The relation between scarcity and willingness to pay moderated by message framing

(1)

The relation between scarcity and willingness

to pay moderated by message framing

Master Thesis

Business Administration, Marketing Track

University of Amsterdam

By Maartje Kokkeel (11397586)

Under supervision of Max Nohe

26th of January, 2018

(2)

2

Statement of originality

This document is written by Maartje Kokkeel 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.

(3)

3

Acknowledgement

This Master thesis is written as a completion of the Business Administration – Marketing Track – Master programme of the University of Amsterdam. I would sincerely like to thank my

(6)

6

Abstract

One of the first studies into scarcity dates back to 1968 when Brock came up with the

commodity theory. This theory states that “any commodity will be valued to the extent that it is

unavailable” (Brock, 1968). In the following years multiple scholars have built upon this notion

and a general consensus is found: product scarcity increases both product preference and

product valuation. Psychological explanations are: consumer’s desire for uniqueness, increased

quality inferences or perceived loss of freedom to choose.

This paper takes a new perspective on scarcity by investigating the relation between scarcity and willingness to pay and therefore how far prices can be stretched. Moreover, researchers have never investigated whether message framing, with regard to rational (focus on product) or emotional (focus on social norms of the snob and bandwagon effect) appeals, affect this relationship. Therefore, this research extends the existing literature by trying to find mechanisms that explain the effect of scarcity on willingness to pay, moderated by message framing.

A between subjects online experiment shows that scarcity has little to no effect on consumer’s willingness to pay. Moreover, message framing does not significantly influence this relationship. However, the results do show that emotional message appeals leads to a significant higher willingness to pay compared to rational message appeals. Therefore, the main finding of this study provides evidence that message framing, with regard to rational and emotional appeals, impact price stretch and scarcity doesn´t in an online environment.

(7)

7

Chapter 1 Introduction

Since 2004, H&M has the yearly tradition to invite famous fashion designers to design one collection for H&M. Designers such as Karl Lagerfield, Viktor & Rolf, Versace and Isabel Marrant reached huge successes with the H&M collaboration (Yotka, 2016). Mass hysteria is created as it is widely known that a limited edition is produced. People even camp overnight to get their hands on a piece and the collection is sold out in only a few minutes. This is a classic, though modern, example of scarcity marketing in the apparel industry. Booking.com, one of world’s largest online travel companies, uses a similar technique to persuade consumers. When

somebody is considering to book a room, Booking.com presents scarcity messages on their website. For example, the website uses messages such as “only 3 rooms left” or “5 minutes left to check out” to seduce consumers to purchase the commodity.

Both examples build upon the notion of marketing scarcity and show that businesses can play ‘’hard to get’’. People desire what they cannot have and marketers build upon this by telling

consumers that the product is very rare in order to trigger consumers to buy the commodities. It is well researched that scarcity messages increase product valuation and purchase intention. Psychological explanations are: consumer’s desire for uniqueness, increased quality inferences

or perceived loss of freedom to choose. However, little is known about the relation between scarcity and willingness to pay and therefore how far prices can be stretched. Moreover, researchers have never investigated whether message framing, with regard to rational (focus on product) or emotional (focus on social norms) appeals, affect this relationship. Therefore, the present research extends the existing literature by trying to find mechanisms that explain the effect of scarcity on willingness to pay, moderated by message framing.

(8)

8

Up until today, the most common findings in the literature is that scarcity increases both purchase intention and product valuation due to overestimation of the quality of the product (van Herpen et al., 2005; Lynn, 1991). According to pricing theory, the scarcity principle should cause prices to rise until an equilibrium is reached between supply and demand. Moreover, the prospect theory states that scarcity increases people’s purchase intention as people prefer the

probability of a modest gain (obtaining a scarce product) over the probability of a larger gain (obtaining it for a discount).

Multiple researchers have defined mechanisms to explain why and how scarcity affect valuation and purchase intention. The two most supported theories are (1) need for uniqueness theory, which suggests that behavioural responses are driven by a desire for moderate level of social distinctiveness (Fromkin, 1970; Fromkin & Snyder, 1980); and (2) reactance theory, which posits that positive attitudes towards scarce commodities are driven by the threat or fear to lose choice freedom, which increases the perceived attractiveness and desirability of scarce products (Brehm, 1972; Wicklund, 1974).

More recent studies looked at the different effects of framing, in terms of time scarcity or quantity scarcity, with a focus on the likelihood to purchase. Although little research has been conducted, the main finding is that time limit restrictions are most useful for attracting customers to the brand. In contrast, quantity restrictions are necessary for stockpiling and are associated with higher product quality (Inman et al., 1997).

In sum, current literature on scarcity effects often focus on the impact of commodity scarcity on the attitude towards the product, in terms of perceptions of value and purchase intention. Nonetheless, the effect of scarcity messages need more investigation. Therefore, the main intent of this study is to investigate how rational or emotional message appeals underlie the relation between scarcity and willingness to pay?

(9)

9

This study theoretically contributes to the limited knowledge about the effect of scarcity on willingness to pay. Moreover, this research builds upon the little attention that has been paid in the literature on how message framing could affect the relation between scarcity and willingness to pay.

Previous literature on scarcity marketing solely focused on quantity- (e.g. only 5 items left, limited edition) and time scarcity (while supplies last, last day) messages. However, this research will also take into account message appeal, both rational and emotional. Rational message appeals focus on the product and its characteristics. For example, “this exclusive product is of the highest quality and contains unique features such as…” whereas emotional appeals focus more on social factors that influence consumer behaviour and emotions.

According to van Herpen, Pieters and Zeelenberg (2005; 2009) scarcity increases product choice and product valuation due to two social factors: (1) snob-effect and (2) bandwagon effect. The snob effects focuses on a shortage in supply. In fact, this theory focusses on the desire of people to disassociate from the crowd and to be unique. In contrast, the bandwagon effect focuses on excess demand due to a desire to “join the crowd”. For example, emotional

appeals such as “are you one of the lucky people to get your hands on…” or “we granted you the opportunity to be one of the few worldwide to obtain…”, could elicit a feeling of uniqueness

(snob effect), whereas scarcity messages such as “in popular demand” or “over [n number] sold..” signals to consumers that the product is in popular demand (bandwagon effect) and could

therefore have different impacts on willingness to pay, hence price stretching, than rational framing. Therefore, this study will contribute to the consumer behaviour literature by investigating which mechanism is most effective to optimize pricing and willingness to pay.

From a managerial perspective, this study contributes to a better understanding of the relationship between scarcity messages on consumer behaviour which might provide managers

(10)

10

with novel insights on how to approach consumers, in terms of scarcity messages and message framing, to achieve highest margins.

This researched will be addressed by means of an experimental design. This study will focus on the effects of scarcity messaging on willingness to pay in the material goods industry.

This paper is organized as follows: chapter 2 discusses the existing literature with regard to scarcity and message framing. Chapter 3 introduces the conceptual framework and hypotheses. Chapter 4 explains the methods and chapter 5 presents the results. Finally, chapter 6 provides a thorough discussion of the findings followed by a conclusion chapter 7.

(11)

11

Chapter 2 Literature review

In this chapter, a detailed review is provided of literature on the theoretical and psychological underpinnings of scarcity, purchase intention, message framing and product valuation. Section 2.1 takes a commodity view on the principle of scarcity. Section 2.2 examines how the commodity theory can be applied to material goods. Section 2.3 explores the effect of quantity and time scarcity on product choice and product valuation. Section 2.4 explains price effects along with the heuristic-systematic dual processing model. Lastly, section 2.5 reviews the literature on message framing. Prior research findings will be analysed to provide clear insights and a better understanding of current knowledge.

2.1 Commodity theory: the principle of scarcity

The principle of scarcity relies on the notion that products become more attractive as they become limited in availability, even when the product is not desirable on its own (Cialdini & Rhoads, 2001). According to Aggarwal, Jun and Huh (2011), there are two types of limited availability or scarcity: (1) limited time scarcity, which is referred to as a limited duration of an offer and (2) limited quantity scarcity, which is marked by a predefined quantity of products (Aggarwal et al., 2011).

Practitioners argue that any kind of scarcity enhances the perceived value of products. Therefore, limited editions are often issued by marketers to communicate limited supply and to drive up prices.

(12)

12

Brock (1968) tried to unravel the scarcity effect on value enhancement and used scarcity as central focus of the commodity theory. This theory explains the psychological responses of consumers to scarcity and the main premise of this theory is: “Any commodity will be valued

to the extent that it is unavailable” (Brock 1968; Lynn, 1991). The lower the availability of

goods, the higher its valuation. In order to absolutely understand this principle three concepts underlying the commodity theory will be discussed: (1) commodity, (2) value, and (3) unavailability (Lynn, 1991).

To be able to label a product or service as a commodity it has to meet three criteria. First of all, a commodity must have some utility to a person. Second, commodities must be interchangeable between people. Finally, commodities must have the potential to be possessed. Using this definition, all goods and services in the consumer environment can be labelled as commodities (Lynn, 1991).

Furthermore, unavailability refers to the limited availability, or scarcity, of product and service offerings (Lynn, 1991). Brock (1968) defined four operational definitions of unavailability: (a) limit supply of a commodity (limited editions), (b) cost of acquiring or providing a commodity, (c) restrictions that limit the possession of commodities (maximum order size) and (d) delays in providing a commodity.

Lastly, value is defined as “the potency for affecting attitudes and behaviour” (Brock, 1968). By way of explanation, the enhancement of the value (potency) of a commodity will increase its utility. Accordingly, this makes the commodity more desirable. Therefore, there is a positive relation between value and utility. The more desirable and sought after a product, the more value people attach to the product (Lynn, 1991).

(13)

13

A critical remark is that the commodity, used in the theory of Brock (1968), is limited to information availability which is provided in a message to recipients. Therefore, the question among scholars arises whether the commodity theory can be transferred to material goods.

2.2 The commodity theory applied to material goods

When buying a product, consumers start with making a value cost trade off, which is derived from elements such as a products’ perceived quality, it´s retail price or the products’ country of origin. However, an aspect that has received little research attention in the literature, but does seem to affect choice behaviour, product valuation and purchase intention, is scarcity (van Herpen et al., 2014).

Fromkin, Olson, Dipboye and Barnaby (1971) were among the first who investigated the relation between any type of scarcity and product valuation. The researchers found that scarce products were rated as more expensive than abundantly available products.

Worchel, Lee and Adewole (1975) investigated the effect of scarcity (scarce versus abundant), change in scarcity and the cause of scarcity (accidental versus popularity) on the valuation of cookies. The researchers presented people two jars of identical cookies – one with only two cookies and another jar with ten cookies inside. The results were consistent with the commodity theory of Brock (1968). Scarcity increases product valuation and product liking. In addition, the results showed that scarcity caused by popularity led to the highest ratings with regard to attractiveness and value. Lastly, a change from abundant availability to scarce availability enhances the value of a product more than constant limited availability (Worchel et al., 1975).

Verhallen (1982) elaborated on the research of Fromkin et al. (1971) and Worchel et al. (1975) by investigating whether intrinsic product information and/or scarcity information affect choice

(14)

14

behaviour. Where Fromkin et al. (1971) and Worchel et al. (1975) confirmed the commodity theory, Verhallen (1982) only partially confirmed this theory. The researcher argues that the commodity theory is only valid if consumers rate the product as attractive. Next to this, Verhallen (1982) referred to the reactance theory of Brehm (1989) to clarify the effect of unattainability, which is complementary to the commodity theory.

The reactance theory is based upon the notion that people perceive the unavailability of commodities as a threat to or elimination of their behavioural freedom (loss of freedom), causing their state of psychological reactance to be aroused. Scholars found that this perceived threat increases people’s motivation to restore their freedom that was threatened. Accordingly, this increases the perceived attractiveness and desirability to obtain the commodity (Brehm, 1972; Wicklund, 1974). Consequently, scarcity causes reactance that is based upon increased attraction towards the threatened commodity. Noteworthy, the reactance response often entails an overreaction. The threatened commodity will be valued more than before (Clee & Wicklund, 1980).

The reactance theory relates to the cognitive dissonance theory of Festinger (1957). If people’s freedom to engage in certain behaviour is threatened, people experience a feeling of discomfort. This is causing an alteration in people’s attitudes, beliefs or behaviours leading again to cognitive consistency (Festinger 1957). For example, a person can pick between two alternatives: commodity A and commodity B. If the person is told to choose commodity A, the person will experience a loss of freedom to choose alternative B. In order to restore this freedom, the person will most probably pick commodity B. Hence, alternative B becomes more desirable.

(15)

15

2.3 The effect of quantity and time scarcity on product choice, purchase

intention, product valuation and quality inferences.

This section will discuss thoroughly the effects of quantity scarcity and time scarcity on product choice, product valuation and quality inferences. Section 2.3.1 will discuss quantity scarcity and section 2.3.2 will elaborate on time scarcity.

2.3.1 Quantity scarcity

Limited quantity scarcity is referred to as a promotional offer that is applicable to a predefined quantity of a product (Lee et al., 2015). Quantity scarcity can be divided into two categories: scarcity due to limited supply and scarcity due to excess demand. A most common example of scarcity due to limited supply can be seen in the case of “limited edition” products (Gierl & Heuttl, 2010). Scarcity due to excess demand is often communicated with phrases like “nearly sold out” (Gierl and Huettl, 2010).

Lessne and Notarantonio (1988) studied the effects of quantity scarcity on purchase intention in the retail advertisements. The scholars found that people in the conditioned group (exposed to soda advertisement with limited availability) experienced a greater purchase intention than the unconditioned group (exposed to soda advertisement without a limited availability).

Verhallen and Robben (1994) examined the effect of scarcity on the valuation of goods. The researchers conducted an experiment of recipe books, where people were exposed to different causes of quantity scarcity (accidental, excess demand or limited supply). The main finding in this study is that scarcity due to market circumstances (excess demand or limited supply) increases the perceived product uniqueness and cost evaluation (perceived as more costly).

(16)

16

Moreover, the researchers argue that this effect might be seen as a generalization of the price-quality relationship, explained by the heuristic systematic dual processing model (Chaiken, 1980; Verhallen and Robben, 1994). This model suggests that, in times of scarcity, people rely on simple cognitive rules and perceive product price as a signal of product quality. Therefore, people prefer higher priced products as this signals higher quality. This model will be elaborated on in section 2.4.

2.3.2 Time scarcity

Limited time scarcity is identified by several scholars as a variable that influences consumer behaviour (Park et al., 1989; Nowlis, 1995; Bettman et al, 1998; Suri & Monroe, 2003). Time scarcity or time pressure is perceived as a limitation to consider information and to make decisions (Suri & Monroe, 2003). Aggarwal and Vaidyanathan (2003) examined the relation between time-limited offerings and willingness to buy. They found that time scarcity is linked to both higher product valuation and higher willingness to buy and a lower intent for extensive product research (Aggarwal and Vaidyanathan, 2003).

Bettman, Luce and Payne (1998) came up with an integrative framework for understanding consumer choice behaviour under limited time scarcity. For example, Dhar and Nowlis (1999) found that people engage in less information processing when they experience time pressure compared to consumers who do not experience any time pressure. Nowlis (1995) examined in an earlier research how consumers trade off price with quality when making product choices under time constraints. The scholar found that under time pressure consumers were inclined to choose higher-quality and higher-priced brands. An explanation for this effect, provided by Wright (1974) and Hansson, Keating and Terry (1974), could be that time constraints lead to

(17)

17

the avoidance of negative information by consumers and to risk averse behaviour. Another explanation is again provided by the heuristic-systematic dual processing model (Chaiken, 1980).

2.4 Heuristic-systematic dual processing model

Price plays an important role in product choice assessment. When buying products, price serves as a signal of sacrifice. However, in times of scarcity, price could also serve as a signal of quality. This effect is explained in the heuristic-systematic dual processing model of Chaiken (1980).

The model distinguishes between a systematic and heuristic view of persuasion. According to the systematic view, people actively engage in performing the task. Meaning that they seriously attempt to comprehend and to evaluate message arguments. Moreover, people try to assess the validity of the arguments in relation to the message conclusion (Chaiken, 1980). In other words, people both have the motivation and the ability to process information, which is referred to by Petty et al. (1983) as the central route of persuasion. In this route of persuasion people base their product choice on the quality and strength of arguments. Price is perceived as an indicator of sacrifice and the true value is derived from the products quality, obtained from a thorough assessment of product information. This is referred to as systematic information processing (Suri & Monroe, 2003). Noteworthy, Suri and Monroe (2003) stated that, under high involvement, systematic information processing decreases if time pressure increases.

In contrast, according to the heuristic view of persuasion, scarcity leads to thoughtless and automatic responses (Cialdini, 1987). People put little effort in judging message validity. In the heuristic view people rely on more accessible information, for example source’s identity or the

(18)

18

communicator´s likability (Chaiken, 1983). This view can be linked to Petty and Cacioppo (1983) peripheral route of persuasion, marked by low ability and motivation to process information. Researchers investigated the link between scarcity and cognitive effort (Rao & Monroe, 1988; Park et al., 1989; Monroe, 2003). The researchers suggest that if people experience time pressure to process product information or if they feel pressured due to scarce availability of products, they tend to use the price-quality heuristic to evaluate an offer (Rao & Monroe, 1988; Park et al., 1989; Suri & Monroe, 2003). Meaning that products become more valuable if they become less available (Cialdini, 2001; Suri et al., 2007).

Next to this, Suri and Monroe (2003) found that the effect of scarcity and motivation on perceptions of product value is mediated by the products perceived quality and sacrifices that must be made. If the price-quality relationship dominates, consumer’s perceptions of value will imitate perceptions of quality. This means that higher priced products will be perceived as of being of higher value (Suri & Monroe, 2003). On the contrary if the price-sacrifice relationship dominates, consumers perception of value will mimic their perception of sacrifice. This indicates that low sacrifices are perceived to be higher in value.

2.5 Message framing

Smith and Petty (1996) claim that marketers should focus on how to present information in an advertisement. They found that the way in which information is framed affect consumer choices and product opinions. Message framing is all about shaping a message in such a way that it has different outcomes using different perspectives (Chong and Druckman, 2007). Section 2.4.1. looks at the differences between rational and emotional message framing.

(19)

19 2.5.1 Rational versus emotional appeals

People are exposed to around 300 to 400 persuasive messages on a daily basis. Many of those messages use rational appeals, meaning messages that contain product information, facts and benefits (Albers-Miller & Stafford, 1999). For example, when selling a car, marketers often focus on product attributes such as its engine, its power, its elegance or its handling performance. According to theories of persuasion, rational appeals impact attitudes as the appeals influence the evaluative cognitions about the object. If a person is motivated and able to process persuasive information, the generated cognitions are likely to reflect inferences about the quality of the message content (Petty & Cacioppo, 1986; Chaiken et al., 1989; Roselli et al., 1995). Strong and valid arguments produce favourable cognitive elaborations whereas weak arguments result in unfavourable elaborations. If the message contains more favourable elaborations than unfavourable elaborations, the statement advocated in the message will most likely be accepted. Therefore, researchers suggest that if object based processing occurs, attitude change is mediated by the number and type of cognitive elaborations (Roselli et al., 1995). Petty and Cacioppo (1996) refer to this type evaluative elaborations of information processing as the central route of persuasion.

Petty and Cacioppo (1996) and Chaiken et al. (1989) also discuss a non-content route of cognitive persuasion. Think of source characteristics such as expertise, credibility or attractiveness. If individuals either lack the motivation or the ability to process information, peripheral processing is likely to occur. This means that messages are evaluated on surface-level characteristic (e.g. source expertise, number of arguments) rather than the actual arguments (ideas, content) of the message (Roselli et al., 1995; Suri & Monroe, 2003).

However, not all messages are intended to affect cognitions about attitude objects. Marketers could also use emotional appeals, which is referred to as messages that serve to bring about

(20)

20

emotional and experimental feelings. For example, car advertisements could, instead of focusing on rational elements, also focus on the fun, comforting, unique and exciting aspects of the car or its sexy appeal, which influence people´s emotions and feelings. The main purpose of emotional appeals is to elicit emotions such as fear, uniqueness, love, humour, popularity or pride to persuade consumers. Marketers aim to create object-related emotions and to make the consumer feel good about the product or brand (Roselli et al., 1995; Albers-Miller & Stafford, 1999). Furthermore, several scholars have found that emotional arguments disrupt cognitive elaboration of message contents and therefore attitude change occurs via non-content based routes (Roselli et al., 1995).

Van Herpen et al. (2005; 2009) examined this non-content component of value perceptions. According to their study, scarcity increases product choice and product valuation due to two social factors: (1) snob-effect and (2) bandwagon effect.

The bandwagon effect focuses on excess demand due to a desire to “join the crowd”. Researchers have found an increase in demand due to the fact that people purchase products because other people have chosen them before. An explanation is that people aim to conform with people with whom they want to be associated with (Leibenstein, 1950).

On the contrary, the snob effects focuses on a shortage in supply. The snob effect refers to the extent to which demand decreases due to the fact that other people also consume the same product. Instead, this theory focusses on the desire of people to disassociate from the crowd and to be unique. According to Lynn (1991), consumer preferences for scarce commodities that communicate uniqueness are greater than for commodities that simply communicate limitation. Researchers have found that uniqueness could serve as an element that activates the feeling of specialness, leading to a desire for personal distinctiveness and the possession of scarce commodities (Fromkin & Snyder, 1980; Snyder, 1992).

(21)

21

Chapter 3 Hypotheses development

Previous literature provided knowledge about product scarcity, message framing and message processing. Based on the existing literature an interesting research gap is found. In the first place, scarcity effects on purchase intention and product valuation has been researched, but no research has been conducted to investigate the effect of product scarcity on consumer’s willingness to pay. Next to this, current research examined differences between limited time scarcity and limited quantity scarcity communication and between rational and emotional appeals in marketing communications. However, little attention that has been paid in the literature on the interaction between product scarcity and message framing, with regard to emotional and rational appeals.

Therefore, this research will focus on how message framing, by looking at the difference between emotional and rational appeals, could affect the relation between scarcity and willingness to pay. To fill this gap, the following research question has been formulated.

How does emotional and rational appeals underlie the relation between scarcity and willingness to pay?

In this chapter the conceptual framework is presented and hypotheses are developed. The conceptual framework shows the relation between time and quantity scarcity and willingness to pay, moderated by rational or emotional message appeals. Section 3.1 will focus on the main effect between quantity scarcity and time scarcity and willingness to pay. Section 3.2 concentrates on the interaction effect of rational and emotional message appeals on the relation between product scarcity and willingness to pay.

(22)

22

3.1 Relation between quantity scarcity versus time scarcity and willingness

to pay

Restrictions, such as ‘’limited edition’’ or ‘’only available today’’ are often used in product

advertisements. Inman, Peter and Raghubir (1997) argue that the presence of a restriction can activate cognitive resources used to render either favourable or unfavourable judgments regarding the attractiveness of the promoted products. Research has found that scarcity tactics are effective and increases product preferences (Bozzolo & Brock, 1992). However, practitioners found differences in deal value when using limited time scarcity or limited quantity scarcity framing restrictions to promote the commodity (object).

Time limits restrictions, such as Final Week!, are most useful for attracting customers to the brand. In contrast, quantity restrictions such as limited edition, are necessary for stockpiling (Inman et al., 1997). Furthermore, quantity restrictions are associated with higher quality and therefore higher product valuation. In fact, brand quality perceptions of unbranded products are the highest for quantity restrictions (limited number of products offered), lower for time restrictions (offer expires at certain date) and lowest for purchase preconditions (only available with the purchase of X) (Inman et al., 1997).

Figure 1: Conceptual Model: Relation between product scarcity and willingness to pay, moderate by message framing

(23)

23

Tan and Chua (2004) investigated the effect of vague scarcity restrictions (e.g. ‘’While stock lasts’’) on deal evaluation. The researchers found that vague restrictions caused consumers to

focus on the potential loss aspect - “I can obtain a unique piece but I must do so quickly or I will lose the opportunity since I do not know when the stocks will run out” (Tan & Chua, 2004) - as those price reductions contain high uncertainty.

The prospect theory elucidates this as it posits that people tend to place more weight on outcomes that are considered as certain than on outcomes that are considered probable. However, there is a difference between gains and losses. People prefer the probability of a modest gain over the probability of a larger gain. On the contrary, when people need to choose between a modest loss or a larger loss they most often prefer the risk of the larger loss (Kahneman & Tversky, 1979). Based on this principle, Gabler and Reynolds (2013) argue that in the case of general promotional offers, people would wait for the lower price. However, when there is an infinite number available (scarcity) the researchers found that people tend to choose the certain modest gain (paying full price) over some probability of a larger gain (discount). According to this study, people forsake discounts to avoid the probability of missing the opportunity to purchase the product and therefore increases people’s purchase intention (Gabler

& Reynolds, 2013).

Based on these findings a positive relation between scarcity and willingness to pay is expected. Therefore, the following hypotheses are developed to test this:

Hypothesis 1a: Willingness to pay is higher when there is scarcity than unlimited availability.

(24)

24

3.2 Moderating influence of message framing, in terms of emotional and

rational appeals, on the relation between product scarcity and willingness to

pay

The most common used approach to influence consumer behaviour in advertising is message framing, with regard to rational appeals or emotional appeals (Johar & Sirgy, 1991). Rational appeals focus on informing consumers about one or multiple functional aspects or key benefits of the commodity. This is also named “informational advertising” (Johar & Sirgy, 1991).

Researchers have found that scarcity messages hinder people’s ability to make rational decisions (Cialdini, 1993). Consequently, people purchase products without thorough consideration or comparison with other products (Jang et al., 2015).

Emotional appeal strategies concentrate around the creation of an image for the product or for the product user. Researchers argue that scarcity increases the perceived value of a message via the peripheral route of persuasion. This is referred to as elements that facilitate persuasion through reactance, need for uniqueness or other content related processes under non-thoughtful conditions (Bozollo & Brock, 1992. Cialdini (1987; 2001) found that scarce products are deemed valuable due to hindrance-to-thinking or scarcity heuristic (automatic responding).

Van Herpen et al. (2005; 2009) examined in two researchers the image component of scarcity on value perceptions. The scholars argue that product choice and valuation is affected by social influences such as need for uniqueness or conformity. Product choice and inferences about the value of products is determined by two social factors: (1) snob-effect (supply related scarcity) and (2) bandwagon effect (demand related scarcity). The effects exist due to either a shortage in supply or due to changes in demand.

First, the snob effect is based upon shortage in supply. This mechanism implies that product exclusiveness increases quality inferences of products. In other words, individual demand

(25)

25

decreases because of the knowledge or presumption that other people consume the same product (oversupply). Scarce goods increase “snob value” as limited availability provides the

opportunity to dissociate from others (van Herpen, et al., 2005). Van Herpen et al. (2005) suggest that, in the case of scarcity due to limited supply, “need for uniqueness” mediates the

relationship between scarcity and product valuation. An explanation for this effect is that a sense of specialness and dissimilarity is evoked by unavailability due to limited supply, whereas excess demand shows high product popularity and similarity between people (van Herpen et al., 2005).

More scholars have argued that people’s “need for uniqueness” influences the relation between scarcity and valuation. The need-for-uniqueness theory deals with peoples’ behavioural and emotional responses towards information about their level of uniqueness (Fromkin, 1970; Fromkin & Snyder, 1980). Fromkin and Snyder (1980) found that the highest value is created if people feel moderately unique. The researchers propose that people tend to feel uncomfortable when experiencing extreme similarity or dissimilarity compared to others as this could lead to social isolation or social disapproval. However, the scholars also argue that there are individual differences in the desire to feel unique. The stronger the need for uniqueness, the higher people aim for dissimilarity. Lynn (1991) confirmed this view by showing that greater value is assigned if people have an internal desire for a unique social status and exclusiveness.

Second, the bandwagon effect investigates scarcity effects from the perspective of excess demand. The theory states that consumer behaviour is determined by the purchase behaviour of other consumers (product popularity). This effect is not only triggered by the direct observation of other consumers purchase behaviour but also by the traces of those behaviours. For example, empty shelf spaces or long queues. Moreover, researchers found that excess demand increases inferences about the quality of the product.

(26)

26

According to Anguirre-Rodriguez (2013) supply related scarcity appeals (snob effect) are more persuasive than demand related scarcity appeals (bandwagon effect) due to greater perceived ad credibility and will therefore result in more favourable purchase behaviour.

In line with the above mentioned findings, the following hypotheses are formulated to test moderating effect of rational and emotional appeals on the relation between product scarcity and willingness to pay:

Hypothesis 2a: Willingness to pay is stronger for emotional appeals than for rational appeals.

(27)

27

Chapter 4 Methodology

This chapter covers information about the (4.1) method, (4.2) pre-test, (4.3) research design, (4.4) variable operationalization and (4.5) measures.

4.1 Method: Population, sample and data collection

The nature of this research is explanatory as this study aims to identify the nature of the cause-and-effect relationship between scarcity and willingness to pay. A deductive approach is used to identify these presumed effects where hypotheses are derived from the information that is given in the existing literature (Evans, 1982).

The aim of the study is to expand the existing literature on product scarcity by investigating differences in willingness to pay between time and quantity scarcity and abundant availability. Moreover, this research will investigate whether emotional or rational message appeals positively or negatively influence this relationship. Therefore, the following research question will be answered: “How do rational or emotional message appeals underlie the relation

between scarcity and price sensitivity?”

To test the hypotheses that a causal relationship exists between two or more variables and to answer the research question, an experiment is designed using the research platform Qualtrics. An advantage of an online experiment is that it keeps costs down and limits time that is needed for a laboratory or field experiment. However, a downside is that the internal and external validity is expected to be lower as it is not possible to closely control variables in the experiment. Moreover, subjects are exposed to a hypothetical situation. Therefore results could lead to the hypothetical willingness to pay instead of the actual willingness to pay.

(28)

28

Data collection is conducted using the Amazon Mechanical Turk platform and Facebook. Next to this, convenience sampling is used, where subjects are selected from the researchers peers and Facebook followers. The sample has been drawn from a large population worldwide and subsequently led to variety in age, gender, nationality and education level.

In total 464 subjects successfully participated in the experiment. 48.5% of the participants are male and 51.5% are female. 54.7% of the subjects is between 25 and 40 years and 9.7% of the subjects is aged under 25. Subsequently 35.5% of the subjects is older than 40 years.

4.2 Pre-test and stimulus development

Before administering the experiment to the research sample, a pre-test is conducted among 30 subjects, who are closely related to the researcher, to determine the effectiveness of the experiment and whether questions are missing or should be eliminated. Another purpose of the pre-test is to assure that the advertisement and promoted product fit with the objective of this study.

Verhallen (1982) and Lynn (1989) argue that the scarcity effect is the strongest when (1) there is uncertainty about the intrinsic quality of a product, (2) product consumption is visible to others and (3) when the product is appealing. Therefore, the model will be tested using material goods.

At first, 10 subjects were interviewed and exposed to a cookbook as a test stimulus. However, most of the subjects in the pre-test implied that the product did not elicit any emotions of either uniqueness or popularity. The subjects indicated that a high-tech and special product would be more suitable. Moreover, the scarcity manipulation and message appeal did not seem to have an effect as the subjects only recalled the picture of the cookbook and mentioned that the

(29)

29

scarcity message and or rational/emotional message appeal conveyed in the advertisement did not appeal to the eye as it was too long and/or not catchy enough. Finally, the subjects recommended to incorporate questions about the product category and product usage.

Based on the first pre-test, the test stimuli was changed to a Fitbit and questions regarding the product category were added to the questionnaire. A Fitbit is selected as stimulus as it is a high-tech product that enables to highlight both functional product aspects (rational appeal) and symbolic aspects (emotional appeals of uniqueness and popularity). Moreover, the product is appealing, there is uncertainty about the intrinsic quality of the product and consumption is visible to others. Repeatedly, a pre-test was conducted among 20 subjects. The subjects reacted positively towards the questionnaire and stimulus. Moreover, the subjects reported that the messages claimed in the advertisement are appealing.

4.3 Research Design

To test the conceptual model an experiment is conducted. The experiment has a 3 (time scarcity, quantity scarcity, unlimited availability) * 3 (emotional appeal – snob effect, emotional appeal – bandwagon effect, rational appeal) between-subject design. In total, there are nine conditions

(table 1). Hence all conditions contained identical questions.

Rational message appeals Emotional message appeals (bandwagon effect) Emotional message appeals (snob effect) Time scarcity Time is running out!

Tested best product.

Time is running out! Everybody has it.

Time is running out! For top performers only. Quantity scarcity Limited Supply!

Tested best product.

Limited supply! on Everybody has it.

Limited Supply! For top performers only. Abundant

availability

Sufficient stock! Tested best product.

For top performers only.

(30)

30

A short (5-10 minutes) questionnaire is created in English and administrated online using the Qualtrics platform. Participants are recruited via peers of the researcher, Facebook and Amazon Mechanical Turk.

In the beginning of the experiment, the participants were informed that the research concentrates on marketing of products and its impact on consumer buying behaviour. First, participants were asked about the importance of a healthy lifestyle, their exercise habits, and about the importance of tracking their exercise behaviour. Next, all participants are confronted with the same, hypothetical, scenario.

“Imagine, recently the new trendy Fitbit Smart Fitness Watch is launched powered by Usain Bolt. The watch band is available in Black, Silver, Gold, and Pink. Take a look at the advertisement and answer the questions. You are allowed to look at the advertisement as many times as you want to”.

Afterwards, participants were randomly assigned to one of the nine advertisements (conditions), without knowing of the existence of the other conditions. They were asked to study the add carefully before continuing the questionnaire. The questionnaire included open-ended questions concerning the dependent variable (willingness to pay), and Likert scale questions regarding the perceived product quality and purchase intention. At the end of the survey, subjects were asked four demographical questions regarding their age, gender, nationality and education level. At the end of the experiment subjects were thanked for their participation (Appendix 1).

4.4 Variable Operationalization

The independent variable in this study is scarcity and is manipulated in the experiment. The subjects are confronted with either quantity- or time scarcity. Abundant availability is added as

(31)

31

a control condition in the experiment. In this condition subjects are told that there is sufficient stock available.

The scarcity message communicated is vague rather than specific. For example, “limited stock” illustrates vague quantity scarcity whereas “only three items left” defines specific quantity scarcity. Another example is provided for time scarcity, where “time is running out” exemplifies vague time scarcity and “today’s deal ends in ten hours, eight minutes and eleven seconds” specific time scarcity. Vague scarcity messages are expected to minimize differences

in perceived purchase urgency due to different perceptions of scale. Moreover, vague messages create equal feelings of uncertainty between time and quantity scarcity. According to Tan and Chua (2004), the introduction of uncertainty into the scarcity restriction (by not knowing when the scarcity condition materializes) causes people to focus on the potential loss aspect of the offer: “I must purchase the commodity quickly as I do not know when the offer is running out”.

The moderator in this study is message framing, which is split into emotional and rational appeals. Rational message appeal strongly focuses on the communication of product information (e.g. tested best product) whereas emotional message appeals strongly focus on either communicating consumer uniqueness (e.g. for top performers only) or consumer conformity with the crowd (e.g. everybody has it).

Appendix 2 presents the scarcity messages that are communicated in the nine different conditions.

4.5 Measurement of willingness to pay

Willingness to pay is defined as “the maximum amount of money a consumer is willing to pay,

(32)

32

Westendorp Method, using the Price Sensitivity Meter (PSM). This method determines pricing

strategies based on consumer’s perception of price. In other words, it is used to find the perceived true value of products. Subjects will be asked four open questions to value a product.

The first two questions are asked to determine the indifference price (perceived normal price)

(Lewis & Schoemaker, 1997; van Westendorp, 2003).

(1) At what price would you consider the product to be priced so low that you would

feel the quality couldn’t be very good? To determine the “too cheap” price.

(2) At what price would you consider the product to be a bargain—a great buy for the

money? To determine the “cheap” price.

The last two questions are asked to determine the optimal pricing point (Lewis &

Schoemaker, 1997; van Westendorp, 2003).

(3) At what price would you consider the product starting to get expensive, so that it is

not out of the question, but you would have to give some thought to buying it? To

determine the “expensive” price.

(4) At what price would you consider the product to be so expensive that you would not

consider buying it? To determine the “too expensive” price.

The questions use open-ended numeric responses as this imposes subjects own view on the true valuation of the product offered. The results will be displayed graphically for analysis by plotting cumulative percentages. Based on the graph, analyses are run.

(33)

33

Chapter 5 Data analyses and results

In this chapter the results of the data analyses are presented. First, descriptive statistics are given after which each hypothesis, as described in chapter 3, is tested and the results are discussed. Finally a robustness checks are conducted.

5.1 Preliminary analysis and descriptive statistics

Before testing the hypotheses, a preliminary data analysis has been conducted in SPSS. The first step consists of cleaning the dataset. A frequency analysis is performed to look for missing data. For all nine conditions and variables no missing data was found. Second, some cases were removed from the dataset as those cases showed contradictory responses, mainly with regard to willingness to pay. For example, subjects set the “too cheap price” above the “too expensive price”. Therefore, prices are ordered incorrectly and invaluable when using the van Westendorp

method (Apollonsky, 2016). In total 54 cases are removed from the dataset. After removing the contradictory cases, the dataset contains 464 correct cases (N=464).

Next, the variable Nationality is recoded. The variable was a continuous variable and is recoded to a categorical variable of (1) Europe, (2) North America, (3) South America, (4) Asia and (5) Africa.

5.1.1 Operationalization of the dependent variable – Willingness to pay

A price sensitivity analysis is performed as described by van Westendorp (1976). The Price Sensitivity Model (PSM) works by asking consumers four questions to determine consumers

(34)

34

interest in the product. Basically, the questions determine the too cheap-, cheap-, expensive- and too expensive price. Next the cumulative frequencies of the values of the four variables

TooCheap, Cheap, Expensive and TooExpensive are aggregated and graphed into a price map

(Figure 2). The cumulative frequencies of the variables TooCheap and Cheap are inverted in order to create four intersection points that produce a range of prices the market considers acceptable (figure 2, range of acceptable prices). The intersection points shown in the graph are the optimal price point (OPP), indifference price point (IDP), point of marginal cheapness (PMC) and point of marginal expensiveness (PME).

Since the van Westendorp analysis yields four values per case, the data needs to be collapsed into one value to be used as a dependent variable in the model. To achieve this the following approach is used. To start with, the optimal price point is used as a reference point (value $75). Based on this value, the variables TooCheap, Cheap, Expensive and TooExpensive are aggregated to one variable (Price) such that the mean gives a similar result to the optimal price point of $75 The mean of the variables TooCheap, Cheap and Expensive most closely reflect the optimal price point with a value of $85.74. Descriptive statistics and a histogram with a normal curve of Price are presented in table 2 and figure 3.

(35)

35

Variable N Minimum Maximum Mean SD Variance Price 464 5.00 2200.00 85.74 116.12 13484.23

The results show high variations for the variable Price. Moreover, the variable is positively skewed (long right tail). To normalize this variable, the natural logarithm of the variable Price is taken. An overview of the descriptive statistics and histogram with normal curve are provided in table 3 and figure 4.

Variable N Minimum Maximum Mean SD Variance Price_LN 464 1.61 7.70 4.17 0.73 0.54

Table 2: Descriptive Statistics Price

Figure 3: Histogram with normal curve of Price

Table 3: Descriptive Statistics Price_LN

(36)

36

No other variables needed to be recoded as all the questions in the questionnaire are positively keyed items, where agreement with the item represent a relatively high level on the scale. For example, from (1) Strongly disagree to (7) Strongly agree.

A fourth preliminary step is conducting a manipulation check to test whether the message appeals were effective stimuli. As two statements were used to indicate subjects´ attitude towards the product and advertisement with regard to the snob effect and the bandwagon effect, the mean of the two items is taken. A comparison is made between the means of the snob conditions and the non-snob conditions. The same is done for popularity, differences in mean are investigated between bandwagon conditions and non-bandwagon conditions. Next, an independent t-test is conducted to find out whether the means of the snob- and/or bandwagon effect in the conditioned groups statistically differ from the unconditioned groups. The results are presented in table 4 till table 7.

Variable G_Unique N Mean SD

Std. Error Mean Avuniqueness 1 156 4.256 1.477 0.118

2 308 3.945 1.477 0.084

Table 4: Descriptive Statistics Snob Effect (Group 1 = snob condition; Group 2 = other conditions)

(37)

37

The results from table 4 and 5 show that t is equal to 2.135 and it has 462 degrees of freedom. The significance level is p=0.033. Therefore, null hypothesis should be rejected (p=0.033<p=0.050). This means that the snob effect in the conditioned groups (group 1) is different from the unconditioned groups (group 2). Accordingly, it can be assumed that the snob effect manipulation was successful.

The results for the bandwagon effect are presented in table 6 and table 7. Table 7 shows that t is equal to 1.559 and degrees of freedom are 462. The significance level is p=0.045. Therefore, the null hypothesis should be rejected (p=0.045<p=0.050). It can be assumed that the bandwagon manipulation was effective.

To test whether there are differences in perceived quality and purchase intention between scarcity and abundant availability, an independent t-test is conducted. The results in table 8 and 9 show statistical insignificance at a 5% level for AVquality (p=0.475>p=0.050). The results indicate that the perceived quality is not significantly higher when there is scarcity compared to no scarcity.

Variable G_Popularity N Mean Std. Deviation

Std. Error Mean

AVpopularity 1 156 3.914 1.445 0.116 2 308 3.612 1.561 0.090

Table 6: Descriptive Statistics Bandwagon effect (Group 1 = bandwagon condition; Group 2 = other conditions)

(38)

38

Table 10 and table 11 provide the outcomes for the perceived difference in purchase intention between scarcity conditions and the control condition. Again, no significant differences in means are found (p=0.864>0.050).

A possible explanation for insignificance for the variables AVquality and AVpurchaseintention could be that quality and purchase intention are triggered by message appeal rather than by the availability of products.

In order to determine scale reliability or internal consistency of perceived product quality, uniqueness, popularity and purchase intention (7-item Likert scale), Cronbach’s alpha is

Variable G_Quality N Mean Std. Deviation Std. Error Mean

AVquality 1 157 5.112 1.034 0.825

2 307 5.181 0.965 0.551

Table 9: Independent samples T-test, t-test for equality in means

Table 8: Descriptive Statistics AVquality (Group 1 = control condition; Group 2 = scarcity conditions)

G_PI N Mean Std. Deviation Std. Error Mean AVpurchaseintention 1 157 3,828 1,464 0,117 2 307 3,803 1,508 0,086

Table 10: Descriptive Statistics AVpurchaseintention (Group 1 = control conditions; Group 2 = scarcity conditions)

(39)

39

calculated. A Cronbach’s alpha analysis is performed for the variables AVuniqueness,

AVpopularity, AVquality and AVpurchaseintention (7-item Likert scale), to check whether all

items on the scale measure the same. Table 12 provides an overview of the outcomes of the reliability test. All reliability coefficients are perceived as reliable as the coefficients are higher than 0.70. The reliability of an item could not be increased by deleting one or more items of the scale.

AVquality ` 0.840

AVuniqueness 0.750

AVpopularity 0.752

AVpurchaseintention 0.714

A final preliminary step is the calculation of the means and standard of the variables in the model. The results are exhibited in table 13.

Variabele M SD Price_LN 4.17 0.73 Scarcity 1.98 0.71 Appeal_type 2.01 0.72 Gender 1.52 0.50 Age 2.49 0.99 Nationality 2.24 0.85 Education 3.92 0.85

Variable Cronbach’s Alpha

Table 13: Summary descriptive statistics of Means, Standard Deviations and Correlations Table 12: Cronbach’s Alpha

(40)

40

5.2 Hypothesis testing

In order to test the hypotheses 1a, 1b, 2a and 2b, a t-test and factorial ANOVA analyses are conducted. First descriptive statistics are presented in table 14. The means of time scarcity (M=4.143) and quantity scarcity (M=4.178) are slightly lower than the mean of abundant availability (M=4.187). Based on nearly equal means, significant differences between abundant availability and scarcity are not expected.

A t-test is performed to test whether significant differences in willingness to pay can be found between any type of scarcity (quantity scarcity and time scarcity) and abundant availability. The results are presented in table 15 and table 16.

Availability level N Mean Std. Deviation

Price_LN ₁ _{157 4.187} _0.715

2 307 4.161 0.736

DV: Price_LN

Scarcity Appeal_type M SD N

Abundant availability Rational appeal 4.091 0.678 54 Snob effect 4.297 0.730 53 Bandwagon effect 4.176 0.738 50

Total 4.187 0.715 157

Quantity Scarcity Rational appeal 4.012 0.679 49 Snob effect 4.252 0.822 58 Bandwagon effect 4.250 0.771 52

Total 4.178 0.767 159

Time Scarcity Rational appeal 4.024 0.646 49 Snob effect 4.327 0.549 45 Bandwagon effect 4.097 0.837 54

Total 4.143 0.704 148

Table 14: Descriptive statistics Scarcity, Appeal, dependent variable Price_LN

(41)

41

The results show that the main effect of scarcity on willingness to pay is, as expected, not significant (p=0.711). Therefore, no significant differences in willingness to pay exist between any type of scarcity and abundant availability.

A 3 (abundant availability, quantity scarcity, time scarcity) * 3 (rational appeal, snob effect, bandwagon effect) factorial ANOVA analysis is performed to investigate the relationship between quantity scarcity, time scarcity and abundant availability more deeply. Moreover, it will be tested whether message framing, with regard to rational and emotional appeals, affect the relationship between scarcity and willingness to pay.

The results of the analysis are shown in table 17. The results show no significant differences in willingness to pay between the three types of scarcity, which confirms the results of the t-test (F=0.107, p=0.898, η2=0.000). The analysis does show a significant main effect, at a 5% significance level, of message appeal type on willingness to pay (F=4.539; p=0.011,η2_=0.020).

However, the interaction term of scarcity*appeal_type is not significant (F=0.204, p=0.818, η2_{=0.003). This means that the relationship between scarcity and willingness to pay is not}

significantly influenced by message framing. The explanatory power of this model is 0.023, meaning that 2.30% of the variance in Price_LN is explained by the model.

(42)

42

Figure 5 shows a graphical presentation using the estimated marginal means. The results indicate that willingness to pay is the lowest when communicating messages using rational appeals. However, the graph also suggests that rational message appeals in combination with abundant availability leads relatively to the highest willingness to pay compared to quantity scarcity and time scarcity.

Furthermore, the graph shows that a communication focus on the snob effect (uniqueness of the consumer) leads relatively to the highest willingness to pay when there is time scarcity or abundant availability compared to quantity scarcity. Focusing on the bandwagon effect (popularity) seems to lead to the highest willingness to pay when there is quantity scarcity. However, no statistical support has been found for these differences.

DV: Price_LN

Variables Sum of squares df Mean Square F Sig. η2

Corrected Model 5.600a 8 0.700 1.326 0.228 0.023

Scarcity 0.113 2 0.057 0.107 0.898 0.000

Appeal_type 4.791 2 2.396 4.539 0.011 0.020 Interaction Term 0.817 4 0.204 0.387 0.818 0.003

a. R Squared - 0.023 (Adjusted R Squared = 0.00

Table 17: 3*3 Factorial ANOVA analysis

(43)

43

To conclude, based on the data retrieved in this research, it can be assumed that there is no significant difference in willingness to pay between abundant availability and (any type of) scarcity. Therefore hypothesis 1a (willingness to pay is higher when there is scarcity than unlimited availability) should be rejected.

The interaction term is also insignificant, indicating that the relationship between scarcity and willingness to pay is not significantly influenced by message framing, in terms of rational and emotional appeals. Therefore hypothesis 2a (Willingness to pay is stronger for emotional appeals than for rational appeals) should also be rejected.

However, the research does show a main effect of message framing on willingness to pay. Willingness to pay seems higher when emotional appeals rather than rational message appeals are used.

A 3 (abundant availability, quantity scarcity, time scarcity) * 2 (rational appeals, emotional appeals) factorial ANOVA analysis is run to test for significance. The results presented in table 18 show significant differences between rational and emotional appeal on willingness to pay (F=6.853, p=0.009)

A single factor analysis of variance is conducted to test hypothesis 1b. The summary statistics are presented in table 19 and the results of the factor analysis in table 20.

DV: Price_LN

Variable Sum of squares F Sig. η2

Corrected Model 0.785 1.486 0.193 0.016 Scarcity 0.094 0.178 0.837 0.001 Appeal_type 3.162 6.854 0.009 0.015 Interaction term 0.076 0.145 0.865 0.001

a. R squared = 0.016 (Adjusted R Squared = 0.005)

(44)

44

Variable Mean SD N

Quantity Scarcity 4.178 0.767 159

Time Scarcity 4.143 0.704 148

The results of table 19 demonstrates that willingness to pay is slightly higher for quantity scarcity than for time scarcity. However, the results in table 20 show insignificance at a 5% significance level (p=0.678, F=0,.73, η2_{=0.001). This assumes that quantity scarcity not}

significantly leads to a higher willingness to pay compared to time scarcity. Hypothesis 1b (willingness to pay is higher when there is quantity scarcity than time scarcity) should therefore be rejected.

Finally, a 3 (abundant availability, quantity scarcity, time scarcity) x 2 (snob effect, bandwagon effect) factorial ANOVA design is conducted to test hypothesis 2b. The results of the summary statistics are presented in table 21 and the results of the factorial ANOVA analysis in table 22.

DV: Price_LN

Variable Sum of squares df Mean Square F Sig. η2

Scarcity 0.094a 1 0.094 0.173 0.678 0.001

DV: Price_LN

Scarcity Appeal_type M SD N

Abundant Availability Snob effect 4.297 0.730 53 Bandwagon effect 4.176 0.738 50

Total 4.238 0.733 103

Quantity Scarcity Snob effect 4.253 0.822 58 Bandwagon effect 4.250 0.771 52

Total 4.252 0.795 110

Time Scarcity Snob effect 4.327 0.549 45 Bandwagon effect 4.097 0.837 54

Total 4.202 0.726 99

Table 20: Single Factor analysis Table 19: Descriptive Statistics

(45)

45

The results of appeal type are insignificant at a 5% significance level, both for the main effect (p=0.169, F=1.899 η2=0.006) as for the interaction term (p=0.554, F=0.593, η2=0.004). Hypothesis 2b (willingness to pay is stronger for the snob effect than for the bandwagon effect) should therefore be rejected.

5.3 Robustness Check

In order to diagnose whether the findings hold under different assumptions, a robustness check is performed. If the significance of the regressors does not change a lot it serves as evidence that the findings are robust. In order words there is evidence of structural validity of the core regression coefficients, which increases the validity of the results (Lu and White, 2013). In total, three robustness checks are performed. In the first robustness check, control variables are added to the model. For the second robustness check, the dependent variable price is adjusted, and for the final robustness check the dataset is split into low and high involvement purchases.

5.3.1. Robustness check 1: adding control variables to the model

The first robustness check is performed by adding control variables to the model. According to Filatotchev and Mickiewicz (2001) control variables should be included in the analysis in order to deal with the possibility that multiple variables conjointly affect willingness to pay. The control variables added to the model are Gender, Age, Nationality and Education.

DV: Price_LN

Variable Sum of squares df Mean Square F Sig. η2

Corrected Model 1.814a 5 0.363 0.639 0.670 0.010 Scarcity 0.081 2 0.041 0.072 0.931 0.000 Appeal_type 1.078 1 1.078 1.899 0.169 0.006 Interaction term 0.673 2 0.336 0.593 0.554 0.004

The relation between scarcity and willingness to pay moderated by message framing