2004). Branding is used to position goods and services and to distinguish it from that of other organizations in the market (Keller 2008). The main function of branding is creating, coordinating, monitoring and adjusting interactions between an organization and its customer’s ideas about the brand (Ewing and Napoli 2005). In other words, marketers must actively build and exploit brands in order to conquer and defend a position in the mind of the consumer.
# Associative network
From the AMA definition can be derived, that brand elements (e.g. name, term, design, symbol or any other feature of a brand) activate certain areas in the mind of the consumers, hence an activation of the consumer’s memory. This activation process is thoroughly studied and many scholars have conceptualized it as ‘the associative network model‘(Collins & Loftus, 1975; Anderson 1983). Kotler & Keller (2009) describe this associative network model as:’ a conceptual representation that views memory as consisting of a set of nodes and interconnecting links, where nodes represent stored information or concepts and links represent the strength of association between this information or concepts’. To put it more simple, when one area in the brain is activated by a stimulus, other related areas are activated as well. For example, a person hears the sound of a fire engine, and immediately associates this with danger, vehicle, red and fire (Figure 1). Based upon the associative network model, Aaker developed the brand associative model (Aaker, 1996). He argues that a node in the associative network can also be a brand to which a variety of associations are linked (Aaker, 1996). In other words, when a consumer comes into contact with a brand element, such as the logo of McDonald’s, nodes such as social environment, service and meals are activated through the brand
associative network (Figure 1). All these nodes of information in the mind of the consumers is what Keller defines brand knowledge (Keller, 1993).
Figure 1- Associative network model
# Brand knowledge
Keller (1993) defines brand knowledge as a function of brand awareness and brand image (Figure 2). Rossiter and Percy (1987) describe brand awareness as the ability of consumers to identify a brand under different conditions. Brand awareness has two separate dimensions: recognition and recall. Brand recognition relates to consumers' ability to indicate that he or she has seen been in contact with a brand element when given the brand as a cue (Keller, 1993). For example, you hear a distinctive Tv-show tune and you immediately recognize this is the tune of The Voice of Holland. Brand recall relates to the extent to which a customer can mention the brand when he hears the product category, or needs that the product category solves (Keller, 1993). For example, you are asked to name, a radio channel and the first brands that come to mind are SkyRadio, Radio538 and 3FM. These three brands have a high recall in the category radio because these come to mind first.
As far as brand image, Keller (1993) defines this ‘as perceptions about a brand as reflected by the brand associations held in consumer memory’ (p3). Brand image is all about knowledge of the consumer about the brand and which associations come to mind. It is conceptualized as a construct which exists out of several dimensions. The first dimension is the ‘type of brand associations’. In three major categories, brand associations are categorized based upon the amount of information that the categories behold about the brand (Keller, 1993). In short, (1) attributes are the descriptive features that characterize a product or service; (2) benefits are the personal value consumers attach to the product or service attribute and (3) brand attitudes are defined as consumers' overall evaluations of a brand’ (Keller, 1993). The sum of these three categories is the dimension ‘type of associations’ (Figure 2). The other three dimensions classify the qualitative nature of these associations in the mind of the consumers, according to their favorability, strength, and uniqueness. In brief, brand knowledge is the all the information and associations of a brand that is stored within the memory of consumers.
# Brand equity
The value of a brand, created by the branding efforts of marketers, can be expressed as brand equity. Although a lot of different views on brand equity exist, most scholars agree that the concept can be defined in terms of marketing effects that are uniquely attributable to a brand (Keller, 1993). In other words, the results of a marketing program are strongly determined by the power of the brand the program is intended for. The higher the brand equity, the better people will react to marketing efforts. Positive brand equity leads to greater perceived differentiation, stronger brand loyalty, larger profit margins, higher trade support and increased marketing communication effectiveness (Keller, 1993) Additionally, brands that have established desired brand equity in the marketplace can have potential to leverage its equity through line extension,
brand extension, ingredient branding, cobranding, brand Alliances, and/or social goodwill (Ghodeswar, 2008).
Companies are interested in building strong brands with great brand equity, but it is not always easy to reach that point. In order for marketeers to understand the process a brand has to go through before brand equity is realized, Keller designed his Customer-Based Brand Equity Model. Keller defines his Customer-Based Brand Equity (CBBE) conceptual approach to brand equity as: “The differential effect that brand knowledge has on consumer response to the marketing of the brand (Keller, 2009). In other words, CBBE measures the consumers’ reactions to marketing efforts for the brand in comparison with their reactions to the same marketing mix element attributed to a fictitiously named or unnamed version of the product or service (Keller, 2009).
The model is based upon the assumption that the strength of a brand lies in what consumers have learned, felt, seen and heard in the course of time about the brand (Keller, 2009). For this reason Keller states that the strength of a brand is determined by the brand knowledge of consumers. The challenge for marketers is to ensure that consumers have the right experience with the products and services. Furthermore, marketers must develop marketing campaigns that link the desired thoughts, feelings, images, perceptions and attitudes to the brand (Keller, 2001). In the end, the CBBE model guides marketers in achieving, measuring and controlling brand equity in the mind of consumers.
# Brand Building
As stated before, the CBBE model provides marketers a step-by-step approach to building brands. In his approach, Keller states that strong brand can only be created if a brand goes through all the 4 steps of the model and that continuing to the next step depends on successfully achieving the previous step. The first step and building block of the CBBE model is brand salience. This refers to “how easily and often customers think of the brand under various purchase or consumption situations” (Keller, 2009). This step is all about establishing a distinct brand identity that consumers can recall and recognize (Keller, 2001). Furthermore, it is important to ensure that consumers identify with the brand and have several associations when thinking of it. Additionally, by creating brand salience, consumers will become aware of the category membership of the product (Keller, 2001). Creating brand salience is closely related to brand awareness of the Keller 1993 article.
The next step in the CBBE model has two building blocks, namely performance and imagery. Brand performance refers to “how well the product or service meets the customers’ functional needs” (Keller, 2001). The brand performance depends on the actual experience of using the product. Product characteristics and features, reliability, durability and price are, among others, factors that are of influence on performance (Keller, 2001). In contrast with brand performance is brand imagery. This describes ‘the extrinsic properties of the product or service, including the ways in which the brand attempts to meet customers’ psychological or social needs” (Keller, 2001). While performance is all about functionality, imagery is all about the emotional associations and the intangible assets such as heritage, personality, values and purchase situations. In the end, through branding efforts on brand performance and brand imagery, marketers can create a meaning or essence of the brand in the mind of the consumers (Keller, 2001).
The third step in the CBBE model also has two building blocks, namely brand judgments and brand feelings. Brand judgments “focuses on the consumers’ own personal opinions and evaluations” (Keller, 2001). They evaluate the associations of brand performance and create an opinion based upon quality, credibility, consideration and the superiority of the brand. In contrast, brand feelings refer to “consumers’ emotional responses and reactions with respect to the brand” (Keller, 2001). The feelings that can arise with the brand are warmth, fun, excitement and security amongst others. Furthermore, certain brand feelings can be a reason for consumers to buy products. Especially in the fashion industry, branded products that evoke certain feelings are used as goods to create an identity with (Dubois & Duquesne, 1993).
The final step of the CBBE model concerns the relationship of consumers with the brand and consists out of the building block resonance. Brand resonance refers to the nature of this relationship and the extent to
which consumers feel connected to a brand (Keller, 2001). Resonance is needed for customers to stay loyal to a brand. Brands that perform well in all the six building blocks have high brand equity and are perceived as strong (Keller, 2003).
# Brand extensions
Once a strong brand is established and companies want to leverage the brand equity for new opportunities, managers can implement a brand extension strategy. When a company enters a new market or category and uses a brand name of an already existing product this is known as a brand extension (Aaker and Keller, 1990). Companies implement a brand extension strategy because they can use the brand equity of the parent brand as a strategic tool to create a competitive advantage (Aaker, 1991). Brand extensions are capable of contributing to the growth and improvement of the parent brand by reinforcing the brand associations (Aaker, 2002) and strengthening the position of the brand on the market (Park et al., 1986). Due to these advantages, brand extensions are part of many strategic plans and have a major impact on the branding of products nowadays. Estimations indicate that 80% of all the new products are launched under the brand name of an existing brand (Simms, 2005).
According to Keller (2013) brand extensions can be divided into two categories. When a company launches a new product under the same name within an existing product category this is called a line extension (Keller, 2013). Coca-Cola introducing Coca-Cola Zero is a well-known example of a line extension. When a company launches a new product under the same name in a new product category this is called a category extension (Keller, 2013). A famous example of a company that often extends to new categories is Richard Brandson’s Virgin with ventures in music, air travel and banking.
Brand extensions have several advantages over developing a new brand. By using the strengths of an already existing brand and its image, marketers can save time and resources when compared to creating a completely new brand (Filipsson, 2008). In addition, consumers link the brand values and associations of the parent brand to the new product (Aaker, 1991). Maybe even more important is the advantage of facilitating product acceptance through brand extensions (Keller, 2009). According to Erdem (1998), brand extensions facilitate in acceptance of new products through their risk reducing qualities. Erdem (1998) describes that consumers are uncertain about product quality, but they believe in the existing brands and their value. Therefore, in the mind of the consumer choosing a branded product reduces the perceived risk of buying an unsatisfying product. Other advantages of brand extensions are increased efficiency of promotional and marketing communications expenditures (Erdem & Sun, 2002), avoid costs for brand development (Keller,
2009) and an increased probability of gaining distribution channels (Montgomery, 1975). Finally, brand extensions provide feedback benefits to the parent brand and company (Keller, 2009). More literature on feedback will follow.
A brand extension strategy is not a guarantee of success. According to Franzen (2000) a risk of brand extensions is that consumers may become confused by the extension. When a brand is extended to a product category that does not directly matches with the parent brand (e.g. perceived fit), consumers are not able to make their buying decision based upon the characteristics of the parent brand which leads to confusion (Franzen, 2000). For example, if Dutch religious public broadcaster EO would set up a new techno festival this could lead to confusion. The religious and conservative values of the EO do not match the product category of techno festivals. Additionally, the key values of the EO do not help consumers in their buying behavior due to the discrepancy with techno. As a consequence of this confusion the consumer will experience negative associations with the parent brand and the brand extension. These negative associations can arise when a brand extension evokes associations that are inconsistent or conflicting with existing associations about the parent brand (Tienstra, 2001).
In Table 1 an overview is provided of all the advantages and disadvantages of brand extensions according to Keller 2009.
Table 1 - Advantages and disadvantages of brand extensions (Keller, 2009)
Advantage Disadvantage
Faccilitate new product acceptance Provide feedback benefits to the parent brand and company
Can confuse or frustrate consumers Improve brand image Clarify the brand meaning Can encounter retail resitance Reduce risk perceived by consumers Enhance the parent brand image Can fail or hurt parent brand image Increase the probability of gaining
distribution and trial
Bring new customers into brand franchise and increase market coverages
Can succeed but cannibalize sales of parent
Reduce costs of introductory and follow up marketing campaign
Revitalize the brand Can succeed but diminish
identification with any one category
Avoid cost of developing a new brand Permit subsequent extensions Can succeed but hurt the parent brand image
Allow for packaging and labeling efficiences
Can dilute brand meaning
Permit consumer variety seeking Can cause the company to forgo the chance to develop a new brand
# Feedback effects
Brand extensions are an effective strategy when they are able to contribute to the growth and improvement of the parent brand by enriching the brand associative network and if they can strengthen the position of the parent brand in the market (Aaker, 2002). Although a brand extension can be a way to exploit the parent brand image, there are risks associated with brands extensions that could devaluate the parent brand image. An unsuccessful extension, could add negative and/or damaging associations to the associative network, which subsequently could harm the consumer evaluation of the parent brand (Ries & Troet, 1981).
When we talk about the influences of the brand extension on the parent brand this is sometimes referred to as the feedback or reciprocal effect (Czellar, 2003; Salinas and Perez, 2009). The feedback effect can be defined as the positive or negative reciprocal impact of the brand extension on the parent brand (Dwivedi et al, 2010). This influence is usually seen as reciprocal since the initial image of the parent brand influences the perception of the brand extension, and the perceptions and valuations of this extension in turn can influence the perception and appreciation of the parent brand (Dwivedi et al, 2010, Czellar, 2003).
The feedback process can be divided into two processes; (1) the evaluation of the brand extension and (2) the re-evaluation of the parent brand (Dwivedi et al, 2010). In process one, a transfer of knowledge and attitude of the parent brand to the brand extension occurs, creating (among other things) an easier acceptance of the extension (see paragraph Brand extension, page 14). In process two, the new associations that are created by the brand extension, are added to the associative network of the parent brand (Gurhan-Canli, Maheswaran, 1998). These new associations of the brand extension, the perceived fit between the parent brand and brand extension and the initial parent brand image together ignite incremental changes of the consumers' attitude toward the brand (Gurhan-Canli, Maheswaran, 1998). Overall, the image of a brand is a constantly evolving phenomenon since consumers and marketers constantly enrich the associative network with more and new associations. The introduction of a brand extension will therefore by definition affect the associative network and the overall perception of the parent brand.
# Summary
To summarize, a brand is a name, term, design, symbol, or any other feature that identifies one seller's good or service as distinct from those of other sellers (AMA, 2007). Managers can use brands to create a competitive advantage by building and leveraging brand equity. One way to do this is by extending the brand to new categories in order to reduce the risks and increase the chance of acceptance. However, brand managers need to be aware of the feedback effects that can occur when extending a parent brand.
2.2. S
TIGMAThe previous section provided elaborate insights into branding and brand extensions. In the next section an overview will be provided about the theories of stigmatization. The section will start with stigmatization on an individual level, followed by organizational and eventually by industry level stigmatization. After this section, the branding and stigma literature will be combined to formulate hypotheses.
# Social stigma
For many years sociologists such as Émile Durkheim (1895) and Gerhard Falk (2001) have studied the socially constructed phenomenon of stigmas. Scholars from other fields such as health care (Sheldon & Caldwell 1994, Fife & Wright 2000) and psychology (Walsgrove 1987, Coleman et al 1996) all use their own variations on the concept. This has led to a lack of a broadly accepted definition of the phenomenon. Undoubtedly the most influential work on stigmas comes from Erving Goffman (1963). He defined a social stigma as an "attribute that is deeply discrediting" and that reduces the bearer "from a whole and usual person to a tainted, discounted one" (Goffman 1963, p. 3). Stafford & Scott (1986, p. 81) elaborate by adding the element of social norm which can be understood as the shared belief that a person ought to behave in a certain way at a certain time. Interesting is the perspective of Crocker et al (1998). They describe that the attribute that ignites the stigmatization, conveys a social identity that is devalued in a particular social context (Crocker et al, 1998, p. 505). This thesis will use a conceptual refinement of Goffman’s definition by Jones et al (1984) as the definition of a social stigma because of the completeness and preciseness of the formulation. It states:
‘Stigmatization occurs when a social category about which others hold negative attitudes, stereotypes, and beliefs, or which, on average, receive disproportionately poor interpersonal or economic outcomes relative to members of the society at large because of discrimination against members of the social category’ (Jones et al, 1984).
Goffman (1963) distinguishes three different conditions in which stigmatization of an actor can occur. The first stigma described is an abomination of the body; deformities of the body or other physical defects. The second stigma is the shortcomings of the individual character such as a person’s nature to be violent, addictions or other mental illnesses. The last stigma condition are the tribal identities such as race, nation and religion. These are passed on from generation to generation and include all members of a family (Goffman, 1963).
When an individual is stigmatized, there is an important and negative difference between the virtual (expected or projected) and the current (actually or present) social identity of the actor. Due to this difference all the other characteristics of the actor are threatened to be overshadowed by this one attribute or characteristic (Goffman, 1963).
# Organizational stigmas
As stated before, stigmatization has been researched in many fields such as sociology and psychology as well as in management and organizational sciences. Particularly in the light of the economic crisis, stigmas have been mentioned in management papers as an effect of financial distress and bankruptcy on individual organization members (Wiesenfeld et al., 2008; Semadini et al., 2008). However, due to the focus of these studies on the individual members of an organization, these papers do not use the definition of stigma when examining organizations. The term “stigma” is often used as a label or a description for a negatively evaluated event (e.g. bankruptcy), rather than a specific organizational-level construct (Mishina & Devers, 2012).
Other scholars such as Pozner (2008) and Husdon (2008) do focus much more on the effects of stigmas on an organizational level. Pozner (2008) differentiates between an individual stigma and an organizational stigma, by looking at the liability for the stigmatization in the eyes of the observer. Did an employee or manager ignited the stigmatization or can the cause be much better related to organizational systems? Devers et al (2009) elaborate on this by proposing that organizational-level stigmas differ from individual-level stigmas in the types of conditions that stigmatize, how preventable or removable the stigma is, and how pervasive the stigmatizing categories are across contexts to organizational sciences (Mishina & Devers, 2012). In their paper ‘A general theory of organizational stigma’ Devers et al (2009) have created a theory that explains the conditions under which organizational stigmas are likely to arise, how this process unfolds, and the initial effects stigmas inflict on organizations. They defined organizational stigma as “a label that evokes a collective stakeholder group-specific perception that an organization possesses a fundamental, deep-seated flaw that deindividuates and discredits the organization” (Devers et al, 2009 p157). This definition will be used throughout this paper when mentioning organizational stigma.
# Core and event organizational stigma
Hoover (2008) distinguishes between two types of organizational stigma; event and core stigma. The event stigma results from episodic actions such as bankruptcy and the effects normally reduce over time. In contrast, organizations can become core stigmatized due to their own nature and attributes of its business
conduct (Hoover, 2008). These attributes are conflicting with the expectations that social audiences have about the organization (Elsbach & Sutton, 1992). Organizations that social audiences perceive as core stigmatized are for example, strip clubs (Schlosser, 2003), abortion service providers (Hudson, Wong-Mingji, & Loree, 2000), and gambling and tobacco companies (Galvin, Ventresca, & Hudson, 2005). These types of organization face several negative effects due to their core business that they conduct.
# Stigmatized industries
Stigmatization also occurs on an industry level (Galvin, Ventresca, & Hudson, 2005). An industry can be described as a complex interorganziational field of commercial activities (Galvin, Ventresca, & Hudson, 2005). It is a group of organizations, linked to each other by mutual belief systems, core technologies of production and actors (Galvin, Ventresca, & Hudson, 2005). Studying an industry can take place on various levels, such mutual strategies, products and business conduct. If the core business of an industry actor is stigmatized by society, often the other industry actors have to deal with stigmatization as well. For example, the private military industry has to deal with high degrees of stigmatization due to the very core service industry members create (Baum and McGahan, 2013).
The reason that the entire industry is labeled as stigmatized is, amongst other reasons, due to the mutual stigmatized product. Hudson and Okhuysen (2003) studied the core stigmatized industry of men’s bathhouses and found that the service which these organizations provide is associated with a lifestyle (changing sexual relations between men) that society disapproves. In this specific situation, the product or service is the reason for stigmatization. However, in other industries, the product is not directly the reason an industry is stigmatized. The gambling industry for example is stigmatized due to the organizations and locations where the gambling games can be played (Galvin, Ventresca, & Hudson, 2005). These organizations are linked to money laundering, mafia and criminal activities while the actual games are not per definition wrong (Dunstan, 1997). Yet, gambling adversaries have been able to frame the entire industry in such a way that it has become stigmatized and all industry actors, products and services can no longer be seen as individuals but as a representation of the deviating characteristics that caused the stigma to occur in the first place (Link and Phelan, 2001).
However, important to note is that this stigmatization is always in the eyes of certain social audiences. While some audiences will define the adult movie industry as controversial and tainted due to some of the attributes and traits, others will perceive this differently. Employees, partners and customers will not perceive
the attributes of the adult movie industry so negatively, and the negative effects of the core industry stigma will not occur or be less effective.
# How to deal with stigmatization
Firms that experience stigmatization need to act appropriately in order to reduce the impact of the stigmas on their business endeavors. Hudson (2008) describes three ways in which firms can deal with organizational core stigma. All these responses can be implemented simultaneously, depending on the nature of the experienced stigmatization. The first answer to organizational stigma is a strategic response (Hudson, 2008). Strategic responses include (1) the use of specialist strategies-operating in only one or a highly limited number of domains; (2) hiding strategies-avoiding scrutiny from hostile audiences: and (3) challenging strategies-mounting challenges to the expressions of stigma or the values of stigmatizing audiences (Hudson, 2008 p259). With these 3 strategies the negative effects of an organizational stigma can be decreased. The second answer to organizational stigma is a structural response. Firms that experience organizational stigma tend to keep their organizations closed. According to Hudson (2008) this is the effect of resource scarcity related to the stigma in combination with the specialist and hiding strategies mentioned earlier. The final response to organizational stigma is network related. Partners and customers of a stigmatized organization could stop collaborating due to the negative effects of a stigma and the fear of transfer to their own organization or person. Stigmatized organizations are therefore eager to shield their network partners from transferred stigma by disguising their association (Hudson, 2008).
# Summary
To summarize, stigmas can be found in all types of fields and have slightly different meanings depending on the level of analysis. Being part of a stigmatized industry will result in the deindividuation and discrediting of an organization by stakeholders with numerous social and economic consequences. Organizations have to actively manage these group specific perceptions in order to cope with the effects of stigmas and shield their partners. In the next section the hypotheses and conceptual model will be outlined.
2.3. H
YPOTHESISBased upon the findings in the previous sections of this chapter, hypotheses are formulated that collectively provide insight into the impact of stigmas on brand extension strategies and feedback effects. A conceptual framework is created that functions as a guideline and represents the several variables that interact.
# Model development
Based on the literature of branding, feedback and stigmatization, five hypotheses are formulated that collectively provide insight into the direct impact of stigmas on the parent brand image post extension and indirect effect via several mediators and a moderator. A framework (see Figure 4) was developed in order to visually connect the hypothesis and provide insight into the process of answering the research question.
Figure 4 - Framework
The framework is based on the literature of the feedback effect which state that the parent brand image post extension (PBIpe) can be predicted by the parent brand image (PBI), the perceived fit (FIT) and the brand extension attitude (BEA) (Dwivedi et al, 2010; Martinez and Chernatony, 2004; Hem, Chernatony, and Iversen, 2003; Grime, Diamantopoulos and Smith, 2002). The framework is developed on the findings and research model of Dwivedi et al (2010) that posited that brand extension feedback effects can be studied in a holistic framework. The reason for choosing this framework as a guide for developing a new framework for this thesis is that, as far as known to the author, it is the most recent statistically tested framework regarding this matter. Added to the framework of Dwivedi et al (2010) is the independent variable Degree of stigma. Now each of the model’s constructs will be discussed along with the hypothesized effects
# Constructs
The independent variable in the framework is the degree of stigmatization (DOS); how much do consumers rate the product or industry as stigmatized. A stigma on an industry has a negative effect on the organization's performance and that of associated actors (Lacey, 2003; Sutton & Callahan, 1987). Devers et
Perceived fit
Parent brand image post extension Degree of stigmatization
Brand extension attitude H1
H4
H5
al (2009) describe two main effects of stigmatized firms; (1) deindividuation and (2) discredit to the organization. According to Link and Phelan (2001) deindividuation is the effect that occurs when an actor is no longer seen as an individual, but as a representation of the deviating characteristics that caused the stigma to occur in the first place. In other words, the organization is no longer seen as an unique entity with individual characteristics, but is stereotyped in terms of the negative attributes of this category (Selznick, 1984). The second effect of stigmas on business is the discrediting label of stigmatization. Sutton & Callahan (1987) describe this discrediting in the context of the Federal Bankruptcy Code in the US. According to the authors, once firms receive the discrediting label ‘Chapter 11’ this ignites a series of negative reactions of stakeholders towards the firm such as disassociation, disengagement, resource scarcity, reduction in the quality of participation, bargaining for more favorable exchange relationships, and denigrating an organization and its leaders (Sutton & Callahan, 1987). The effects of stigmas on the organization can therefore have severe implications for strategy, marketing and other business conduct.
While much of the research in this field has focused on the effects of stigmas on organizations, less is known about the effects on brands. It can be assumed that the previously described findings also hold for brands active in a stigmatized industry. I would like to argue that consumers do not only disassociate themselves from organizations active in stigmatized industries, but also from brands. Disassociating from both brands and organizations is caused by the incongruence of values of certain social audiences with those of stigmatized industries (Ellemers et al. 1999, Elsbach and Bhattacharya 2001). It can be assumed that the deindividuation and the discrediting actions towards the brand will have implications for the attitude that consumers have towards it. Based on this assumption it can be hypothesized that:
H1: Degree of stigmatization negatively affects the parent brand image post extension.
Likewise, the deindividuation and discrediting actions toward the brand will have implications for the brand extension. Therefore, it can be hypothesized that:
H2: Degree of stigmatization negatively affects the brand extension attitude.
Dwivedi et al (2010) have demonstrated that extending a brand adds new associations to the associative network of the parent brand and thereby incrementally changes the perception of consumers towards it. If the brand extension is appreciated by consumers, positive associations are added to the network and this has a positive effect on the PBIpe (Martinez and Chernatony, 2004) However, if the brand extension attitude is negative, the feedback effect will also be negative. Assuming that consumers experience the industry as stigmatized, the brand extension will therefore by definition also be poorly rated which subsequently lowers
the perception of consumers towards the parent brand. This way DOS has an indirect effect on PBIpe via BEA. This leads to the following hypothesis:
H3: Brand extension attitude has a mediating role on the relationship between degree of stigmatization and parent brand image post extension.
The brand extension literature is unambiguous about the importance of perceived fit. It is seen as the best indicator of brand extension success (Aaker & Keller, 1990). Perceived fit can be defined as the extent to which consumers accept the new product as logical and would expect it from the parent brand (Tauber, 1988). If there are more shared associations between the parent brand and the brand extension, a better fit is perceived (Dwivedi et al, 2010). Aaker and Keller explain the importance of perceived fit through the studies on categorization. The categorization theory implies that branding of a new product with a parent brand, enables consumers to match the product to an existing category based on the associations that they have with the parent brand (Goodstein, 1993). The associations that the consumer has with the parent brand are transferred through the so called ‘spill-over effect’ to the associative network of the brand extension (Balachander & Ghose, 2003). The degree of fit between the parent and the extension determines the amount of this spill-over effect (Aaker and Keller, 1990). For this reason, the bigger the perceived fit, the greater the transfer of attitude from the parent to the extension (Czellar, 2003).
It can be assumed that when a legitimate brand is extended into a stigmatized industry, the perceived fit between the parent brand and the brand extension is low. In the case of low fit, consumers do not understand the extension and confusion is being created. For this reason, it can be hypothesized that:
H4: Degree of stigmatization negatively affects the perceived fit.
In addition, perceived fit also has a feedback effect on the parent brand image. Dwivedi et al (2010) describe that fit strengthens the brand associative network and thereby improves the parent brand image. However, a low fit can result in a negative feedback effect (Martínez & de Chernatony, 2004). By extending a legitimate brand into a stigmatized industry, it can be assumed that the perceived fit is low, which leads to negative associations for the brand extension, which subsequently are transferred to the parent brand as well. It can be argued that a serial mediated effect is present between DOS and PBIpe via (1) FIT and (2) BEA. The higher the DOS, the lower the fit between the parent and the extension, the lower the extension is rated and the more the parent brand is damaged. For this reason, it can be hypothesized that:
H5: Perceived fit (1) and brand extension (2) attitude have a serial mediating role on the relationship between degree of stigmatization and parent brand image post extension.
3. M
ETHOD
This section will describe the methods that were used in order to conduct this research. First will be provided a description of the empirical setting in which the effects of stigmatization were studied. Then, research design and data collection will follow. Finally, the sample and variables will be described.
3.1. E
MPIRICALSETTINGThe empirical setting that was chosen for this study is the Dutch TV-industry. Many of the Dutch broadcasters and production companies are searching for new ways to create revenue due to the decreasing TV advertising budgets (Neele, 2015). Experts indicate that one of the potential new revenue streams for media companies is gambling (Van Oerle, 2015). High profit rates (see Table 2) combined with low investments costs are amongst the reasons why media companies want to enter this industry. However, this revenue stream is controversial within these companies due to the association with addiction, money laundering and crime (Galvin, Ventresca, and Hudson, 2004). In line with these thoughts are the papers of Grougiou et al (2015), Leventis et al (2013) and Hudson (2008) that portray the gambling industry as stigmatized.
There are 2 main concerns that need to be dealt with before the broadcasting companies can launch their own gambling-products. One is the delay of legislation by the government, leading to uncertainty with the broadcasters about when to launch a gambling product. The other concern involves the potential danger for existing brands once they get associated with gambling. Management of the Dutch commercial broadcasters
Table 2 - HIGHLIGHTS ANNUAL REPORT BWIN/POKERSTARS 2014 PER CATEGORY
Sports betting Casino Poker Bingo
Gross revenue 261.5 248.2 93.5 114.0 Net revenue 233.4 199.4 78.7 51.5 Total Revenue 237.1 203.7 81.7 51.9 Gross Profit 180.8 192.2 72.4 47.3 Clean EBITDA 50.1 43.5 7.9 11.8 Clean EBITDA margin 21.1% 21.4% 9.7% 22.7%
is afraid that the negative image of the stigmatized gambling industry might spill-over to their legitimate TV-brands. This fear for spill-over effect is the main reason that this particular empirical setting was chosen. The study was partly conducted for RTL Nederland; the biggest Dutch commercial TV broadcaster. RTL Nederland was interested in the effects of stigmatized industries on parent brands and brand-extensions. The research specifically focused on the effects of extending an RTL brand with different gambling games, namely soap-opera Goede Tijden, Slechte Tijden. The brand was hypothetical extended into the gambling industry with a branded Bingo-, Slots- or Casino- proposition. As stated before, the stigmatization of the industry caused that all industry products and services can no longer be seen as individuals but as a representation of the deviating characteristics that caused the stigma to occur in the first place (Link and Phelan, 2001). Bingo, slots and casino are the games that represent a stigmatized industry.
Some ethical issues can be raised by studying this specific industry. Should TV broadcasters start their own gambling games? Are these gambling games broadcast during the day when children are watching television? These and other ethical issues were discussed with RTL Nederland. Their response was that RTL wishes to examine its opportunities and make a well informed decision on whether to launch gambling products. Examining the effects of gambling on the existing brands is part of the entire decision making process.
3.2. R
ESEARCHDESIGNThe research was designed with the objective to test whether extending an RTL brand with a gambling game would have a negative, positive or no effect at all on the initial brand image. In order to achieve this objective, to answer the research question and to test the hypothesis, an experiment was conducted. An experiment is a research method that can study causal links between variables (Saunders, 2009). It looks at the change in one independent variable, and sees if this produces a change in another dependent variable (Hakim 2000). By using a control group in the study, the measured effects can be quickly related to the treatments (Saunders, 2009). This makes experiments one of the most effective ways to test hypotheses.
To acquire the data for the experiment a self-administered online survey was distributed online via social media, email and forums between 06-07-2015 and 20-07-2015. This data collection technique requires the respondents to individually answer a set of questions in a pre-determined order without an interviewer being present (Saunders, 2009). Most of the participants were recruited through social media and could participate through a link of Qualtrics. Qualtrics is an online tool in which surveys can be quickly and easily distributed
via the Internet. Respondents had to have a computer, laptop or mobile device with internet connection at their disposal in order to participate.
In order to make sure that not only young adults from the writers own personal network would respond, an active search for an older audience on online forum was conducted. By posting on forums such as FOK.NL, Radar.nl, the Koffietijd Facebook-page an older group of participants was reached. Another stimulant for people to participate was their chance of winning a 15 euro voucher for De Bijenkorf.
On the 21 of July 2015, the survey was closed and respondents could no longer participate in the study. To analyze the data, the statistical program IBM SPSS (v22.0) was used.
3.3. S
URVEYFLOWThe flow of the survey was designed specifically to test the effect of the treatments in the four categories. The groups were randomly formed: (1) Bingo, (2) Slots, (3) Casino and (4) Control. The random selection and allocation of the conditions was done by Qualtrics. All groups were asked to assess statements about the parent brand in order to establish the PBI.
After the PBI was established, group 1, 2 and 3 were provided information about the hypothetical extension of the brands in the gambling industry. To be more precise, the extension of the brand with one of the three treatments; bingo, slots or casino. Respondents were asked to rate the fit (FIT) between the brands and games. Subsequently, the three groups were asked to assess the actual brand extension (BEA). Then, in order to measure the PBIpe, respondents were asked to re-evaluate the parent brand with the knowledge of the extension in mind. Next, respondents were asked to rate the degree of stigmatization (DOS) of their assigned game while the control group rated all games. Finally, age, gender and education were asked as control variables for the study.
3.4. M
ATERIALSThe survey contained a total of 67 questions and statements measuring 5 variables. 7-point Likert scales were used to determine the agreement of the respondents with the statements; a score of 1 = ‘strongly disagree’ and a score of 7 = ‘totally agree’. The entire survey can be found in appendix 7.1. The items that formed these variables are derived from or based on the literature discussed in the literature section such as the research of Dwivedi et al (2010) as well as on the CBBE-model of Keller (2009). The reason to go for this
approach is that these statements and scales have been proven and used in several academic papers. Furthermore, this approach creates an opportunity to compare results with existing literature, making it more relevant to the academic world.
Parent brand image and Parent brand image post extension – The scales for the parent brand image
and parent brand image post extension are two somewhat identical scales. The scales consist of 9 items and measure the initial brand image before extension. Later, in order to measure the effect of the treatment, respondents were asked to re-evaluate the parent brand with the knowledge of the extension in mind. In order to do so, statements that were used to measure the PBI were slightly adjusted and used to measure the PBIpe. The scales are based on items originating from various studies such as Aaker & Keller (1990), Martinez & Chernatony (2004) and the scale used by BAV Consulting (BAV Consulting, 2015). Awareness, attributes and attitude of the brands are measured by the scale.
FIT – The scale of Dwivedi et al (2009) was used in order to measure the fit between the parent brand and
the extension. Respondents were asked to rate the fit between the brands and the games. The scale consists of 4 items and measures Tauber’s (1988) definition of FIT: ‘consumers accept the new product as logical and would expect it from the parent brand’ (p. 28). By using statements such as ‘According to me, the decision to launch GTST Bingo is very surprising’ or ‘The launch of GTST Slots in the market was expected’, the scale provides respondents the opportunity to rate the fit on an emotional level. This contrasts other studies that focus on the perceived category and image fit (Park et al., 1991).
Brand extension attitude – To measure the brand extension attitude scales from Dwivedi et al (2009) were
altered for the purpose of this study. The 5 items measure the general attitude towards the extension, assumptions about the quality of the extension and the likelihood that respondents would try the extension when they were looking for a product in this category (Aaker & Keller, 1990).
Degree of stigmatization - Items to measure this construct were based upon the research findings of
Thomas & Lewis (2012). They conducted an extensive study on gambling and the perceived stigmatization of extensive gamblers. Based upon their key results, the 6 items were made to measure the degree of stigmatization and include statements concerning dangers, image and overall attitude towards the various gambling games.
3.5. S
AMPLEThe sample size of this study was 152. 152 respondents started the survey and 146 finished it (N=146). This amount was necessary since the study has four different categories, all with their own treatment. According to Saunders (2009), in order to perform a statistical analysis the sample size per category should be at least 30. Based on this rule of thumb 120 respondents was the bare minimum needed.
As stated before, the respondents were randomly assigned to one of the categories by Qualtrics. The distribution over the different categories can be seen in table 3.
In order to participate, respondents needed to be at least 18 years old, since this is the minimum age for gambling in the Netherlands. Furthermore, they needed to be able to understand Dutch since this is the main language in which RTL communicates.
The respondents were predominantly high educated females of their late twenties. 60.9% of the respondents are female. The age range of the respondents was from 18 until 78 with an average age of 32.4. The sample is predominately high educated; 52.4% went to university, 26.0% went to HBO and 14% went to MBO or lower.
TABLE 3 – Sample
Frequency Valid Percent Cumulative Percent
GTST Bingo 30 20.5 20.5
GTST Slots 30 20.5 41.1
GTST Casino 32 21.9 63.0
Control 54 37.0 100.0
4. R
ESULTS
This section will provide an overview of the research findings. First a reliability analysis will be provided, followed by an overview of the differences between treatments. The last part of this section is dedicated to hypothesis testing.
4.1. R
ELIABILITYANALYSISFirst a check for variables with missing data was conducted by using frequency tests. For a few case variables there was missing data. By list wise deletion I made sure to only analyze cases with complete data. Then, I recoded seven items due to its counter indicative origin. 16.4, 16.7, 21.3, 26.3, 29.3, 71.3, 73.3 and 75.3 were inversely recoded into 16.4r, 16.7r, 21.3r, 26.3r, 29.3r, 71.3r, 73.3r, 75.3r to match other results. In order to check if the different items can be combined into the constructs parent brand image, brand extension attitude, perceived fit, degree of stigma and parent brand image after extension, for all the treatments, Cronbach's Alpha was tested to check internal consistency. The Cronbach’s alpha for the parent brand image (9 items) is α= 0.809, indicating that 80.9% of the scores are internally consistent for the construct PBI. For the treatment bingo, brand extension attitude (5 items α=0.846), perceived fit (4 items; α=0.83) both had high levels of internal consistency. Questions associated with the degree of stigma had low internal consistency. A double check for counter indicative items was done, but none were evident. Deleting 3 items increased Cronbach’s alpha from 0.322 (9 items) to 0.607 (6 items). This is still a questionable internal consistency, however acceptable according to George & Mallery (2003). The final construct to check for internal consistency is the parent brand image after extension (PBIpe). In order to compare the results of PBI with PBIpe, the items needed to be the same. A check for the same 9 items led to a Cronbach’s alpha of 0.812, indicating good internal consistency.
The same constructs were checked for the other treatments and the results can be found in table 4. It shows that for all the treatments, sufficient amounts of internal consistency were found for the items associated with PBI, BEA, FIT, DOS and PBIpe.
Next, mean scores were created in order to test the hypothesis. An overview of the development of mean scores for all treatments can be found in Appendix 7.2 Development of variables.
4.2. D
IFFERENCEBETWEENTREATMENTSAn overview of mean scores is provided to see how the respondents rated the parent brand image after an extension into a stigmatized market. This overview provides insight into the differences between the 3 treatments (see Figure 5). At first sight, it appears that all treatments have a lowering effect on the parent brand image. It also appears that the treatment Slots has the biggest negative effect on the parent brand image, followed by Casino and Bingo. To specifically test these first interpretations, a one-way ANOVA was conducted in order to test the effects between the groups.
A one-way ANOVA was conducted to determine if the parent brand image post extension (PBIpe) was different for groups with different stigmas. In order to conduct this test a new variable (Mean PBIpe Total) was formed consisting out of the Mean PBIpe for all treatments. To conduct this test, several assumptions must be met before testing is allowed. First, no significant outliers with values greater than 1.5 box-lengths
TABLE 4 – RELIABILITY ANALYSIS
Treatments PBI Items BEA Items FIT Items DOS Items PBIpe Items
Bingo .809 9 .846 5 .830 4 .607 6 .812 9
Slots .809 9 .907 5 .980 4 .582 6 .876 9
Casino .809 9 .829 5 .887 4 .854 6 .864 9
Control .809 9
Figure 5 - Parent brand image post extension per treatement (PBIpe)
0 1 2 3 4
from the edge of the box were detected. Secondly, PBI was not normally distributed for treatments, as assessed by Shapiro-Wilk's test (p < .05). However, due to the robustness of the One-way ANOVA test, it was decided to proceed anyway (Field, 2009). Finally, there was homogeneity of variances, as assessed by Levene's test for equality of variances (p = .378).
The parent brand image was statistically significantly different for different treatments, F(3, 142) = 18.350, p < .001, η2=0.279. PBIpe score decreased from the Control group (M = 3.24, SD = 0.79) to the Bingo group (M = 2.68, SD = .78), to the Casino group (M = 2.07, SD = .90) and Slots group (M = 1.92, SD = 1.18), in that order. Tukey post hoc analysis revealed that the mean decrease from control to Bingo (-.56, 95% CI [-1.09, -. 02]) was statistically significant (p = .037), as well as the decrease from control to Casino (-1.17, 95% CI [-1.67, -.64], p < .005) and the decrease from control to Slots (-1.31, 95% CI [-1.84, -.78], p < .005). See table 5 for an overview of all the mean differences. Since all the groups rate the PBIpe more negative than the control group, it can be concluded that associating with one of the stigmatized gambling games has a negative effect on the parent brand image. In order to better understand this relationship and to test the hypothesis, several regression analyses will be conducted in the following section.
Table 5 – One-way ANOVA
Sum of
Squares df Mean Square F Sig.
Between Groups 44.861 3 14.954 18.350 .000 Within Groups 115.719 142 .815 Total 160.580 145 Multiple Comparisons (I) CategoryGT ST (J) CategoryGTS T Mean Difference (I-J) Std. Error Sig. 95% Confidence Interval Lower Bound Upper Bound Tukey HSD Control Bingo .55746* .20556 .037 .0231 1.0919 Slots 1.31440* .20556 .000 .7800 1.8488 Casino 1.16718* .20139 .000 .6436 1.6907
4.3. H
YPOTHESISTESTINGIn order to study the hypothesis, it was decided to focus on the casino treatment. The casino treatment group has the constructs with the highest internal consistency and is therefore most reliable for further research. A correlation matrix was assembled which includes the means, standard deviations, inter correlations of the studied composites and their reliability coefficients (Table 6c). In appendix 7.3 the correlation matrix for Bingo (Table 6a) and Slots (Table 6b) can be found as well.
In order to test the first hypothesis and see if a stigma can negatively influence the parent brand image post extension, more than the variables of PBI, FIT and BEA can and after controlling for gender and education, a hierarchical regression analysis was implemented. PBI, FIT and BEA are proven predictors of PBIpe (Dwivedi et al, 2010). The Durbin-Watson test indicated the residuals are slightly positively correlated (1.12) but still acceptable according to Field (2009) (1<X>3). Approximately linear relationships between the independent and dependent variables were shown in the partial regression plots. The P-P Plot indicates homoscedasticity since the spread of the residuals is not increasing or decreasing severely as you move across the predicted values. Tolerance and VIF indicate that there are no collinearity problems (all Tolerance values > .1). No outliers were detected by case wise diagnostics. By looking at the histogram, it can be inferred that the residuals are normally distributed. Also the mean and standard deviation have values of approximately 0 and 1 indicating a normal distribution. The assumptions of linearity, independence of errors, homoscedasticity, unusual points and normality of residuals were met and analysis can start.
TABLE 6c.1 | CORRELATION MATRIX Casino
Constructs Mean Std. Deviation 1 2 3 4 5
Mean PBI
M_PBI or PBIpe control
3.1681 0.76104
Mean BEA Casino M_BEA_C 1.7538 0.73479 0.162
Mean FIT Casino M_FIT_C 1.7436 0.9554 -0.26 0.275
Mean DOS Casino M_DOS_C 5.3977 1.07909 0.045 -.384* 0.108
Mean PBIpe Casino M_PBIpe_C 2.1054 0.86328 .458** .364* -0.07 -0.043 ** Correlation is significant at the 0.01 level (2-tailed)
