Irrationally predictable? : an insight in the irrational behavior of consumers when rejecting product innovations

(1)

Irrationally predictable?

An insight in the irrational behavior of consumers when rejecting product innovations

Jonathan Parr

Student number: 10901825

University of Amsterdam | Amsterdam Business School

Master Thesis MsC Executive Programme of Management Studies Marketing Management Track

(2)

Statement of Originality

This document is written by Student Jonathan Parr 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)

List of Tables

1. Frequencies 27

2. Results of experimental groupsfor willingness to pay (WTP) and likeliness to click (LTC)

33

3. Means, Standard deviations and Correlations of study variables 52 4. Results and bootstrap results of level of information on current monthly

costs.

56

5. Results of car ownership as a predictor for willingness to pay and likeliness to click further.

56

6. Bootstrap results of car ownership as a predictor for willingness to pay and likeliness to click further.

57

7. Results and bootstrap results of car ownership as a predictor for first impression of private lease.

57

8. Results and bootstrap results of car satisfaction as a predictor for first impression of private lease

58

9. Results and bootstrap results of first impression of private lease as a predictor for likeliness to click further

58

10. Results of social proof as a predictor for willingness to pay and likeliness to click further

59

11. Bootstrap results of social proof as a predictor for willingness to pay and likeliness to click further

59

12. Results of scarcity as a predictor for willingness to pay and likeliness to click further

60

13. Bootstrap results of scarcity as a predictor for willingness to pay and likeliness to click further

60

14. Results of authority as a predictor for willingness to pay and likeliness to click further

61

15. Bootstrap results of authority as a predictor for willingness to pay and likeliness to click further

61

16. Results of reciprocity as a predictor for willingness to pay and likeliness to click further

(5)

17. Bootstrap results of reciprocity as a predictor for willingness to pay and likeliness to click further

62

18. Results of consistency as a predictor for willingness to pay and likeliness to click further

63

19. Bootstrap results of consistency as a predictor for willingness to pay and likeliness to click further

63

20. Results of liking as a predictor for willingness to pay and likeliness to click further

64

21. Bootstrap results of liking as a predictor for willingness to pay and likeliness to click further

64

22. Results and bootstrap results of consistency as a predictor for first impression of private lease

65

23. Mediation effect of first impression on the relationship between consistency and likeliness to click.

(6)

List of figures

1. Resistance hierarchy model 5

2. Behavioural reasoning theory 6

3. Loss aversion 8

4. Hindsight bias 10

5. Conceptual model 17

6. Homepage for control group 22

7. Homepage for experimental group “authority” 22

8. Indication of current monthly costs based on level of information provided

27

9. Willingness to pay for private lease for car owners and non-owners 29 10. Means of willingness to pay for all experimental groups 32 11. Means of likeliness to click for all experimental groups 32

(7)

Abstract

High failure rates for product innovations are a significant challenge for companies when introducing these innovations. When discussing the possible causes for failure, several aspects are mentioned related to the product. However, consumers also reject products because of behavioural reasons. Some of which some are linked to consumer irrationality. This irrationality consists of many aspects, the most important for this study being confirmation bias, loss aversion and, related to that, the endowment effect. A case study of private lease of cars, being a high involvement product innovation, was designed. This study consists of an experiment under 286 respondents by means of an online survey. Participants were asked for their response to private lease in general and a specific private lease homepage. The results showed that the endowment effect plays a large role when consumers are confronted with the option of adopting private lease. In the experimental design, consumers who currently owned a car were willing to pay significantly less for private lease then consumers who did not own a car. A wide range of influencing strategies were used in the research design, but failed to reduce the endowment effect for private lease. However, irrationality also worked in favour of adoption. This study showed that consumers were more inclined to search for further

information on private lease if a they had a good first impression of the product. This was the case even if this first impression was based on a peripheral cue, such as the influencing strategy of consistency. Several suggestions for further research on possible ways of reducing the endowment were given based on the results of this study.

(8)

1. Introduction

When deciding which product to choose, companies have long expected consumers to make a rational comparison of what the company has to offer. Consumers are expected to be looking to maximize their utility when making decisions (Von Neumann & Morgenstern, 1944). However, consumers are consistently making choices which do not fit this expectation. They are fooled by arbitrary anchors (Prelec et al., 2003), search for information that confirms their own views (Nickerson, 1998) and are extremely loss–averse (Kahneman et al., 1975). When making decisions consumers use mental shortcuts, or heuristics, and are therefore biased in making these decisions (Thaler, 1980). Although many of these biases are useful in certain social settings or when people lack time, there are also disadvantages for both consumers and companies because of these biases. The consequence is that simply maximizing the utility of a product might not be enough to persuade consumers to purchase the company’s products (Kahneman & Thaler, 2006). In several researches (Ram, 1987; Sheth & Stellner, 1979; Foxall, 1994), consumers show no desire to change when they are satisfied with their current situation. As this prevents consumers from looking for and evaluating better alternatives, it can lead to consumers sticking with inferior products for longer than necessary.

A good example of the influence of heuristics and biases is the challenge companies face when introducing product innovations. We know that companies are faced with high failure rates when introducing these product innovations (Gourville, 2006). There is a wide range of possible reasons why consumers resist product innovations (Kleijnen et al. 2009; Ram & Sheth, 1989). Obviously, some products are not an improvement of their predecessors in the eyes of the consumer, or the product is more expensive. Companies also sometimes

overestimate the improvements in their own products. However, consumers also can also act irrationally when facing the adoption of product innovations, by being biased in their decision making. As Gourville (2006) argues, consumers are not aware of these biases. Current

literature offers little insight in the influence this might have in the adoption process.

One of the most prominent theories on decision making (Kahneman & Tversky, 1975) shows that losses loom larger than gains. This makes it likely that consumers will consider “losing” their current product in a different light then the possible gains of a product innovation. This can be made clear in the difference in consumers’ willingness to pay for new products and

(9)

their current products, and is described as the endowment effect (Kahneman et al.,1991). Additionally, it might be case that endowed consumers are not even willing to continue in the buying process by searching for further information (Talke and Heidenrich, 2014). Current literature on consumer biases in decision making is mostly descriptive in nature. These researches do not show what companies, who are obviously keen on improving the acceptance of new products, can do to reduce the aforementioned behavioral effects. This despite a great deal of research in marketing on influencing strategies (e.g. Cialdini, 2007), its antecedents and limitations (Griskevicius et al., 2009: Goldstein et al. 2008; Schultz et al., 2007) and the possible advantages some of these heuristics and biases might offer companies when introducing their products. In this thesis I will address this gap in current research by looking at the effects of consumer’s irrational decision making on the adoption of product innovations, and ways to influence this. I will do this by answering the following research

question: “How does consumer irrationality influence consumer’s adoption of product

innovations through a reduced willingness to pay and reduced search for information? How successful are influencing strategies in reducing this endowment effect? And how can companies use consumer irrationality to their advantage when introducing product

innovations?”. Firstly, the current literature will be reviewed on the main concepts, and these will be related to each other. Based on this review, the conceptual model and hypotheses will be drawn up. Along with the research design and method section they will be offered in chapter three. Results from the collected data are presented in the following chapter. In chapter five, the most important conclusions and implications of the results of this study are discussed along with the most important limitations and suggestions for further research.

(10)

2. Literature review

This chapter provides an overview of the concepts used in this study, and relevant findings on these concepts in current literature. Firstly, consumers adoption of production innovations is discussed. In the following paragraph, an overview of the key concepts on irrational consumer behavior is provided. After that, further relevant examples of irrational consumer behavior are given. Willingness to pay is discussed in the fourth paragraph. The final paragraph offers an overview of the available influencing strategies.

2.1 Consumers adoption of product innovations

To understand why consumers reject or adopt certain product innovations, it is important to understand consumer decision making. Traditionally, the factors mentioned in literature are

cognitive (a person’s belief/knowledge about something), affective (their feelings/emotions)

and conative (the way a person behaves in a certain situation) (Eagly & Chaicken, 1993). Or more simply, how a person learns, feels and does (Ramond, 1976). Another main contribution to this topic is the elaboration likelihood model (Petty & Cacioppo, 1986). This describes under which conditions consumers are influenced by different cues. This model shows that the ability and motivation to process are key elements before engaging a cognitive processing route. When consumers are less able or motivated to process information, they are more likely to be persuaded by their affective judgment based on peripheral cues. This means consumers can be influenced by cues which might not be relevant to the issue. They argue that only cognitive processing can lead to an enduring attitude change. Petty and Caccioppo (1986)

argue that “the central route views attitude change as resulting from a diligent consideration

of information that is central to what people feel are the true merits of the advocacy” (p. 3). Terms such as comprehension and learning are key in the central route. More recent research has shown that these routes are not fully separate but have an overlap (Sparks et al., 2013).

Consumer’s attitude and purchase intentions towards corporate social responsibility were both

influenced in ways which could be attributed to the central route (reviews, trust in the resort) as well as more peripheral cues, such as award logos.

Despite this knowledge of the consumer decision making process, there is still a large failure rate in the introduction of product innovations. Although the exact numbers differ between studies, most cite numbers between 40% and 90%, depending on the industry. (Andrew &

(11)

Sirkin, 2003; Gourville, 2006; Kleijnen et al., 2009). Companies therefore still lack

knowledge on how consumers evaluate product innovations. Garcia et al. (2007) show that current models on consumers decision making (such as Petty & Cacciopo, 1986) fail to fully address the issues companies face when introducing product innovations. To understand what is missing in current literature, I will first look at what is available.

In adopting new products, two key concepts are described by Manning et al. (1995). They show that consumers are looking to make their own decision when adopting new products as well as seeking a novel experience. They show that novelty seeking is related to the earlier part of the adoption process and independent judgment making with the latter decision process. This shows that when introducing new products, focusing on novelty might only produce short term positive effects. Of course consumers will accept some new products, of which Gourville (2006) argues that there are two ways of looking at change in a product innovation. One is change in the product specifications and the second is behavioral change in using the product. The products that require little change in consumer behavior can be very successful; products requiring high behavioral change for consumers can be successful if changes in the product specifications are high enough. These products are called long hauls, and typically require a long term implementation strategy for companies. The fact that positive evaluation of a new product can occur after a longer period of experiences, is

consistent with the findings of Warlop et al. (2005). They show that consumers can learn from experiencing a product, especially by introducing extrinsic cues which remind them of their prior evaluation of the product. Labrecque et al. (2015) argue that consumers habits play a large role in adoption of product innovations. Products which conflict with existing habits have a higher chance of failing, whilst products which were integrated in to existing habits were more successful. Moreau et al. (2001) report that both existing knowledge and

innovation continuity play a major role in the individuals adoption process. They argue that knowledge is more of a limitation in discontinuous innovations, which many failed product innovations are.

Although learning (Gourville, 2006; Warlop et al., 2005) might increase success rates, not all products are suited for a learning curve or can function without behavioral change.

(12)

reducing this when introducing new products. To understand more about the adoption of product innovation, it is important to understand why consumers resist product innovations. Ram and Sheth (1989) offer two explanations for consumers resisting product innovations: it challenges their prior beliefs or the innovation will change their daily routines. These

behavioral problems show that assessing product quality, although important, is an inadequate

measure of judging an innovation’s chance of success. They also suggest that there are five

major barriers that create consumer resistance: usage, value and risks barriers (functional) and tradition and image barriers (psychological), as well as offering strategies to overcome these barriers. Although useful, these barriers offer very little in terms of the behavioral challenges consumers face.

Kleijnen et al. (2009) offer their resistance hierarchy (see figure 1) in which they research the main antecedents of resistance to product innovations. The findings show that risk (mostly functional and economical), product usage patterns, perceived image and traditions and norms are the most mentioned antecedents by consumers. The antecedents lead to different

resistance types (see figure below). Perhaps due to the survey nature of the study , there is little irrationality found in the outcomes. However, the types of rejection offer an interesting possibility for further research on the effects of irrational behavior, to further understand the rejection based on this behavior.

(13)

Figure 1: Resistance hierarchy model

Rejection often occurs when the products are incongruent with their predecessors, or offer some psychological newness (Jhang et al., 2012). This is consistent with the argument of Gourville (2006) that products which require large behavioral change of consumers, show the highest failure rates. Products which are more similar to its predecessors are less likely to be influenced by consumers biases such as the endowment effect, as these have less uncertainty and require lower behavioral change from consumers (van Dijk & van Knippenberg, 1996; Chapman, 1998). Another concept which is often relevant when choosing a new product are

switching costs: “the costs consumers associate with switching from one provider to another “

(Burnham et al., 2003). Even when staying with the same brand or product category, some of these costs might apply. Examples of these are learning costs and evaluation costs. All types (procedural, financial, relational) of switching costs provided by Burnham et al. (2003) must be addressed by companies in their strategy when introducing new products and are part of the cognitive process, mostly as a risk to consumers. Ram and Sheth (1989) offer an example of switching costs (diesel cars) as very active resistance to product innovations.

(14)

Figure 2: Behavioral reasoning theory

They show that reasons against adoption have a twice as strong influence on consumers adoption intentions than reasons for adoption have on their adoptions intentions. This is consistent with Kahneman et al. (1991) findings on loss aversion, and shows that consumers indeed value losses in product adoption more highly than gains. To understand what drives consumers to their behavior, it is essential to first look at the consumers context (Simonson & Tversky, 1992) and their current situation. Bell (1985) showed that the status quo is an

important reference point for individuals, whilst Falk et al. (2007) found that people prefer the existing situation, regardless of whether the alternative has higher utility, confirming the arguments by Gourville (2006), Samuelson and Zeckhauser (1988) and Hartman et al. (1991) concerning the effect of the status-quo bias.

Talke and Heidenreich (2014) argue that consumers show either passive or active resistance. Passive resistance results from a more generic disposition to resist innovations which results prior to the evaluation of the product, whilst active resistance results from product evaluation. They also argue that the product consumers currently posses is a central specific factor in the product adoption process. Szmigin and Foxall (1998) also found that when the attachment to current products is so strong, that alternatives (such as new innovations) do not get considered by consumers. High satisfaction with the current product increases that status-quo bias

(Szmigin & Foxall, 1998; Falk et al., 2007) which means that improving existing products of high quality might be even more difficult then replacing poor products.

Chatterjee and Heath (1996) argue that trade-off size and reference states influence a

consumers decision in choosing a new product. They also show these findings are consistent with loss aversion and framing theory. Therefore it is likely that because of biases, many new

(15)

products fail to be a success, despite them being objectively better than current offerings. Gourville (2006) argues that most consumer resistance stems from behavioral change required when adopting the new product. When introducing new products which are incongruent with previous products, Jhang et al. (2012) demonstrate that positive affect, future frame,

consideration of multiple alternatives, and a benefit rationale increase evaluations of a new product. Falk et al. (2007) show that consumers who are highly satisfied with their current offering, will show stronger resistance to product innovations, even if it provides more utility. They argued that consumers satisfaction of their current product (in their case an offline channel) would influence their adoption of a new product (an online channel). This can also be seen in the different view that buyers and sellers of the same product have of the traded item. Nayakankuppam et al. (2005) showed that sellers are more focused on the positive effects of a product, whereas buyers are relatively more focused on the negative effects. Talke and Heidenrich (2014) argue that companies must anticipate on these consumer responses if they want their new products to be adopted. Current literature in marketing only recently has started looking at alternative routes of consumer decision making based on heuristics and biases. To fully understand the power these have on consumer decision making, an overview of irrational consumer behavior is necessary. In the following paragraph I will address the most common biases and heuristics, and the most important theory these are based on.

2.2 Irrational consumer behavior

Two psychologists from Israel have shown that there is another path that people take when making decisions. Kahneman and Tversky (1975) introduced prospect theory in which they showed that decision making by humans is not in convergence with the expected utility model which was dominant in economics until then. They showed that decisions are often viewed in terms of gains and losses. Their work in the next decades would consistently show that people use all kinds of heuristics and biases in their decision making, even if these are not consistent with their cognition or affective/emotional state. This was highly inconsistent with the earlier theories, such as Zajonc and Markus (1982). Kahneman et al. (1991) demonstrated that loss aversion is a key concept in decision making. An example of this was showing the existence of the status-quo basis. Here subjects in their experiments were shown to have a irrationally strong preference to remain at the status quo in certain situations, because the disadvantages of leaving loom larger than the advantages (see figure 3).

(16)

Figure 3: Loss aversion

Although these effects have shown to hold in several conditions, there are boundaries to their influence (Novemsky & Kahneman, 2005). The key concept of loss aversion is that people experience pain when giving something up. In situations where goods are consciously given up, loss aversion does not generally affect the subject. A useful example of this is the loss of money when purchasing a product. Companies are generally involved in these types of transactions and buying a new product generally involves a money transaction. However, the loss aversion discussed in this thesis is not related to this transaction, but to the transaction between two products: old and new.

Another example of prospect theory on consumer behavior is found in a study by Hartman et al. (1991), who showed that consumers require monetary compensation for switching to a

more reliable electricity provider. They argue that consumers are “irrationally reluctant to move from the status quo” (p.160/21). As shown in this study of Hartman et al. (1991), biases

and heuristics can heavily influence a consumer’s adoption of new products. They showed that an increase in the quality of a new product actually requires a compensation towards the consumer, because they are moving from the status quo. This study also showed that

willingness to pay is a useful measure to discover this irrational behavior by consumers.

Kahneman et al. (1991) also introduced the endowment effect, which is based on their concept of loss aversion. Zhang & Fishbach, 2005 provide a definition of the endowment effect “the gap between the price that buyers are willing to pay in order to acquire an object

(17)

and the price that sellers would demand in order to part with this object” (p. 1). The most important aspect of this effect is that people are concerned about losing what they already have. This finding gave a richer understanding of the practical implications of prospect theory. It showed that consumers experienced a certain pain when giving up a product they owned, even if had been randomly assigned to them. Interestingly, this shows that people do not see more appeal in the good or product they already own, but are mostly driven by the pain of giving up the product.

Saqib et al. (2010) discovered that there are moderators for the endowment effect. The level of involvement with a decision moderates the strength of the endowment effect: higher involvement will lead to a stronger endowment effect and lower involvement to a reduced endowment effect. According to Lastovicka and Gardner (1978), a low involvement product

class in one in which “most consumers perceive little linkage to their important values and is

a product class in where there is little consumer commitment to brands. Less frequently purchased products, and more brand differentiated product classes, such as automobiles or stereo equipment are often given as examples of high involvement product classes. Classic marketing theory shows that consumers are engaged in a more central route of persuasion when in high-involvement situations (Petty, 1981). There have been arguments that the effect of heuristics and biases are mostly found in low-involvement situations (Petty & Cacioppo, 1986; Zhang & Zinkhan, 2006). However, Saqib et al. (2005) found that the endowment effect is greater with high involvement products. It therefore is possible that ways of

influencing that might or might not work in this study, do work for other products. The effects in different conditions of involvement are unclear until now. Additionally, Mandel et al. (2002) found that motivational factors also have influence on the strength of the endowment effect. The results of Mandel et al. (2002) study showed that a difference in transaction

demand, or “the motivation to complete a transaction” (p. 1) had a significant effect on the

strength of the endowment effect.

Fischhoff (1975) found that people are poor judges of their decisions when the outcome is available to them, called hindsight bias. They overestimate their prior knowledge based on the outcome. This effect could influence the endowment effect when people do not evaluate a new product positively. Consumers are naturally inclined to prefer the status quo, and then

(18)

confirm their own feelings with the hindsight bias. Lee et al. (2006) showed this particular finding in an experiment in which participants evaluation of a new product (beer) was influenced by the moment they received information (see figure 4). These findings are also consistent with the confirmation bias (Nickerson, 1998) showing that people seek or interpret evidence in ways that are consistent with to existing beliefs or expectations. People also go searching for evidence that confirms their prior beliefs, which makes it even more difficult to change these beliefs.

(19)

2.3 Other forms of irrational consumer behavior

Although much work in the field of irrational consumer behavior is based on prospect theory, the past few decades have seen many more additions to the field. Thaler (1978; 1980) offers strong examples in consumer behavior based on discussed theories. He argues that people under weigh opportunity costs in situations where they either sell or purchase a certain good.

He finds another consumer bias in the “sunk cost fallacy”, where consumers fail to see that

the historical costs are economically irrelevant to their decision. As these costs cannot be compensated by future behavior, rational decision makers would not take these into account. However, Okada (2006) argues that consumers do not act rationally when thinking about sunk costs when upgrading to a new product. This could partially explain the lower acceptance rates of product innovations. Garland and Newport (1991) also found that individuals evaluate the lost investment from abandoning a project with a reference state that includes the total resources that they had initially allocated to the project. This meant subjects disregarded the

absolute monetary loss. This is highly consistent with Kahneman and Tversky’s (1984)

findings on mental accounting, which shows that how people perceive a transaction (costs or loss) heavily impacts their decision making in correspondence with prospect theory. It again shows the strong effect loss aversion has on consumer decision making.

Ariely et al. (2006) show another example of consumer irrationality by introducing the concept of arbitrary cohesiveness. Consumers can be seen behaving consistently to their prior beliefs, despite challenging the correctness of these prior beliefs. In line with these findings, Hoeffler et al. (2006) discusses the role early experiences have on people’s choices. A more favorable initial experience will decrease the amount of search for alternatives. This means early success for product innovations is critical, as it will decrease consumer search.

Consumers have also shown a particularly strong preference for free options (Shampanier et

al., 2007). This “zero-price” effect has a large pull factor on consumers, even when other

products offer greater utility. Message framing has also shown to have a strong impact on the decision consumers make. This is based on the work of Kahneman and Tversky (1981), who showed that people produces predictable changes in choice when a problem is framed in different ways. Ganzach and Karsahi (1995) followed this up in a business setting and showed that loss-framing had a much larger effect on consumers than gain framing, in their

(20)

innovations, that marketers know what the gains and losses are for consumers, and understand that the losses loom larger than the gains.

Framing the aspects of a product also offers opportunities when introducing innovations. Levin and Gaeth (1988) showed that framing effects can be influenced by the timing of the message and introducing samples before consumption. Park et al. (2000) discussed that the way of framing additional options in products influences the eventual number of options chosen by consumers. When comparing additive option framing or subtractive option framing, consumers showed a strong preference for a subtractive framing. They had higher satisfaction and increased their spending when removing options from a starting point compared to adding them to a base product. This experiment showed that anchoring, consistent with Prelec et al. (2003) has practical implications in marketing decisions.

2.4 Willingness to pay

The price that consumers are willing to pay reflects two aspects, the value of the product to the consumer and the sacrifice involved in acquiring the product. This assessed value consists of the product’s acquisition value, which is based on the relationship between the product’s perceived benefits to the perceived sacrifice, and the transaction value (Thaler 1985), which depends on the perceived gains or losses relative to reference prices (Simonson & Drolet, 2004). When introducing product innovations reference prices are readably and easily available to the consumer. Therefore it is reasonable to presume that the possible gains and losses play a significant role in the adoption of product innovations.

One of the main advantages of using the concept of willingness to pay is its usability in new product development (Wertenbroch & Skiera, 2002). It is also a concept which has been heavily discussed, and several ways of measuring willingness to pay have been introduced, such as the Vickrey auction (Vickrey, 1961) and the Becker-deGroot-Marschak (BDM) method (Becker et al., 1964). A comparison between these measures by Wertenbroch and Skiera (2002) argues that the BDM is the most reliable measure for consumers’ willingness to pay. One of the main criticisms of directly measuring willingness to pay, for example by survey data, is that consumers might not reveal their true willingness to pay (Hoffman et al., 1993). However, several studies have produced reliable and consistent results by simply

(21)

asking consumers indicate consumers’ willingness to pay (e.g. Green et al, 1998, Kahneman

et al., 1991, Prelec etc., 2006, Hartman et al., 1991). Also, in the following study, reliability of the actual price is less of an issue as discovering consumers absolute willingness to pay is not the goal, but the relative difference in it between endowed and un-endowed owners is.

2.5 Influencing strategies

It is interesting for companies to find out how to influence the irrational part of consumers decision making. Zhang and Fishbach (2005) offer some insight into the effect of emotions on the endowment effect. They found that people feel less risks and more psychological

resources to cope with negative feelings under positive mood conditions. On other hand, they also found that people feel more risk and less psychological resources to cope with negatives under negative mood conditions. They conclude that negative or positive mood conditions can increase (negative) or decrease (positive) the endowment effect. This shows that these biases can be influenced, for example by creating a positive mode. Similarly, Lerner et al. (2004) found that different emotional states could eliminate and even reverse the endowment effect. I will be looking at other ways of influencing this effect. As can be seen the models of both Claudy et al. (2015) and Armitage and Christian (2003), behavioral intentions only partially predict the eventual behavior of consumers. This means there are other ways of influencing consumer behavior then simply adding reasons for adoption or changing a consumers attitude towards the product. After many years of research and observation, Cialdini (2007) found six main principles of influencing consumers behavior which are observable and often used in our society:

● social proof: Cialdini (2007) defines social proof as “one of the means we use to

determine what is correct, is to find out what other people think is correct” (p. 116). Consumers are influenced by the behaviors of other people. One example of this is the use of common people in video-advertisements (Cialdini, 2007) which should

convince potential buyers that people similar to them are happy users of the product. There are many examples of the power of social proof in other fields (for examples, Cialdini p. 140-156). The use of social proof in reusing towels in hotel is also evidence of the power of social proof (Goldstein et al., 2008). Persuading people to adopt

(22)

● liking: Cialdini (2007) used the following definition for liking: “the preference of

people to say yes to the request of someone we know and like” (p. 167). Although this finding in itself is not surprising, the broader application makes this influencing strategy very useful in marketing. Weapons such as physical attractiveness, similarity, compliments, contact and cooperation (Cialdini, 2007) are very useful for marketers and salesman. Nicholson et al. (2001) found that liking has a profound effect on the relationship between the sales representative and the consumer, mainly through increased trust. Similarly, Kahneman (2011) has shown solid effects of the halo bias,

in which liking the subject was more relevant to someone’s decision than the actual

message. Using the weapons mentioned above could therefore increase the trust in the quality of the product innovation.

● authority: humans tendency to show obedience to authority. Or as Milgram (1973) put it, ”the extreme willingness of people to go almost any lengths on the command of an

authority” (p. 75). In marketing such tools as titles and clothes are used (Cialdini, 2007) to show consumers the authority of a person making a certain statement. Persuading people to leave the status quo in product adoptions might be more successful if that statement is made by an authority.

● scarcity: Scarcity is defined by Cialdini (2007) by the fact that “an opportunity seems

more valuable to people when its availability is limited” (Cialdini, 2007, p. 238). The prospect of losing an opportunity is something people do not enjoy, and is consistent with prospect theory (Kahneman & Tversky, 1975). Companies can offer products for limited time, price or amount to persuade to consumers to purchase. When introducing product innovations, companies could use framing to imply a scarce offering of the product innovation. Framing has shown to have an impact on peoples judgment and decision making (Levin et al. 1998).

● consistency/commitment: defined by Cialdini (2007) as “quite simply, our nearly

obsessive desire to be (and to appear) consistent with what we have already done.” (p. 57). Companies have used this desire to their benefit by getting consumers to commit and use their desire for consistency to follow through on that commitment (Cialdini, 2007). Whether this commitment is in our best interest or not, is not as relevant to consumers as their desire to be consistent. This corresponds with the findings of Prelec

(23)

et al. (2003) and Ariely et al. (2006) on path dependency and anchoring. Framing product innovation to be consistent with consumers prior beliefs can be a useful in persuading consumers.

● reciprocity: Cialdini (2007) defined reciprocity as “a rule that says we should try to

repay, in kind, what another person has provided for us” (p. 17). One of the main advantages of this rule its strength (Cialdini, 2007). It can sway consumers to accept offers that otherwise would surely have been refused, based on their feeling of indebtness. An example of reciprocity in merchandise is the use of a free sample, where the consumers feels obligated to purchase the good after the free sample, even if it was not of satisfactory quality. Combined with the findings of Shampanier et al. (2007) on the extraordinary pull “free” has on consumers, this certainly is a powerful persuasion mechanism.

These influencing techniques should be applied carefully by trying to convince consumers of new products. Le Garrec and Torregrosa (2016) found that consumers with high motivation to process information are less likely to be socially influenced than people with low motivation to process information. There is sufficient evidence that these persuasion tactics increase effectiveness of ads and sales pitches, and these persuasion strategies are often used in marketing activities (Griskevicius et al., 2008). There are also certain conditions in which these persuaders show the strongest effects, and situations in which these effects are less strong. Griskevicius et al. (2008) and Schultz et al. (2007) offer examples of situations in which certain persuaders are more or less effective. These conditions are based on the emotions the participant is experiencing, as this makes participant more or less likely to be persuaded by a such a mental shortcut. An example being that scarcity is less effective when someone is in an emotional state of fear (Griskevicius et al., 2008).

(24)

3. Methodology

This chapter makes a start with the empirical part of the study. Firstly, based on the literature review in the previous chapter, the conceptual model and hypothesis are presented. In the second paragraph, the research design is drawn up. Following that, the data sample and

distribution are shown. Paragraph four consists of the measures and their background. Finally, a brief description of the way of data analysis will be provided.

3.1 Conceptual model and hypotheses

To answer the research question “How does consumer irrationality influence consumer’s adoption of product innovations through a reduced willingness to pay and reduced search for information? How effective are influencing strategies in reducing this endowment effect? And does the confirmation bias increase the search for information of a product innovation for consumers?”, the following aspects are of importance. Because of high failure rates in consumer adoption of product innovations, it is important to further analyse the decision making of consumers when faced with such an innovation. Consumer irrationality has shown to play a significant role in consumers decision making. Two forms of consumers irrationality are expected to play a role in the adoption of product innovations: the endowment effect and the confirmation bias. These will be tested by measuring the difference between consumers’ willingness to pay and by measuring consumers likeliness of continuing in the buying process. If biases lead to a lower adoption of product innovations (Gourville, 2006), proven influencing strategies (Cialdini, 2007) might moderate this relationship. Measuring

willingness to pay gives an insight the presence of the endowment effect (Thaler, 1990). The effect of these peripheral cues on reducing the endowment effect is missing in current literature.

Additionally, consumers continuing in the buying process will be measured be looking at their search for further information.Although the confirmation bias is often used in literature its effect on the adoption of product innovations is unclear. A consumers first impression

(positive or negative) can be used to measure whether the confirmation bias is present (Rabin & Schrag, 1999). It is expected that consumers with a more positive first impression are also more likely to continue in the buying process.The conceptual model can be found in figure 5. This gives an overview of the most important concepts that are used in this study.

(25)

Figure 5: Conceptual model

The main effect discussed in this study, is that of consumer irrationality on consumer adoption of product innovations by a difference in their willingness to pay (Thaler, 1990). It is expected that consumers who are endowed with a current product display a lower willingness to pay for a product innovation, as they involve this endowment in their decision making. People who already own a product also show a reduced interest in considering adopting new products. They end the adoption process even before it has really begun, so called passive resisitance (Talke & Heidenrich, 2014). A decreased tendency to search for information shows that this passive resistance is higher for endowed consumers than for un-endowed consumers. This leads to the following hypotheses:

H1A: The endowment effect has a negative influence on consumer adoption of product innovations through a difference in willingness to pay between endowed and un-endowed consumers.

H1B: The tendency to search for information is negatively influenced by the endowment effect. This tendency results in a decrease of the adoption of the product innovation.

(26)

The endowment effect could therefore prove a major barrier for companies when introducing product innovations. However, consumer irrationality could also work in their favor. This could be done by creating a positive first impression, as this could lead to a higher tendency to search for information. As shown by Talke and Heidenrich (2004) and Puccinelli (2008) continuing the search for information is an important phase in the buying process and with that the adoption of product innovations. Hoeffler et al. (2006) show that the starting point is an important factor in the further search for information amongst competing offerings, consistent with the findings of Nickerson et al. (1998) on the confirmation bias. This leads to the following hypothesis:

H2: A more positive first impression of a product innovation will lead to an increased

tendency to search for information. This tendency results in an increase of the adoption of the product innovation.

If consumer irrationality, specifically the endowment effect, leads to a lower adoption of product innovations, companies will have to find a way of influencing consumers decision making to reduce the endowment effect. Based on the work of Cialdini et al. (2007), there are several ways in which companies can influence consumers decision making process by using six universal principles of persuasion: social proof, scarcity, reciprocity, authority,

consistency and liking (Cialdini et al, 2007). They have shown to increase the likelihood consumers will make a purchase decision or create a favorable attitude towards a product (Schultz et al., 2007). It is unclear whether these influencing strategies reduce the endowment effect, specifically in adoption of product innovations. If these strategies are effective, a difference in willingness to pay and search for information is expected between influenced and un-influenced consumers. The consumers who are provided with the influencing strategy are expected to show a higher willingness to pay and higher search for information compared to the un-influenced consumers. This leads to the following hypotheses:

H3A: The use of influencing strategies reduce the endowment effect on consumer adoption of product innovations, through a difference in willingness to pay between influenced and un-influenced consumers.

(27)

influencing strategies. This tendency results in an increase of the adoption of the product innovations.

3.2 Research design

It is preferable to observe consumer behavior, in comparison to filling out a survey. This because consumer’s behavior could differ from their behavioral intentions or evaluations (Armitage & Christian, 2003). However, the practical limitations are too large to conduct such an experiment. Therefore a mixed method methodology case study with an experimental design was chosen. This gives a realistic picture of consumers behavioral intentions. The case study used to research the hypotheses is that of private lease of cars. Cars are often cited as high involvement products (e.g.: Lastovicka et al., 1978), which makes them highly suitable for this study, as high involvement products have shown strong endowment effects (Saqib et al., 2010). Private lease in the Netherlands is still a small but growing industry. The

Vereniging Nederlandse Autoleasemaatschappijen or “VNA” (2016) reported 36.000 private

lease cars in 2015, an increase of 19.000 cars to 2014. First However, the total market share is still only 5%, with almost 90% of drivers (non-business) still owning a car. Another study (Barten & Nouws, 2016, p. 17) shows that 72% of private lease drivers uses private lease as a replacement for owning a car themselves. The reason private lease can be seen as a product innovation is its relatively low but growing market share and the required behavioral change. The consumer does no longer own the car he is driving and is no longer required to make a onetime payment to the car company. Barten and Nouws (2016, p.32) also showed that 34% consumers would not consider a private lease offering, even if the offering would be much cheaper. Although this number is much lower than in 2015 (57%), this is still a relatively high amount.

As there is a sufficient theoretical background in the current literature, further qualitative research such as interviews, action research or ethnography (Saunders & Lewis 2012) is not necessary to investigate the research topic. This experimental design is chosen to make sure that owning a previous car truly is the differentiating factor. This study is concerned with the difference between current car owners and people who have not owned a car, car ownership is therefore the predictor variable. To find out the difference between these two groups in the outcome variable an experimental design is very suitable. Compared to a conjoint analysis,

(28)

best solution in providing this direct link. Conjoint analysis would be more suitable when interested in the effect of different variables, such as cognitive argument or affective states. It is less suitable for isolating the endowment effect, which fits more with a linear model. The study is cross-sectional as it measures the effect on consumers when they are confronted when the introduction of a product innovation.

An online survey distributed through Qualtrics was used to measure consumer’s attitude

towards the product. This was chosen to make participation as accessible as possible, which is necessary in a convenience sample such as this. It also has the additional benefit of low cost for the researcher and low effort for the participant. There was no time limit set on the survey, so respondents could search for further information on the concept of private lease if the description provided is unclear. Respondents were not be able to retake any questions to ensure reliability in the tested influencing strategies. To address the issues described by Saunders & Lewis (2012) on self-selection, some questions will be added to ensure a reasonable population and to control for certain effects. These questions concern income, gender, education level, age, kilometers driven per year, number of previous cars owned and satisfaction with current car. Questions on ownership of a car were included in the beginning of the questionnaire, and other items in the final section. Participants were also asked to give their first impression on private lease.

Participants who owned a car were instructed to think of this as a moment in which they would want to replace their current car. Participants who did not own a car were instructed to think of this as a moment in which they would purchase their fist car. Consumers were asked to fill in objective criteria on their current car as well as expectations on their new car. An introductory text before the first question was used to ensure that consumers understood what was asked of them, and that even people who were currently happy with their car could fill in the survey.

It is likely that the current monthly costs are a relevant predictor for consumers new monthly payment, as it shows consumers share of wallet towards their car. Research (Mediaxplain, 2011) showed that 85% of consumers do not know the monthly costs of their car. Participants will therefore be either given full information on the elements of monthly costs concerning

(29)

cars or no information at all. The information provided on costs will be based on a recent research by the Dutch Consumers Association (Kamp, 2016). This helps differentiating the role of current monthly payment in the study, whilst showing the effect the available information has on respondents. Participants were divided in two groups when asked about the current costs of their car. In one question, 117 participants were given no further information and simply asked to estimate the current costs. In the other question, the costs (Kamp, 2016) which apply to car owners were mentioned to 106 participants. Based on these groups a comparison was made between the current monthly payment of participants based on the level of information they received in the question.

The effect of the different influencing strategies on consumers decision making is measured by introducing different messages on a (fake) homepage of private lease website. For the control group, no additional message was added. For the six experimental groups, six

influencing messages have been designed based on the work of Cialdini (2007) and Goldstein et al. (2008). In figure 6 the message for the control group can be found, and in figure 7 the

message for the experimental group with the influencing strategy “authority”. All questions

and homepages can be found in appendix B. These messages were based on current literature on this topic, and were tested on several consumers. Further clarification on these questions and strategies used are given in paragraph four of this chapter. They were also discussed with two employees of private lease company JustLease.

Because this thesis is written in English the survey was also primarily written in this language. Since most respondents to the survey had Dutch as their first language, the questions and answers were translated into Dutch. In order to assure that the content of the items remains unchanged, a professional translator reviewed both the Dutch and English version and offered suggestions to make sure the translation was accurate and similar terms were used.

(30)

Figure 6: Homepage for control group

(31)

3.3 Data sample and distribution

The sample consists of the 8.100.640 car drivers in the Netherlands (CBS, 2016). The survey was shared on social media (Facebook, LinkedIn, Yammer) and sent by email to several participants. The survey was tested qualitatively on two participants who were not familiar with the goal of the survey. They were asked to give feedback on clarity of the questions and answers, the time needed for the survey and encouraged to offer further comments. These comments were used to improve the survey before it was distributed to respondents.

3.4 Measures

The measure of adoption of product innovation has been used in several studies such as Gourville (2006). The most reliable measure would be to use actual purchase data, however these are not usable for this study. Therefore, willingness to pay will be measured. This concept has been used many times in consumer research, such as Kim et al. (2009). It also is a common measure in measuring the endowment effect and purchase intentions (e.g. Thaler, 1985, Kahneman et al., 1991, Wertenbroch & Skiera, 2001). In this experiments, participants were asked in which price range they would be looking for a private lease car.

Next to the measure of willingness to pay, respondents were asked to indicate their likeliness to click further on the website. Likeliness to click further was introduced as a measure to see whether consumers continue in the their buying process by this increased search for

information. To click further would mean consumers have recognized their need for a new car and see this website as a possibility to search for information as they next step in the buying process (Puccinelli et al., 2009). These two measures are sufficient to indicate the presence of the endowment effect between first time buyers and endowed car owners. The endowment

effect is defined as “the fact that people often demand much more to give up an object than

they would be willing to pay to acquire it” (Kahneman et al., 1991). This means that current car owners, as they have something to give up, should demand much more from a product innovation such as private lease, than people who are un-endowed.

The influencing strategies were mostly based on the work of Cialdini (2007). In addition to that, the strategy including the free test ride was based on the work of Shampanier et al.

(32)

by Lee et al. (2006) was used as the idea for the different questions on current monthly costs. Several questions on the product and current car ownership will be asked to participants based on the research of Barten and Nouws (2016), as these are necessary control variables. Time spent researching the purchase of a new car is measured in hours. Being a high involvement product, cars are expected to produce a high research time. As argued by Saqib et al. (2005), this can have an effect on the arguments consumers are influenced by as well as the presence of the endowment effect (Saqib et al. 2010). Respondents first impression of private lease was asked to be to differentiate for further research on the confirmation bias. Several researches, such as Rabin and Schrag (1999) have shown the importance of first impression in decision making. Because buying a car is not a daily activity and the price of private lease consists of several components which are not transparent for consumers, answer possibilities will be offered on a 5-point Likert scale instead of open questions. A 7-point Likert scale was used for two categories (income, current monthly costs) where further differentiation in categories was necessary based on literature (Barten and Nouws, 2016).

3.5 Data analysis

The data were analyzed with Statistical Processing software SPSS. To measure the difference in willingness to pay groups were coded. The experimental group with the influencing

strategy was numbered as group 2 and the control group as group 1. This meant the effect of the influencing message could be isolated by transforming the scores from the control group and experimental into a new variable. The same method was used in measuring likeliness to click further on the website.

After recoding the reverse coded items, scale reliabilities, descriptive statistics, skewness, kurtosis and normality tests were computed. From the total of 28 variables, none were normally distributed, based on the Kolmogorov-Smirnov test (p < .01). This does not

immediately produce a problem, as this has no impact on the usefulness of the data due to the central limit theorem (Field, 2010, p. 169). However, to make sure the data used are not biased, bootstrapping, a non-parametric re-sampling procedure, was applied using 1000 bootstrap samples. The bias corrected and accelerated confidence interval method was used to calculate the confidence intervals. The use of bootstrap and choosing of adequate bootstrap sample and method follows recommendations by Field (2013, p. 199). To check the data for

(33)

homoscedasticity, Levene’s test was performed with every analysis. This showed no significant results.

(34)

4. Results

There were 316 completed responses with an additional 64 partial responses (response rate 83%). The partial responses did not contain enough useful information and were not used in further analysis. Of the 316 recorded responses, 274 respondents filled in all the questions they were asked on the survey. All responses (n=32) which failed to answer questions

concerning their current car or in their willingness to pay or likeliness to click were excluded listwise. Twelve responses failed to answer questions in the block about general questions and cognitive questions (seven failed to answer one question, four failed to answer two). Missing responses were substituted with the median. 286 responses remained after this procedure and were used for the analysis. An overview of most important frequencies of characteristics of the respondents are presented in table 1.

Table 1: Frequencies

Several participants who did not own a car also responded to the questions with the

influencing strategies, to make a separate comparison. These numbers were not included in the final analyses of these variables, as this was meant to see the difference between endowed consumers and un-endowed. The sample sizes (Social proof N = 14 , Scarcity N = 7,

Authority N = 8, Reciprocity N = 5, Consistency N = 5, Liking N =6) for the un-endowed consumers within these conditions were too small to compare them with the endowed.

Gender Frequency Percent Level of education Frequency Percent

Male 127 44,41% VMBO 4 1,40%

Female ₁₅₉ _55,56% HAVO ₉ _3,15%

VWO 3 1,05%

Year of birth Frequency Percent MBO 26 9,09%

1941-1950 4 1,40% HBO 135 47,20%

1951-1960 20 6,99% WO 109 38,11%

1961-1970 55 19,23%

1971-1980 77 26,92% Net income Frequency Percent

1981-1990 115 40,21% Less than €1000 8 2,80% 1991-2000 15 5,24% Between €1000 and €1999 43 15,03% Between €2000 and €2999 133 46,50% Between €3000 and €3999 55 19,23% Between €4000 and €4999 26 9,09% Between €5000 and €5999 ₁₀ _3,50% €6000 or more 11 3,85%

(35)

First the correlation matrix, which can be found in table 3 in appendix A, will be discussed. Subsequently the results from the regression analyses will be outlined. Direct relationships between the different variables are presented in advance to the researched mediation effect. As mentioned in the previous chapter, bootstrapping was performed because of non-normality of the data. The results of bootstrapping are presented in the tables, and will be included in text when significantly different to the regression results.

A first observation which can be drawn from table 3 is that car ownership significantly correlates with both likeliness to click further as well as willingness to pay for the control group. Both correlations are negative, implying that non-owners of a car are more likely to click further and have a higher willingness to pay. As expected, both monthly costs and income also positively correlates with willingness to pay. These variables will be used as control variables in further analysis, along with gender and age. The level of information provided correlates significantly with monthly costs, further analysis will show whether this has an indirect effect on willingness to pay through these monthly costs. Likeliness to click and first impression also show a statistically significant correlation.

4.1 Relationship monthly costs and willingness to pay

Based on the correlation matrix, current monthly costs showed a statistically significant relationship with willingness to pay. Respondents were provided with a different level of information on monthly costs. A multiple hierarchical regression was performed to investigate whether this level of information would influence the amount of monthly costs respondents would indicate. The results are presented in table 3, the different groups are presented in figure 8. In the first step, four predictors were entered: gender, age, income and education. This model was statistically significant (F (4,218) = 3.819; p < .01) and explained 6.5% of variance in current monthly costs. After introducing the level of information at Step 2 the total variance explained by the model as a whole increased to 15.7% (F (5, 217) = 23.577; p < .01). The introduction of the level of information explained 9.2% additional variance in current monthly costs after controlling for gender, age, income and education (R2 Change = .092; F (1, 217) = 8.087; p < .01). In the final model two of the five predictor variables were

(36)

statistically significant, with the level of information recording a higher Beta value (β = .308,

p < .01) than age (β = -.142, p < .05).

Figure 8: Indication of current monthly costs based on level of information provided

Based on the results in table 3 it was interesting to test whether the level of information moderated the relationship between current monthly costs and willingness to pay. The first model of the process macro bootstrap method of Hayes (2013) was used to test this

moderation effect. The model was significant, explaining 26.0% of variance in willingness to pay. The predictor variable current monthly costs was statistically significant (B = 0,432, t = 7,011, p < .01). This shows that the level of information had a statistically significant effect on monthly costs, and monthly costs has a statistically significant effect on willingness to pay. However, there is no evidence for a direct effect or moderating role of level of information on willingness to pay. The results indicate that including the level of information to the model does not lead to significant interaction effects (B = -0,073, t = -,760, p > .10).

4.2 Hypothesis 1: Endowment effect.

Hierarchical multiple regression was performed to investigate the ability of car ownership to predict the willingness to pay, the results of which are presented in table 5, 6 and 7. The

(37)

difference in means is presented in figure 9. In the first step of hierarchical multiple regression, three predictors were entered: gender, income level and age. This model was statistically significant F (3, 95) = 14.177; p < .01 and explained 30.9 % of variance in willingness to pay. After introducing car ownership at Step 2 the total variance explained by the model as a whole was 34% F (4, 94) = 12.099; p < .05. The introduction of car ownership explained an additional 3.1% of variance in willingness to pay, after controlling for gender, age and income level (R2 Change = .031; F (1, 94) = 4.360; p < .05). In the final model three out of four predictor variables were statistically significant, with net monthly income

recording a higher Beta value (β = .380, p < .01) than car ownership (β = .183, p < .05) and

gender (β = -.263, p < .01). In other words, if a person’s car ownership increases for one (goes

from not owning a car, to owning a car), their willingness to pay will increase with 0.183. As expected, income plays a big role in the amount someone is willing to pay. An increase in a

person’s income with one, and his or her willingness to pay increases with 0.380. On the other

hand, on average women obtain 0.263 points lower score in willingness to pay than men.

Figure 9: Willingness to pay for private lease for car owners and non-owners

If endowed consumers would not be interested in the concept of private lease because they owned a car, it is less likely they would continue the purchase process on a private lease

(38)

ownership to predict the likeliness consumers would click further on the website on private lease, after controlling for age, gender and research time. In the first step of hierarchical multiple regression, three predictors were entered: gender, age and research time. This model was statistically significant F (3, 95) = 5.275; p < .01 and explained 14.2 % of variance in likeliness to click further. After introducing car ownership at Step 2 the total variance

explained by the model as a whole was 26.1% F (4, 94) = 8.280; p < .01. The introduction of car ownership explained an additional 11.8% variance in likeliness to click further, after controlling for gender, age and research time (R2 Change = .118; F (1, 94) = 15.020; p < .01). In the final model out of four predictor variables research time and car ownership were

statistically significant. Research time recorded a higher Beta value (β = .432, p < .05) than

car ownership (β = -.377, p < .01). In other words, if a person’s car ownership increases (goes

from not owning a car, to owning a car) with one, their likeliness to click will decrease with 0.377.

Finally, I expected that consumers first impression of this production innovation would be more positive if they did not previously own a car. A multiple hierarchical regression was performed to measure whether car ownership predicted a difference in first impression. In the first step of hierarchical multiple regression, two predictors were entered: gender and age. This model was not statistically significant F (2, 284) = .292; p > .1 and explained 0.2 % of variance in first impression of private lease. After introducing car ownership at Step 2 the total variance explained by the model as a whole was 5% F (3, 283) = 4.972; p < .01. The introduction of car ownership explained an additional 4.8% of variance in first impression of private lease, after controlling for gender and age (R2 Change = .048; F (1, 283) = 14.306; p < .01). In the final model out of four predictor variables only car ownership was statistically significant, recording a negative Beta value (β = -.226, p < .01). In other words, if a person’s car ownership increases (goes from not owning a car, to owning a car) for one, their first impression of private lease will decrease with 0.226. Based on these results, hypothesis 1A and 1B are supported.

4.3 Hypothesis 2: Confirmation bias

To test this hypothesis, it is necessary to measure whether current car satisfaction has a

(39)

performed in order to test this relationship, results can be found in table 8. In the first step, two predictors were entered: gender and age. This model was not statistically significant (F (2, 220) = .622; p > .1) and explained 0.6% of variance in first impression of private lease. After introducing current car satisfaction at Step 2 the total variance explained by the model as a whole remained 0.6% (F (3, 219) = .460; p > .1). The introduction of current car

satisfaction explained 0,1% additional variance in first impression of private lease, after controlling for gender and age (R2 Change = .001; F (1, 219) = .141; p > .10). In the final model none of the three predictor variables were statistically significant, with current car satisfaction recording a non significant Beta value (β =-.025, p > .10).

To measure whether the confirmation bias occurred and positively influenced a consumer likeliness to search for further information through his first impression a multiple hierarchical regression was performed. The full results are presented in table 9. In the first step, four predictors were entered: gender, age, research time and car ownership. This model was statistically significant (F (4, 282) = 5.467; p < .01) and explained 7.2% of variance in

likeliness to click further. After introducing first impression of private lease at Step 2 the total variance explained by the model as a whole increased to 27.0% (F (5, 281) = 75.643; p < .01). The introduction of first impression of private lease explained 19.7% additional variance in likeliness to click further, after controlling for gender, age, research time and car ownership (R2 Change = .197; F (1, 281) = 20.664; p < .01). In the final model three of the five predictor variables were statistically significant, with first impression of private lease recording a higher

Beta value (β = .456, p < .01) than research time (β = .149, p < .01) and car ownership (β =

-.132, p < .05). Bootstrap results showed similar outcomes, however car ownership showed an increased statistical significance (B = -.404, p < .01). To find out whether car ownership moderated the effect of first impression on likeliness to click, the first model of the process macro bootstrap method of Hayes (2013) was used. This model showed no significant

interaction effects of car ownership on the relationship between first impression and likeliness to click (B = 0.943, t = -,538, p > .10). Based on the results mentioned in this paragraph, hypothesis 2 is supported.

(40)

To establish whether the influencing strategies have a significant effect in reducing the endowment effect for private lease, a hierarchical multiple regression was performed. The ability of the influencing strategy to predict the willingness to pay, after controlling for age, gender, current monthly payment and net income was measured. To measure likeliness to click, different control variables were used: age, gender and research time. On overview of the difference in willingness to pay and likeliness to click between experimental conditions is shown in figure 10 and 11. Results of the regression analysis for the different groups are shown in table 2. The detailed results of the analysis can be found in table 10 to table 21 in Appendix A.

Irrationally predictable? : an insight in the irrational behavior of consumers when rejecting product innovations