Cloaking or flaunting? : the effects of consumer characteristics and involvement on consumers’ willingness to pay

(1)

Cloaking or flaunting?

The effects of consumer characteristics and

involvement on consumers’ willingness to

pay

Anne Schouten 6143903

Msc in Business Studies

Entrepreneurship and Management in the Creative Industries Supervisor: F. Situmeang Msc.

(2)

Abstract

In this research we looked at the informational aspect of advertisements in relation to the social device. Cloaking and flaunting strategies where introduced by Yoganarasimhan (2012). The cloaking strategy contains of making your product more exclusive, while the flaunting strategy shows off the popularity of your product. This research was the first to

experimentally test both strategies in order to see their effect on willingness to pay.

Differences in consumer characteristics will be related to the strategies, regarding consumers seeking for distinction versus consumers that seek for group cohesion. We also checked whether involvement was of any influence. Using the Veylinx auction, 463 participants participated in one of the three treatments. The results did not show significant results in order to confirm the theories on cloaking and flaunting. However, significant results where found with regard to the consumer characteristics. The consumer characteristics explained that the auctioned product was not perceived as exclusive by the participants who where seeking for distinction, wherefore their willingness to pay decreased.

(3)

1.0 Introduction

Exclusive products are often sold with a large price premium. Perceiving a product as exclusive could depend on advertising the product, which makes the chosen marketing strategy important (Atwall & Williams, 2009). Promoting your products is one of the basic marketing activities that most companies execute in order to generate (more) sales (Kotler & Armstrong, 2010). Promoting your product in order to signal exclusivity can take a completely different form than the more standard strategies.

To illustrate; there is a relatively new trend on the rise, called hidden bars. The concept is nothing new, but the media attention in the Netherlands has just started to spread about this trend (RTL News Reporter, 2014). Where hidden bars used to occur since alcohol was forbidden, it has become a “signalling-your-high-social-status” in present days. Where the hidden aspect became a strategic move in order to make the bar more exclusive.

This hidden phenomena is not just happening within the bar scene, but it is also becoming more common for restaurants, clubs, art galleries, fashion stores and articles and so on (Yoganarasimhan, 2012, p. 88). Strategically withholding information is a certain marketing strategy that keeps the social externalities of consumer behaviour in mind and takes the role of information into account in order to make a product more exclusive, as people can show off their good taste (Yoganarasimhan, 2012). While this topic is still underexposed, it can be an important beneficial strategy for certain companies and therefore requires further research.

Much research has been done into consumer behaviour, however there is a lack of focus on the social influences in combination with the role of information. For example Singh, Balasubramanian and Chakraborty (2000) do look at the role of information regarding different advertisements, but fail to include the social roles of consumers in this. Wernerfelt (1990) also looks at advertisement and the role of brand choice and quality signals. He states

(5)

that “cheap talk” advertising can serve as a coordinating cue for consumers to signal that they belong to a certain group. Kuksov (2007) looks at the brand value in social interaction and notes that the associations of the brand users with the brand image may affect the choice of consumers for this specific brand, especially when others observe the brand consumption. There is also research that looks deeper into psychological consumption behaviour processes, like the paper of Nord and Peter (1980), but still they ignore social interactions and influences towards each other. Luxury goods theories do provide more useful insights regarding consumption driven by social aspects (Wu, Lu, Wu and Fu, 2012; Vickers and Renand, 2003; Nia and Zaichkoswky, 2000; Silverstein and Fiske, 2003). So research on the social interactions has been done, but none of them explain the social roles in relation to the effect of signalling information strategies and whether such strategies were effective.

However, as briefly mentioned above, Yoganarasimhan (2012) started to bring these differences in social externalities regarding the role of information in perspective, which is the beginning of filling the gap. He came up with the cloaking and flaunting marketing strategies. According to Yoganarasimhan (2012), a cloaking strategy consists of making your product more exclusive by using the right information or strategically withholding information. The cloaking marketing strategy goes against the more standard marketing strategies where companies want to promote their product excessively to a broad audience, in order to get as many sales as possible (Kotler and Armstrong, 2010). When companies use a cloaking strategy they rely on their sophisticated consumers that tend to look for certain products that are not easily accessible to everyone. The goal of this strategy is to make the product more valuable and attractive to certain types of people, which can lead to a higher willingness to pay. Yoganarasimhan (2012) also identified the opposite of Cloaking. When companies specifically indicate the best-selling items and share more information than the standard information that is shared with a product, it is called a Flaunting strategy. Implementations of

(6)

these strategies are visible within for example the fashion industry and are described in the literature review.

Along with the distinction between cloaking and flaunting strategies is the distinction between two different types of consumers: snobs and omnivores (Roose, Eijck and Lievens, 2012). As Yoganarasimhan (2012) already discussed there are certain types of people that add more value to a product when it is not widely available. By buying such a product of higher value they feel more sophisticated. This phenomenon is called conspicuous consumption and is explained by Veblen’s Theory of the Leisure class (Veblen, 1899, p. 97). Kuksov (2007) also mentioned that recently, brand images are used as forms of personal statements more frequently. Those consumers that like to have a product with a high-brow image, that is not very accessible and which distinguishes them from the crowd, are called snobs. On the other hand, the omnivores are people that consider themselves to be open to all sorts of culture and goods and they weaken both social and cultural boundaries. Since they are open to all sorts of goods, they are willing to try out everything, and they do not care if the products to be consumed are high-brow or low-brow. They do not care about distinguishing themselves in order to feel special, but are also fine with group behaviour (Kuksov, 2007). Yoganarasimhan (2012) proposes that snobs can be brought into relation with the cloaking strategy, whereas omnivores can be brought into relation with the flaunting strategy.

Besides the different types of consumers, Ippolito (1990) found that the ability of signalling your quality of the product throughout the advertising is always due to the bonding principle. Ippolito (1990) signals the importance of consumer involvement that is involved with perceiving the information and the quality of the information regarding products. Consumer involvement is widely recognized as a variable relevant to advertising strategies, and therefore their response to the information provided by the company. A difference in the decision process and the search for information depending on the level of involvement of the

(7)

consumers can be acknowledged (Laurent and Kapferer (1985). This indicates that consumers with a higher involvement actively seek information, which makes it a moderating variable regarding the cloaking and flaunting strategies since a cloaking strategy could make it hard to find the information.

To summarise, previous literature looks at social influences and the role of information, but fails to include the effect of different information strategies on social interactions and buying behavior. Yoganarasimhan (2012) was the first to identify two different strategies regarding different use of information; the cloaking and flaunting strategies. However no research has been done on different consumer types, like snobs and omnivores that might go along with this strategies. Another important factor that is left out in previous research is consumer involvement, which can play a significant role in looking for information. In order to contribute to the literature as well as for practical implications we will search for an answer to the following question: How would consumer characteristics and involvement moderate the effect of cloaking and flaunting marketing strategies on the consumers’ willingness to pay? In order to answer the research question we are doing experiments with cloaking and flaunting strategies. Experiments are used since we can specifically look at the information aspect provided for both strategies, by manipulating the variables in the experiments. We will explore the effect on the consumers’ willingness to pay for both strategies. First the literature will be explored to deeper extent in paragraph two, thereafter there will be a methodology section, followed by the results. Finally, the discussion and conclusion will discuss and summarise the findings.

(8)

2.0 Literature review

In this literature review the theory constructs of the cloaking and flaunting strategies will be further explored, we will be looking at the social device concerning snobs and omnivores and other relevant theories regarding social processes and we will broaden the view on the importance of consumer involvement.

2.1 Cloaking and Flaunting

Yoganarasimhan (2012) created a model that explains the role of information in the market of fashion. According to Yogonarasimhan (2012) an important attribute of fashion is that it is consumed conspicuously. A lot of research failed to include the social externalities that come with conspicuous consumption (Wernerfelt (1990), Kuksov (2007), Amaldoss and Jain (2005), Sorensen (2007), Tucker and Zang (2011)), where conspicuous consumption means that individuals consume highly conspicuous goods and services in order to show their high social status and their wealth (Bagwell and Bernhem, 1996). The fashion industry is a good example since there is a high degree of consumption not only for utilitarian purposes, but it is more than just their inherent utility and fashion can be seen as a social device. This implies that information that is revealed by the firm affects the signalling game between consumers, their willingness to pay, and by extension, the firm’s profit (Yoganarasimhan, 2012, p.76).

Competing on the basis of social status has it fundamentals in the leisure theory of Veblen (1899). This theory was found in the higher stages of the barbarian culture where he focused on such uneconomical behaviour to identify that there are shortcomings of classical economic theories (Bird and Smith, 2005). In order to signal that people are from a high status social rank, they will buy “unnecessary” goods that the lower classes cannot afford. This is the beginning of the so-called conspicuous consumption as we already introduced. The conspicuous consumption is embedded in signalling theory. Bird and Smith (2005, p. 222)

(9)

describe Signalling theory as: “Signalling theory provides a way to articulate idealist notions of the intangible social benefits that might be gained through symbolic representations of self with more materialist notions of individuals as self-interested but socially embedded decision makers (Bird and Smith, 2005, p. 222). Another aspect of this signalling game is the willingness to pay. Consumers are willing to pay a higher price for a functionally equivalent good if they will perceive a higher status by it (Bagwell and Bernheim, 1996). Since the cloaking strategy is mostly derived from the signalling theory, we will first elaborate on the cloaking strategy, where after we will continue with the flaunting strategy.

2.1.1 Cloaking

Due to this signalling game and the role of information Yoganarasimhan (2012) came up with the cloaking and flaunting strategies, as already briefly introduced in the introduction. However he only focused on the fashion industry where the signalling theory is of great importance due to the social device as we just explained. The cloaking strategy is built upon the given that consumers want to distinguish themselves from the crowd. Therefore the strategy comprises of making the product more exclusive. Companies can make their product even more wanted, by for instance strategically withholding information. The fashion industry has some extreme examples that can illustrate this aspect of the cloaking strategy. YSL for instance, had a bag called the Muse Two, which became the most wanted IT bag, while they had not even specifically advertised the bag. They just ran advertisements for their bags as if they were all the same. And there is l’Oreals’ Kiehl who eschews media advertising, which comes down to the same as cloaking information. They are aimed at “people who know” (Catry, 2003). The most extreme example comes from the Japanese designer Takahiro Miyashita. He had to move his store three times because he wanted to stay off the radar, so only people looking for the store could find it (Yoganarasimhan, 2012, p. 74).

(10)

The withholding information aspect ensures a certain “hard to get” factor to the product and although Yoganarasimhan (2012) focuses only on fashion, this strategy might also be effective in other industries as long as there are consumers that want to distinguish themselves. The “hard to get” factor has similarities with the “rarity” factor described by Catry (2003). The fact that the product is rare makes is more wanted. This is also confirmed by Lynn (1991), who states that according to commodity theory, scarcity enhances the value (or desirability) of anything that can be possessed. Lynn (1992) came up with the Scarcity-Expensiveness-Desirebility (S-E-D) model, suggesting that people believe that scarce goods costs more, and where a higher price indicates higher quality and status to offer an explanation for scarcity’s enhancement of desirability (Wu et al., 2012, p. 263). This desire is explained by Snyder and Fromkin (1980) in terms of uniqueness theory. They state that people have a social desire to conserve some sense of uniqueness. In line with this social desire, Catrys’ (2003) research also mentioned that this is very important for people that want to stand out from the mass. While the rarity aspect as Catry (2003) describes it overlaps with the cloaking strategy, he notes that this rarity of products goes along with more personal marketing campaigns like public relations programmes, selective mailings and special events. Scarcity of products may be well managed by companies through the use of “special series” and “limited editions” (Catry, 2003, p. 13). Limited editions bags for example can provoke waiting lists, since the limited edition aspect intensifies the desirability for owning the products (Wu, Lu, Wu and Fu, 2012). According to Wu et al (2012) this phenomena illustrates the psychological effect of scarcity, and marketers can use these effects in their strategies. Like Catry (2003) already mentioned, marketers use phrases like “limited release”, “only until supply last” and “limited time only” to signal the scarcity of the product in order to make it more valuable (Wu et al., 2012, p. 263).

(11)

Since the cloaking strategy is built upon the fact that consumers want to distinguish themselves, theories on luxury goods also add useful insights and therefore we will examine a few. To start with Davies, Lee and Abonkhai (2011), they mention that buying a luxury good goes along with aiming to increase personal well-being and self-pleasure through peer and self-identification of the creation/ continuation of personal brand identity (p. 5). Vickers and Renand (2003) add to this by stating that luxury goods are based on symbols of personal and social identity, whereby the primary value of luxury goods is psychological. Nia and Zaichkoswky (2000) indicate this psychological aspect to be the crucial point on which luxury goods can be differentiated from commodities. According to Vickers and Renand (2003) luxury goods can be defined in terms of a mix of components of functionalism, experientialism and symbolic interactionism. Symbolic interactionism explained 48 percent of the variation in their study, and can also be linked to this study. The component relates to demonstrating internally generated needs, particularly suggesting a high degree of self-enhancement and a desire for membership. Wilcox, Kim and Sen (2009) also look towards the symbolic meanings. They state that consumers buy luxury goods in order to display their individual identity and convey social image. According to Bian and Forsythe (2012) attitudes play an important role in this social image. Attitudes can serve social-functions as self-expression and self-presentation (DeBono, 1987; Grewal, Mehta, & Kardes, 2004; Katz, 1960; Shavitt, 1989; in Bian and Forsythe, 2012). Where self-expression reflects their central values and beliefs and self-presentation reflects the social image they want to present. Silverstein and Fiske (2003) are also in line with this, they consider them “new consumer’s needs”. Whereas they highlight that today (read 2003, when the article was published) consumers have more money and a greater desire to examine their emotional side. Consumers seek products that make positive statements about who they are and who they would like to be (Silverstein and Fiske, 2003, p. 54). Like many others Carcano, Corbetta and Minichilli

(12)

(2011) mention the importance of the social meaning of luxury goods. However they also note that the luxury concept can range from a “hard” to “soft” luxury approach (Carcano and Ceppi, 2010; in Carcano et al., 2011). This means that it can range from unique custom-made products or small series offered to a select group of clients; to the so-called “mass luxury”, which refers to the industrialized manufacturing of line of products that are marketed to a wide variety of customers (Carcano et al., 2011, p. 42). Where the latter (“soft” approach) can be brought into relation with the flaunting strategy, the first approach (“hard” approach) corresponds with the underlying constructs of the cloaking strategy. Carcano et al. (2011) state that external cues like referring to rarity, exclusiveness and heritage are used in the “hard” approach. Wiedmann, Hennings and Siebels (2009) mention, besides all other aspects of luxury consumption we already mentioned, the dimensions of the luxury value. They state that consumer behaviour in relation to luxury brands can be explained by interpersonal and personal aspects and situational conditions. Where interpersonal aspects are for example snobbery and conspicuousness, and are also main factors in this study.

Summarizing the luxury goods literature, social status is a main factor, which is exposed in each study. The underlying principles of the cloaking strategies, like seeking distinction, finding something unique and status-enhancing, has much overlap with the luxury goods theories.

2.1.2 Cloaking and price mechanism

Price technically there are divergent views. On the one hand companies that cloak information keep their price levels the same, as YSL for example, who priced their IT bag at the same price as other bags. While on the other hand companies that entail a cloaking strategy ask for far higher prices (Yogonarasimhan, 2012). Bagwell and Bernhem (1996) explain the role of pricing in the conspicuous consumption by looking at budget versus luxury brands. They state

(13)

that luxury brands sell products of quality that are not necessarily superior to products of budget brands for far higher prices. Consumers that are seeking to advertise their wealth are willing to pay this premium price. Velthuis (2003) goes somewhat more in-depth into the meaning of prices. Economic theories see prices as an expression of a consumer’s willingness to pay. However neoclassical micro-economics contribute to these economic theories by involving signalling theories, since it can be difficult to access information about the quality of the product in some cases. Joseph Stiglitz argues that ‘‘price serves a function in addition to that usually ascribed to it in economic theory: it conveys information and affects behaviour’’. So prices are used to judge the quality of a good and this signalling effect is not confined to uninformed parties who lack other sources of information (Velthuis, 2003, p. 186). Many authors like, Erickson & Johansson (1995), Lichtenstein, Bloch & Black (1988), Tellis & Geath (1990) have demonstrated that the price of a good may have a positive role concerning the quality perception (Wiedmann et al. 2009, p. 629). Status-conscious consumers also use the price as indicator for prestige. Therefore prestige-pricing (setting a price high) is used as a strategy to suggest high quality or status, in order to make a product more desirable (Wiedmann et al., 2009). Another research that is in accordance with this line of reasoning is the research of Catry (2003). He notes that price is one of the first signals that is picked up by the market and the higher the price, the more selective the purchase seems to be (Catry, 2003, p. 14). The results of the study by Wu et al. (2012) show that the higher the perceived scarcity of a product, the higher the assumed expensiveness of the product will be. They also state that increased uniqueness will result in higher prices and consumers are willing to pay more to express their uniqueness, whereas these researchers also found evidence for the fact that people see price as one of the extrinsic cues to signal quality (Wu et al., 2012, p. 271).

(14)

Deriving from luxury goods theories there is Mc Daniel’s (1993) Exclusive Value Principle, which states that a luxury good’s price consists of the sum of the utilitarian value and an exclusive value premium (e.g. external factors such as advertising) (de Barnier, Rodina and Valette-Florance, 2006). However this argument can serve both the cloaking and flaunting strategies, since they are not specific in what kind of external factors and advertising methods this exclusive value premium is based upon.

Since the cloaking strategy in considered to make the product more exclusive, sophisticated and unique, this is likely to have impact on the price. According to Yoganarasimhan (2012) there are companies that ask a far higher price for products sold using a cloaking strategy, which can signal a high quality / status. Theories of Bagwell and Bernhem (1996), Velthuis (2003), Catry (2003) and Wu et al. (2012) are also in line with this higher price higher value idea. However, since we will be using auctioning, meaning that the price is not showed, we would rely on mostly on the exclusive unique point that Wu et al. (2012) made. Therefore we will suspect that a cloaking strategy will lead to a higher willingness to pay, which we will test with the following hypotheses:

H1: A cloaking strategy will lead to a higher willingness to pay

2.1.3 Flaunting

After explaining the cloaking strategy in more detail, we will now elaborate more on the flaunting strategy. Contrary to cloaking information, there are several fashion houses on the upper hand that flaunt their information, to signal their most wanted and prestige items. This strategy entitles their consumers with all information possible, like which item is the bestseller, which items are new arrivals and so on. An example is given by Ralph Lauren, where you can find specific subcategories on their website that are named “bestsellers” and “new arrivals”. A similar concept is researched by Tucker and Zhang (2011), who call this

(15)

popularity information. They state that a product may be popular by having a perceived high quality or because it caters to a broader range of tastes (Tucker and Zhang, 2011, p. 1). So looking at this definition of Tucker and Zhang (2011) in terms of popular information, this goes along with the information provided by a flaunting strategy. The results of their study states that niche markets benefit most from highly salient popular information comparing to mainstream products, because the same level of popularity implies higher quality for the niche market products compared to the mainstream products (Tucker and Zhang, 2011, p. 27).

2.1.4 Flaunting and price mechanism

When it comes down to pricing, the companies’ price their flaunting items at far higher prices. According to the theory of Velthuis (2003) these higher prices could be a signal for high quality. However, Yoganarasimhan (2012) actually states that due to the flaunting strategy the products will decrease in value, since there is a lot of information available and they are easy to find and buy. So people that are seeking to advertise their wealth and sophistication are less interested in products that everyone can easily buy, which is the case in the research of Yoganarasimhan (2012) since he is looking at conspicuous consumption. On the other hand, when we are looking at the technology industry for example, this flaunting strategy might have the opposite effect. Where everyone wants the best technology they are certainly willing to pay a high price and this would not lead to a decrease in value of the product. This already indicates the differences between consumers, which we will discuss in the next sections. Another theory that looks at pricing schemes is the price-advertising model of Zhao (2000). His study adds to the literature by looking at advertising spending not only as a signalling device, but also as an informational device. Zhao (2000) mentions that advertising is used as a signalling device for quality, since only high-quality firms can afford it. But is does matter how they advertise, because without the correct signalling approach the

(16)

advertising is seen as “burning money”. It is important to raise awareness and signal quality. Applying this theory to the flaunting strategy, companies that spend money on advertising to flaunt information about their products are more likely to be perceived as high quality, which will lead to a higher willingness to pay (Velthuis, 2003).

Although there are some contradictions between theories, we believe that a flaunting strategy will also lead to a higher willingness to pay in general. Relying on the research of Velthuis (2003) and Zhao (2000) a higher price suggests a higher value and advertising itself shows that the product should be of high quality. While Yoganarasimhan (2012) only focuses on conspicuous consumption, we believe this would lead to a lower willingness to pay for conspicuous consumers. However the conspicuous consumption is only a niche market. Therefore we will test the following hypotheses:

H2: A flaunting strategy will lead to higher willingness to pay.

2.2 Social Device and Consumer Characteristics

Consumer behaviour is a very broad and well-studied area. There are so many variations in terms of consumers’ values, needs and purchase behaviour. When we look at consumer needs we can broadly outline this into functional, emotional and symbolic purposes (Kim, Forsythe, Gu & Moon, 2002, p.481). When products are used to express consumers’ personality, social status or affiliation we are looking at the symbolic purposes, which we will be concentrating on in this research. Building on the social device of fashion consumption, as mentioned earlier, Schimmel (1904, p. 133, in Yoganarasimhan (2012)) argues that fashion fulfils two contrary social needs, namely the need for group cohesion and the need for distinction. Roose, Eijck and Lievens (2012) also name two different types of cultural consumers in their study, the snobs and the omnivores, which are in line with the two contrary social needs as just mentioned. Besides the distinction between snobs and omnivores, Roose et al. (2012) also

(17)

highlight the fact that different lifestyles can influence the way we look at the different fractions of the dominant positions. Cultural capital, economic capital, gender, age and education are all determinants of a consumer’s lifestyle (Roose et al., 2012, p. 497). And different lifestyles can determine whether someone is more likely to be a snob or an omnivore.

2.2.1 Snobs and distinction

A snob can be explained by the desire to consolidate social hierarchy by defending clear-cut cultural boundaries as part as an on-going class struggle (Bourdieu, 1984; in Roose et al., 2012). The snob is looking for distinction, by focusing on exclusive items (Dubois and Duquesne, 1993). They want people to know that they are consuming this specific positional good, so they can signal that they have good taste and are more high-brow consumers (Roose et al., 2012, p. 491). This type of consumer is also described by Amaldoss and Jain (2005) in terms of consuming conspicuous consumption goods. Besides the desire to signal their wealth, another line of reasoning is that people buy products to be different from other consumers. Uniqueness is therefore an important product characteristic to snobs. Amaldoss and Jain (2005) found that snobs buy a product when the price rises. And when more consumers are buying the product, the value of the product decreases since it does not have the same signalling effect anymore. This is also in line with the reasoning of Bagwell and Bernhem (1996), Velthuis (2003) and Catry (2003) as we described in paragraph 2.1.2. To relate the snob back to the cloaking and flaunting strategies, based on the literature, we can assume that snobs would rather want products that are sold using a cloaking strategy, since this strategy indicates that the value of the goods sold is higher and of more status. Therefore we come to the following hypothesis;

(18)

H3: A cloaking strategy generates a higher willingness to pay to consumers that are seeking for distinction (snobs).

2.2.2 Omnivores and group cohesion

On the other hand, the omnivores are people that consider themselves to be open to all sorts of culture and goods, and they weaken both the social and cultural boundaries. Since they are open to all sorts of goods, they are willing to try out everything and they don’t care if the products to be consumed are high-brow or low-brow. Peterson (1992, in Roose et al. 2012, p. 493) discovered the omnivores and argues that omnivores tend to use culture as a tool that enables them to associate with people from different social and geographical backgrounds in order to enlarge their private and professional network. This suggests that omnivores are the ones that identify themselves with other people, the need for group cohesion, than that they want to distinguish themselves from others.

A similar term that is used to describe this type of people is “bangwagon” (Liebenstein, 1950). According to Liebenstein (1950) there is a bandwagon effect when people desire to wear, buy, do, consume and behave like their fellows (p. 184).

Another research that adds to these two different social needs of group cohesion and distinction is the “herds effect”. Salganik and Watts (2008) state that in the market for books, movies, music and other cultural products, individuals have often information on the popularity of the goods. Consistent with psychological literature on social influence (Sherif 1937; Asch 1952; Cialdini and Goldstein 2004, in Salganiks and Watts, 2008) recent work tells us that this information may influence the consumers’ decision making. However Salganik and Watts (2008) suggest that individuals can influence each other, meaning that they follow each other, which can have great impacts for the cultural markets. The experimental research of Salganik and Watts (2008) shows that people tend to follow the herd

(19)

when it comes to information that is revealed, where consumers base their decisions up on. Venkatesan (1966) also agrees on this, stating that many buying actions come from a desire to identify with a certain group of people. His experiment shows that individuals tend to conform to the group norm, when objective standards are absent. However when there was group pressure to “go along” with the group, individuals tended to resist the group pressure (Venkatesan, 1966, p.384).

Amaldoss and Jain (2005) also look into the social aspect of consumer behaviour. They state that the desire to conform might also influence purchase behaviour. There are consumers that value a product more as the number of consumers that buy the product increases (Jones, 1984; Ross, Bierbrauer and Hoffman 1976; in Amaldoss and Jain, 2005, p. 31). There even is evidence of this so-called conformism in the markets of books, toys and garments. When the product value increases this goes along with a higher willingness to pay for the product.

So concluding from prior literature we suggests that people that “tend to follow the herd” or have a more social need for group cohesion will be more likely to be attracted by flaunting strategies. This is because the information that is revealed by a flaunting strategy suggests that the mass likes the product. And when there are more consumers buying the product the value increases and therefore their willingness to pay. We will be testing this using the following hypothesis:

H4: A flaunting strategy generates a higher willingness to pay to consumers that are seeking for group cohesion (omnivores).

(20)

2.3 Consumer Involvement

Consumer involvement has played an increasingly important role in explaining consumer behaviour (Mittal and Lee, 1989, p. 363). Consumer involvement is identified as a relevant variable to advertising strategies, as it has shown to be a mediator of media exposure, response to persuasion message, depth of processing advertising, extensiveness of decision-making process and on-going product-related behaviours such as word-of-mouth communication (see Mitchell 1979; Batra and Ray 1985; Park and Mittal 1985; Laurant and Kapferer 1985; in Mittal and Lee, 1989). One of many definitions in the literature on involvement is; ‘a motivational state of mind of a person with regard to an object or activity. It reveals itself as the level of interest in that object or activity’ (Mittal 1983).

The importance of involvement for advertising strategies mentioned, Ippolito (1990) looks somewhat more specifically to the importance of consumer involvement that is involved with perceiving information and the quality of the information regarding the products. As mentioned in the introduction, a difference is found in the decision process and the search for information depending on the level of involvement of the consumers (Laurent and Kapferer (1985). The level of involvement could either be high or low, which leads to actively or passively dealing with advertisement communication. Previous research has found involvement to be a critical moderator of the decision-making process, since information processing in a high involved state tends to be more elaborate than when low involved (Koschate-Fisher, Cramer & Hoyer, 2014, p. 2). We will outline high and low involvement further in the following paragraphs.

A difference to take into account regarding information processing and involvement is gender differences. However we will not fully elaborate on this in this particular study, we consider it useful to mention since it might have implications for the cloaking and flaunting strategies. Of course there is much literature on gender differences, but Darley and Smith

(21)

(1995) examine the differences regarding information processing strategies. They claim that males are more analytical and logical in their processing information, while females are more subjective and intuitive (Broverman et al. 1968; in Darley & Smith, 1995, p. 42). O’Cass (2000) looks at information processing and involvement regarding the fashion market. Their study reveals that females report a higher level of involvement in fashion clothing (O’Cass, 2000, p. 562). Where according to O’Cass (2000) consumers become more involved when there is a potential for satisfying salient higher order psychological needs.

Another side note on consumer involvement is about the way we look at it. Zaichkowsky (1985) highlights the problem about involvement regarding different applications of the term “involvement” itself. Prior literature suggests that a person can be involved with advertisements (Krugman 1962, 1965, 1967, 1977; in Zaichkowsky, 1985), products (Howard and Sheth 1969, Hupfer and Gardner 1971; in Zaichkowsky, 1985) or the purchase behaviour (Clarke and Belk 1978; in Zaichkowsky, 1985). However the involvement with these three different aspects leads to three different reactions, which makes measuring consumer involvement somewhat problematic. Therefor Zaichkowsky (1985) came up with a new scale of Involvement measurement, where we derived our involvement scales from for this research.

2.3.1 High involvement

According to Laurent and Kapferer (1985) highly involved consumers are more likely to show particular behaviours like active search, extensive choice process and active information processing. Highly involved consumers are likely to actively search for information, but they are also likely to look for information from alternative sources (Laurent and Kapferer, 1985). Park and Leeb (2008) mentioned that highly involved consumers are more likely to process arguments in a thoughtful and effortful way. This argument of Park and Leeb (2008) fits with

(22)

the elaboration likelihood model (ELM) of Petty and Cacioppo (1983). This model looks at two psychological possible routes advertisements persuade consumers. The first is called the central route, this routes sees attitude changes as a result of diligent consideration of information (Petty and Cacioppo, 1983). The central route is more on the cognitive processes that are involved towards the advertisements of companies. The consequences of this route are the stability over time and that this route is a better predictor for behaviour and intentions. Contrary to the central route, there is the peripheral route. Considering this peripheral route attitudes change because the attitude object has been associated with positive or negative cues. With this route there are certain cues that shape attitudes based on feelings without the need of extensive issue-relevant thinking (Petty and Cacioppo, 1983). Motivation and ability are components that are relevant for this model, and so is involvement. The central route is supposed to have larger impact under high involvement conditions. Petty and Cacioppo (1983) base this on the fact that high involved consumers specifically look for information and they are focussing on strong arguments, and really listen to and think about what experts say.

Based on prior literature, high-involved consumers are particularly looking for relevant information and they thoughtfully process the arguments given on a product. Bringing this into relation with the cloaking and flaunting strategies, high-involved consumers can be attracted by cloaking and flaunting strategies. However, when “strategically withholding information” is using under a cloaking strategy, only high involved consumers are likely to find this information. Low involved consumers are not actively seeking for information and will therefore not find the hidden information, except if it were accidentally. Consumers that are highly involved can be actively searching for information on all kinds of products, so exclusive (hidden) products as well as flaunted products or any other product. Therefore it is interesting to see whether the high-involved consumers perceive the

(23)

information provided by both strategies differently than the low involved consumers. While high involved consumers are processing the advertisements in a more thoughtfull way their willingness to pay relationship could be stronger than for low involved consumers (Park and Leeb, 2008). Therefore we came to the following hypothesis;

H5: High involvement is likely to be a stronger moderator for the relationship between the cloaking and flaunting strategies and willingness to pay than low involvement.

2.3.2 Low Involvement

Where highly involved consumers show these particular behaviours of actively searching for information and so on, the low involved consumers lack these types of behaviours (Laurent and Kapferer, 1985). But on the other hand, where they take a more passive position towards processing information, the low involved consumer looks for signs of popularity of the product (Park and Leeb, 2008). Considering the ELM model we spoke of in the last paragraph, Petty and Cacioppo (1983) found that the peripheral route has more impact on low involved consumers. This would imply that consumers that have a low involvement tend to rely more on feelings and a more unconscious form of making decisions.

Like we already mentioned, low involved consumers are not likely to buy products with a cloaking strategies when they strategically withhold information, since they do not actively seek information. But according to the theory of Park and Leeb (2008) low involved consumers look for signs of popularity instead. Since flaunting strategies signal how popular their products are, low involved consumers are likely attracted by those strategy. However, depending on the interpretation of the advertisement of the cloaking strategy, low involved consumers could as well be attracted by a cloaking strategy. But since the argument for the flaunting strategy had a stronger theoretical background, we came to the following hypothesis;

(24)

H6: Low involved consumers tend to have a higher willingness to pay for products promoted by a flaunting strategy.

After reviewing the cloaking and flaunting strategies, consumer characteristics and consumer involvement we will provide you with an overview of our hypotheses in our conceptual framework (figure 1).

Figure 1 – Conceptual framework

Cloaking Strategy Flaunting Strategy Consumers’ willingness to pay Consumer characteristics Consumer involvement H1 H2 H3 H4 H5 H6

(25)

3.0 Methodology

In this paragraph we will explain the research design that has been chosen for this research, than we will provide information on the sample, and the data collected.

3.1 Research design

This study can be termed an explanatory research, since we are establishing causal relationships between variables in certain situations (Saunders, Lewis and Thornhill, 2009, p. 140). Yogonarasimhan (2012) has done qualitative research in the direction of the subject of this thesis, however the cloaking and flaunting strategies have never been experimentally tested. With experiments you can see whether a change in an independent variable produces a change in another dependent variable (Hakim, 2000 in Saunders et al. 2009). Therefor the research method used in this thesis will be an experiment. Besides the cloaking and flaunting strategies, we will take the consumer characteristics and involvement into account as moderators.

3.1.1 Experiments

The experiment will be set up with the auctioning platform Veylinx. Veylinx is a project that was developed by researchers at the University of Amsterdam to test the consumers’ willingness to pay using second-price auctions. In a second-price auction the highest bid wins but the winner pays the second-highest bid price for the product. This mechanism is explained by McAfee and McMillan (1987), who note that the bid must be equal to the value the bidder devotes to the product, in order for it to be a profitable auction. Participants bidding in the Veylinx auction do not get to see the actual price of the product, and the bidding is “blind”, which will prevent the bidders for being influenced by others.

(26)

By manipulating the advertisements that people get to see when they bid for a certain product in the auction, changes in independent variables can be brought into relation with the depended variable. The type of experiment used is called a field experiment, since the auction takes place in the field (Sanders et al., 2009).

3.1.2 Sample

There are currently ± 3000 people signed up for Veylinx. Everyone signed up is able to read Dutch and lives in the Netherlands, since the research platform only concerns The Netherlands so far. Approximately seventy per cent of the people has an age above 30, making it more generalizable for the whole population instead of only using students that might have a limited amount of money to spent. For this research the auction was sent out to approximately 500 participants of Veylinx, randomly chosen. Not everyone invited to the auction did participate This might be due to a lack of interest for the product, or time constraints, since the auction only lasts for 24 hours.

3.2 Data collection

The participants that are signed up for the Veylinx auctions where gathered by handing out flyers, and using social media networks. Every person joining the research platform of Veylinx, has to bring in 100 new participants.

For this research, the participants received an email to join the auction. When they open the link to the auction, they first had to confirm information on their names, address, gender and education, than got to the auction, bid, and after bidding they had to fill out several additional questions regarding the bidding.

The product used to test the hypotheses in this research was a fashionable leather laptop case. The leather laptop case is black, and therefore unisex. Participants were divided

(27)

into three treatments. The first treatment contained the control group; the second treatment showed a flaunting strategy and the third treatment showed a cloaking strategy. Treatments are shown in appendix 8.1. After the participants bid on the product, they got 7 questions, which were the same for all treatments. The first three contained questions in order to measure whether they were perceived as snobs or omnivores. questions four, five and six where measuring the involvement of the participant and question 7 was a control question.

3.2.1 Operationalization of the variables

In order to answer the research question we take the consumers’ willingness to pay as our dependent variable and the cloaking and flaunting strategies are the independent variables. Consumer characteristics and consumer involvement are the moderating variables.

The dependent variable consumers’ willingness to pay is operationalized by the bid-price a participant places for the product. By bidding they show how much they are willing to pay for a certain product where the price is not given. As Alfred Marshall, the founding father of modern economics, argued: ‘‘[t]he desire to obtain anything that is ordinarily bought and sold for money, is ... measurable by the price that people are willing to pay for it.’’ By ‘‘voting’’ on the market place with money, consumers reveal their preferences; the price system translates these individual monetary votes into supra-individual market prices (Velthuis, 2003, p. 185).

The cloaking and flaunting strategies, the independent variables, are operationalized by the information that we will show a side of the product we are auctioning. To start with treatment 1, the control treatment. This treatment shows the same pictures as the other treatments, but only has descriptive information of the laptop case. Then we continue with treatment 2, the flaunting strategy. In order to signal a flaunting strategy we used the word “bestseller”, and “must-have” in the advertisement of the product, as we derived this from

(28)

theory (Yoganarasimhan, 2012; Tucker and Zhang, 2011). And last treatment 3, the cloaking strategy. To signal a cloaking strategy we use the words “exclusive” and “limited edition” in order to show exclusivity of the product (Catry, 2003; Carcano et al., 2011). As already mentioned, treatment advertisements can be seen in appendix 8.1.

The moderating variables consumer characteristics and consumer involvement where measured during the short survey at the end of the bidding process. In order to measure the difference between snobs and omnivores we had three statements, which can be found in the appendix 8.2. No snobs versus omnivores scale was found in literature, so we had to compute it ourselves. We used a 5-point nominal scale in order to measure the participants’ agreeableness with the statements. Concerning the involvement moderator we had three statements as well, also found in appendix 8.2. These statements were derived from the involvement scale of Zaichkowsky (1985). We used the same 5-point scale as mentioned before. The last question of the survey contained a manipulation check; “is het getoonde product exclusief”. Reliability checks and factor analysis of the two moderators can be found in section 4, the preliminary analysis.

3.2.2 Pre-tests

In order to do valid research we used pre-tests. Validity can be described as the extent to which the instrument measures exactly what the researcher wants it to measure (Saunders, 2009, p. 157). So by doing pre-tests, we measured whether the strategies where perceived as they were intended to. One survey was sent out, containing the advertisement of the cloaking strategy. After seeing the advertisements the participants were given six statements related to the advertisement, using a 5-point ordinal scale. Another survey was sent out, showing the flaunting strategy advertisement. People were asked to fill out 4 statements on a 5-point ordinal scale. All questions can be found in Appendix 8.3.

(29)

There where respectively 44 respondents gathered through social media, 24 respondents for the cloaking strategy and 20 respondents for the flaunting strategy. In order to make sure that the multiple questions can be combined into one variable it is necessary to control the internal consistency reliability of the scales. Where reliability can be described as the extent to which a measuring instrument is measuring consistently (Saunders, 2009, p. 156). This reliability check is done determining the Chronbach’s alpha, which should be above 0.700 (Field, 2009, p. 675). The Chronbach’s alpha of 0.658 for the cloaking strategy is a bit low, but is reasonably reliable. The Chronbach’s alpha for the flaunting strategy is 0.864. Table 3.2.2 represents the Chronbach alpha’s, means and standard deviations. On the basis of the Chronbach alpha’s and the mean scores (both above 3, meaning they agreed) we can see that the flaunting strategy is reliable, and the cloaking strategy is nearly there. After the pre-test we made several minor adjustments to the advertisements.

Tabel 3.2.2 – Chronbach’s alpha, means and standard deviations

Chronbach’s Alpha

N Minimum Maximum M SD

Cloaking 0.658 5 2.800 3.600 3.380 2.885

(30)

4.0 Results

In this paragraph we will present the results of this research. First we will start with the preliminary analysis; which contains descriptive analysis as well as a normality check, factor analysis, internal reliability check and a correlation matrix. Subsequently, we will start testing the hypothesis, where we will compare the means of the three treatments using a one-way ANOVA and lastly we will test the hypotheses using a General Linear Model and regression analysis.

4.1 Preliminary analysis

We will start with preliminary analysis to prepare the dataset in order to test for the hypotheses. As mentioned we will provide the statistics, test the assumptions of the ANOVA, run a factor analysis and reliability check of scales and compute a correlation matrix.

4.1.1 Descriptive statistics

Out of the approximately 3000 people signed up for Veylinx, only a subgroup was used for this research. This led to 463 participants placing an actual bid. 267 of them where male (57.7 %) and 196 of them where female (42.3 %). So we had slightly more men in our sample. The mean age was 43 (SD = 16.86). In line with this mean age, only 45 participants where students (18.4 %) and 199 participants indicated they were non-student (81.6 %). Concerning education, only 255 participants filled out their education level. We can see that our sample was relatively highly educated since 162 of the participants followed a HBO or WO education (63.5 %), 60 participants went to high school (23.5 %) and 33 participants only went to primary and secondary school (13 %). Table 4.1.1 demonstrates an overview of the

(31)

descriptive analysis. In section 4.1.4, we show the results of a correlation test, to see whether these control variables have an effect on the bid amount.

Table 4.1.1 – Descriptive measures respondents

N=463 Treatment 1 Treatment 2 Treatment 3 Total Condition Control group Flaunting Cloaking Total

Participants 136 165 162 463 Male 79 93 95 267 Female 57 72 67 196 Age 44 (SD = 17.99) 43 (SD = 16.61) 43 (SD = 16.20) 43 (SD = 16.86) Non-student 77% 84.4% 82.1% 81.6% Highly educated (HBO/ WO) 67.1% 62.4% 61.4% 63.5% Bid amount M = € 7.35 (SD = 10.44) M = € 6.50 (SD = 11.01) M = € 6.31 (SD = 10.24) M = € 6.69 (SD = 49.10) Minimum bid € 0 € 0 € 0 € 0 Maximum bid € 50.00 € 70.00 € 72.00 € 72.00 % Zero bids 38.2% 46.7% 44,4% 43.8%

As also shown in table 4.1.1 the mean auction bid values for treatment 1,2 and 3 are, respectively, M = 7.35, M = 6.50 and M = 6.31, with standard deviations of SD = 10.44, SD = 11.01 and SD = 10.24. Also showed in the descriptive table is that the sample did include many zero bids. Out of the 463 there were 203 zero bids, which is 43.8 %. Since zero-bids are likely placed by participants that were not interested in the product, they can be excluded in order to make the research more valid. We will elaborate on this in paragraph 4.1.2.1, where we will check for normality first.

4.1.2 Testing assumptions for ANOVA

In order to start with performing a one-way ANOVA to compare the mean bid values we need to check whether the required assumptions of the ANOVA are met. ANOVA assumes that

(32)

variances in each condition are fairly similar, observations are independent, the dependent variables are measure on an interval scale and the within-condition distributions are normally distributed (Field, 2009).

4.1.2.1 Testing normality

We could derive from the standard deviations and minimum and maximum bid amount that the data is spread over a large range of values. The standard deviations are high for all three treatments when we compare them with their mean bid values. Also, the maximum bid deviates largely from the mean values in all three treatments. Computing the histogram and the boxplot (figure 4.1.2.1) for the bid amount data, we can see that data is highly positively skewed, clustered at the low end and is non-normally distributed. The skewness value is 2.474 with a standard deviation of 0.113. This indicates that the data is highly skewed to the right. Additionally the kurtosis value is 8.162 with a standard deviation of 0.226, which is extremely high, indicating a very pointy and heavy-tailed distribution. The Kolmogorov-Smirnov and Shapiro-Wilk confirm numerically that the data is non-normally distributed. The Kolmogorov-Smirrov shows a significant deviation from normality (Statistic = 0.263, df. = 463, p = 0.000). The Shapiro-Wilk confirmed this as well (Statistic = 0.681, df = 463, p = 0.000). These values can be found in table 4.1.2.1.

(33)

Figure 4.1.2.1 – Histogram and boxplot for bid amount

Table 4.1.2.1 – Tests for normality

Kolmogorov-Smirnov Shapiro-Wilk

Skewness Kurtosis Statistic p-value Statistic p-value

Treatment 1 1.751 2.765 0.243 0.000 0.741 0.000

Treatment 2 2.729 9.655 0.277 0.000 0.647 0.000

Treatment 3 2.853 0.376 0.269 0.000 0.661 0.000

Total 2.474 8.162 0.263 0.000 0.681 0.000

After concluding that the data was not normally distributed, an attempt was made to try to normalize the dataset. We started by deleting potential outliers and non-representative cases from the data. Thereafter, we computed the logarithm of the bid amount.

Looking at the boxplot in figure 4.3.1 we identified 6 major outliers, where major outliers are defined as a bid amount 3.0 x IQR outside the central box (Field, 2009). Computing the z-scores to look at these outliers, and using a cut-off value of 4 only 5 were identified as outliers. Than looking at the non-representative cases, as we already identified there where many zero bids, which indicates that a lot of participants were not interested in

(34)

buying the product. Table 4.1.1 showed that this was 43.8%. To correct for the non-interested participants the data was reduced to only above zero bids. In order to leave out the bids close to zero, which also are likely to indicate non-interested participants, we select the top 50 percentile of bid amounts. The remaining data was transformed using the logarithm of the bid amount. Looking at the histogram of the ln bid amount, shown in figure 4.1.2.1.1, we observe a more normal distribution. However, the Kolmogorov-Smirnov test still shows a significant result (Statistic = 0.106, df = 221, p = 0.000), as well as the Shapiro-Wilk test (Statistic = 0.961, df = 221, p = 0.000). Meaning that we still cannot assume that the data is normally distributed. Although we cannot assume normality, the values of the skewness and kurtosis are reduced as can be seen in table 4.1.2.1.1. From here, we will continue working with the downsized dataset with 221 participants remaining. The downsized data are summarized in the descriptive table 4.1.2.1.2.

(35)

Table 4.1.2.1.1 – Test for normality with downsized data set

Kolmogorov-Smirnov Shapiro-Wilk

Skewness Kurtosis Statistic p-value Statistic p-value

Treatment 1 -0.317 -0.641 0.131 0.007 0.951 0.012

Treatment 2 -0.467 -0.019 0.145 0.000 0.943 0.002

Treatment 3 -0.292 -0.436 0.087 0.200* 0.968 0.051

Total -0.391 -0.227 0.106 0.000 0.961 0.000

* This is a lower bound of the true significance

Table 4.1.2.1.2 – Descriptive of downsized data set

N=221 Treatment 1 Treatment 2 Treatment 3 Total Condition Control group Flaunting Cloaking Total

Participants 65 79 77 221 Male 40 48 43 131 Female 25 31 34 90 Age 43 (SD = 2.11) 41 (SD = 1.77) 41 (SD = 1.88) 42 (SD = 16.37) Non-student 78.4% 81.6% 81.1% 80.5% Highly educated (HBO/ WO) 64.9% 61.2% 57.1% 60.9% Bid amount M = € 14.06 (SD = 1.26) M = € 11.93 (SD = 1.09) M = € 11.56 (SD = 0.95) M = € 12.43 (SD = 9.43) Minimum bid € 1.75 € 1 € 1.11 € 1 Maximum bid € 40.00 € 45.00 € 37.95 € 45.00 % Zero bids 0% 0% 0% 0% 4.1.2.2 Homogeneity of variance

After testing for normality, homogeneity of variance is another assumption that must be tested. We used Levene’s test to do this (Field, 2009). The test shows whether there are equal variances between treatments or not. Using the dataset after downsizing, The Levene test showed us that we can assume equal variances (Levene statistic = 0.988, p = 0.374).

(36)

4.1.3 Reliability of scales

Since we were measuring the constructs of consumer characteristics and involvement using multiple items, we run a principal components factor analysis first, where after we need to control for the internal consistency reliability of the scales as we did at the pre-tests as well.

To run the factor analysis and compute the Chronbach’s alpha we removed the cases that were not filled from the computation (n= 21 removed). The factor analysis showed that all three questions for consumer characteristics load on one factor (EV=1.748, R2 = 58.271). The Chronbach’s alpha for consumer characteristics is 0.850 after deleting question 3 (see appendix 8.3). Although three items is recommended to measure a construct, we chose to delete this one, since the Chronbach alpha of the three items was 0.485, which is far below the needed 0.7. The Chronbach’s alpha (after deleting one item) shows good reliability and internal consistency. For the involvement construct, the factor analysis also showed that the three questions load on one factor (EV = 2.054, R2 =68.465). The Chronbach’s alpha is 0.767 and this cannot be improved by deleting an item. So this construct also shows good reliability and internal consistency. The mean and standard deviations can be found in table 4.1.3.

Table 4.1.3 – Chronbach’s alpha, means and standard deviations

Chronbach’s Alpha N Minimum Maximum M SD Consumer characteristics 0.850 2 2.600 2.860 2.730 2.124 Involvement 0.767 3 3.165 3.870 3.722 2.934

Since the scales where reliable, we computed the questions into a consumer characteristics and an involvement variable. In order to divide them in respectively, snobs (high on consumer characteristics) and omnivores (low on consumer characteristics) and high and low involvement we used a median split, since this results in better normally distributed data. The median of the snobbism variable was 2.75, meaning that >2.75 would result in being a snob

(37)

and <2.75 in being an omnivore. The median of involvement was 3.67, resulting in low involvement for <3.67 and high involvement for >3.67.

4.1.4 Bivariate correlation analysis

In order to control for alternative explanations a bivariate correlation analysis was preformed between control variables gender, age, education, student and duration of the bid and the dependent variable ln bid amount. Age (r = -0.184, p < 0.01), education (r = 0.196, p < 0.05) and bid duration (r = 0.216, p < 0.1) all three significantly correlate with bid amount.

Age is negatively correlated with bid amount, which means that younger people place higher bids. A possible explanation for this could be that younger people are more likely to travel for study / work and therefore could use a laptop case.

Education is positively correlated with bid amount, which means that the higher the education the higher the bid amount. This could be explained by the higher the education, the more likely it is that someone has a higher salary, which could result in willing to pay a higher price.

The correlation of the bid amount with the bid duration could simply be explained by, the higher the auction bid, the more likely it is that it took the participant longer to decide whether to spend a higher amount of money on the bid.

To check whether education, age and bid duration differ within treatments we performed an ANOVA on those control variables. This showed non-significant differences for education (F(2,125)=0.763; p=0.468), age (F(2,218)=0.567, p=0.568) and bid duration (F(2,218)=1.022, p=0.361), meaning that the control variables do not influence the results and we can continue with the ANOVA for testing the hypotheses.

(38)

4.2 Hypotheses testing

After the preliminary analysis, we will test the hypothesis in this section of the results. First we will start with a one-way ANOVA in order to test the first two hypotheses and second we will use a univariate analysis of variance in order to test hypotheses 3 till 6.

4.2.1 One-way ANOVA

Although the assumption regarding the normality is still violated, we will still use the one-way ANOVA, since this test is generally known to be a robust test, even when the data is skewed (Field, 2009). We used the one-way ANOVA to test for the significant differences, in order to compare the mean bid amounts of the three treatments with each other. A one-way ANOVA was calculated comparing the three treatments on ln bid amount. The mean score of ln bid amount for the control group was M = 6.97 (SD = 0.80), the flaunting treatment showed a mean of M = 6.73 (SD = 0.91) and the cloaking treatment had a mean of M = 6.78 (SD = 0.79). Contrary to our expectation, the control group had the highest mean score for willingness to pay. The cloaking treatment showed a slightly higher willingness to pay comparing to the flaunting treatment. However, the ANOVA showed a non-significant effect (F(2,218) = 1.500, p = 0.225). This means that there is no significance difference in willingness to pay found between the different treatments and therefore we cannot draw conclusions. Based on the non-significance result the following two hypotheses can be rejected:

H1: A cloaking strategy will lead to a higher willingness to pay. H2: A flaunting strategy will lead to higher willingness to pay.

Although there were no significant differences found between the different treatments, it could occur that there will be significant results when we use specific parts of the sample.

(39)

First we checked whether gender makes a difference, since theory showed that men and woman interpret advertisements differently. Subsequently, we will check different age categories, education and whether there is a difference in willingness to pay between students and non-students.

To start with, the one-way ANOVA showed no significant differences in bid amount between men and woman (F(1,219) = 0.589, p = 0.444). The mean bid amounts where fairly close to each other, respectively, M = 6.85 (SD = 0.80) and M = 6.76 (SD = 0.90). There were also no significant results found when we looked at the treatments individually. When we excluded all men from the sample, we found a significant difference for mean bid amount between treatments for a significant level of 0.1 (F(2,87)= 2.829, p = 0.064). Meaning that if the research where only among woman, there would be a significant difference. Excluding all woman from the data did not lead to significant differences (F(2,128)= 0.068, p = 0.935). When we solely take the bids of woman, the mean bid amounts where; M = 7.10 (SD = 0.76) for treatment 1, M = 6.54 (SD = 1.07) for treatment 2 and M = 6.72 (SD = 0.78) for treatment 3. The Post Hoc test revealed that only treatment 1 and 2 differ significantly from each other (M difference = 0.56, p = 0.021). However, since treatment 1 has the highest mean bid amount again, both hypotheses will still rejected. It is interesting to see that treatment 3 (cloaking) has a higher mean bid amount that treatment 2 (flaunting), although the treatments do not significantly differ from each other, as the Post Hoc test revealed.

The age of participants showed a negative correlation with bid amount (r = -0.184, p < 0.01), but in order to check whether there are significant differences between different ages, we split the variable up in different age groups. The first age group contains participants younger than 20, the second age group contains participants between 20 and 40, the third age group contains participants between 40 and 60, the fourth age group contains participants between 60 and 80 and the fifth age group contains of participants older than 80. The one-way

(40)

ANOVA shows that the differences between age groups are significant (F(4,216)= 2.537, p = 0.41). The mean score of the 20-40 age group was M = 6.90 (SD = 0.94), for 40-60 it was M = 6.89 (SD = 0.73) and for 60-80 it was M = 6.39 (SD = 0.75). The mean bid amounts are in line with the negative correlation, showing that the younger the participants the higher their bid was. The Post Hoc test revealed that the only significant difference is found between the age group 40-60 and 60-80 (M difference = 1.64, p = 0.040).

Than continuing with education, there is a significant difference found between the different education levels (F(4,123)=2.867, p = 0.26). The Post Hoc test revealed that participants who are high (WO) educated (M = 7.03, SD = 0.79) have a higher willingness to pay than high school educated participants (M = 6.62, SD = 0.66).

There is also a significant difference found between non-students and students in their relation to bid amount (F(1,121)= 3.352, p = 0.070). The mean scores are respectively, M = 6.69 (SD = 0.81) and M = 7.02 (SD = 0.73), which shows that students have a higher willingness to pay for the laptop case comparing to non-student participant. This is also in line with the age aspect.

However, a analysis of variance points out that there is no significant difference in bid amount for different treatment groups for these control variables (p>0.05), expect for the tests excluding men, as we already discussed.

4.2.2 General Linear Model

The main effect of the different treatments on the bid amount is not significant, therefore we cannot say much about the moderators either, which results in rejecting all other hypotheses as well. We still used the General Linear Model, in order to perform an analysis of variance to look at any possible direct effects of the moderators.

Cloaking or flaunting? : the effects of consumer characteristics and involvement on consumers’ willingness to pay