The effect of analogies on consumers’ willingness-to-pay
for innovative products
Bachelor’s Thesis
University of Amsterdam
Vincent Sebastiaan Cieraad
10108874
Amsterdam, June 30
th, 2014
Thesis seminar Business studies Supervisor: Bram Kuijken Academic Year: 2013-2014 Semester 2, Block 3
Abstract
Successful implementation of really new products can lead to a long-term financial success, but the market success rate of these problems is alarmingly low. Buying really new products causes a high amount of perceived uncertainty for a customer due to lack of knowledge and product experience. This negatively affects the consumers’ attitude towards the product and the consumers’ willingness-to-pay. In order to reduce this perceived uncertainty and increase willingness-to-pay, producers need to educate their customers. A promising consumer-learning tool seems to be analogies. Therefor this research sets out to find out how analogies influence willingness-to-pay. The results indicate that analogies can be used as an efficient consumer-learning tool in order to reduce perceived uncertainty and increase willingness-to-pay. In addition, the research shows that analogies also have the function of a reference price. The analogy influenced the value people would attach to the product due to its referencing aspect. Therefor the use of analogies that are valued high by consumers increases the inferred value of the innovative product and raises the consumers’ willingness-to-pay.
Table of content
Abstract ... 2
Foreword ... 4
1. Introduction ... 5
2. Literature Review ... 7
2.1 Innovation and innovation resistance ... 7
2.2 Consumer uncertainty ... 7
2.3 Willingness to pay ... 8
2.4 Analogy and structure-mapping ... 9
2.5 Diffusion of innovation ... 11
3. Theoretical Framework ... 13
3.1 The influences of analogies on the consumers’ willingness-to-pay ... 13
3.2 The effect of analogies on different consumer segments ... 14
4. Methodology ... 16
4.1 Research design ... 16
4.2 Method ... 17
4.3 Sample ... 18
4.4 Variables and measurements ... 19
4.4.1 Willingness-to-pay ... 20
4.4.2 Early adaptors ... 20
5. Results ... 21
5.1 Descriptive Statistics ... 21
5.1.1 Auction bid data ... 21
5.1.2 Early Adaptors ... 24
5.2 Hypothesis ... 26
5.2.1 Auction bid Data ... 26
5.2.2 Early Adaptors ... 29
6. Discussion ... 30
6.1 General Discussion ... 31
6.2 Theoretical Implications ... 33
6.3 Managerial Implications ... 33
6.4 Implications and Future Research ... 34
7. Conclusion ... 36
References ... 37
Appendix A ... 40
Appendix B ... 44
Foreword
This thesis was written for my Bachelor’s degree in Business Studies at the University of Amsterdam. I would like to thank my thesis supervisor Bram Kuijken for his guidance and support during the writing of this research. His comments and helpful insights have helped me a great deal in improving the quality of my paper. I would also like to thank my friends and family
for their help and support. I hope that the reader will enjoy reading my thesis. - Vincent Sebastiaan Cieraad -
1. Introduction
Today’s business environment is more rapidly changing than ever before. Because of this rapidly changing and open environment it’s getting harder and harder for companies to penetrate the market with new products and build and sustain a competitive advantage (Baer & Frese, 2003). While new product development is one of the most important activities of a company, only very few Really New Products (RNP’s) become a success. In fact, between 40 per cent to 90 per cent of RNP fail (Cierpicki, Wright, & Sharp, 2000; Breu, Guggenbichler & Wollmann, 2008). The reason that a lot of products fail to succeed is that buying a RNP from a consumer’s point of view is associated with a high degree of uncertainty (Mitchell, 1999; Sheth 1981; Taylor, 2014). Since the outcome of a choice can only be known in the future a feeling of uncertainty arises for the customer. This subjective experience causes the consumer to feel less in control of the situation. Due to incomplete information, due to lack of information or product experience, the consumer will tend to value the product less (Mitchell, 2008).
Providing the consumer with information has been proven to significantly decrease uncertainty, and therefore perceived risk (Jacoby, Jaccard, Curim, Kuss, Ansari, & Troutman, 1994). A way of providing information and educating the consumer is by using analogies. An analogy is a comparison between one thing and another, typically for the purpose of explanation or clarification. Analogies help the consumer categorize the product and will help them develop a clearer understanding of potential benefits (Gentner, 1983; Chan, Paletz, & Schunn, 2012). Analogies have proven to improve product evaluation and product comprehension, and increase product preferences. This makes analogies a highly suitable mechanism to communicate a
product’s features, informing the customer, reducing perceived risk, and reduce resistance against innovation (Clement & Gentner, 1991).
Research so far on analogies as a consumer-learning tool has shown promising results (Gentner, 1983; Chan et al., 2012). However, within the existing literature there is a lack of strong evidence. Most studies have been conducted in laboratory studies or are hypothetical. This study will use a real life second price sealed bid auction, which will extend the validity of the results and hopefully find proof of the usefulness of analogy. Since perceived uncertainty leads to systematic undervaluation, which in turn leads to a decrease in consumer willingness-to-pay, reducing perceived uncertainty with the use of analogies could be a valuable tool for managers. One important thing that this research will take into consideration is how the use of different
analogies will influence the consumers’ perception about the product. If the producer uses an analogy that is valued lower by the consumer than is the value of the product, the consumer might associate the product with a lower value and visa versa. In this paper we try to identify what the effect is of using analogies as a consumer-learning tool on consumers’ willingness-to-pay, and how different value analogies influence the consumers’ valuation of the products.
Bauer was the first person to make the link between uncertainty and risk in 1960. The perceived risk caused by uncertainty influences the consumer’s innovation decision-making process (Rogers, 1983), and increases the resistance against adapting innovation. A large feeling of perceived uncertainty will lead to a large resistance against innovation (Sheth, 2009).
Uncertainty also tends to lead to undervaluation of products, and therefor decreases the
consumer’s willingness to pay for the product (Wang et al., 2008). The high levels of uncertainty that arise with RNP have been linked to high failure rate (Taylor, 2014). The lack of product experience and product information causes the consumer to systematically undervalue the
product (Schmidt & Calantone, 2003), therefore withholding them from adapting the product and reducing their wiliness-to-pay in general. Decreasing uncertainty could therefor significantly increase the rate of success of RNP (Sheth, 1981; Taylor, 2014; Calantone, 2003). This is why analogies could be an important tool for managers to increase the success rate of their RNP’s. Since R&D is one of the companies’ biggest expenditures this could lead to a dramatic decrease in costs for companies and increase their revenue and long-term success (Gielens, 2008).
In the following section the existing literature will be reviewed. Subsequently a theoretical framework is given in which hypothesis will be introduced that later on will help answer the research question. Next, the methodology used to conduct this research is explained. Next, the results of the data and the analysis will be outlined and discussed. Based on the hypothesis the research question will be discussed, limitations of the research will be given and suggestions for future studies. Finally a conclusion is given.
2. Literature Review
2.1 Innovation and innovation resistance
Innovation imposes change on the consumer, and a normal consumer response to this is
resistance. Since not all innovation is by definition desired or useful. Innovation in this research is defined from a marketing perspective as an innovation is a product that is perceived by the consumer as new. This change can be based on just a single aspect such as a new shape of a product, but this can also entail the development of a completely new product.
There are two main reasons that innovation fails: 1. The producer fails to communicate the innovation of the product, therefor failing to be perceived as new by the consumer and 2. The product is perceived as a Really New Product by the individual, meaning that the consumer cannot relate to the product and fails to see the usefulness of it (Cierpicki, Wright & Sharp, 2000; Schmidt & Calantone, 1998). Not seeing the usefulness of the product will result in resistance against innovation; the individual is not interested in the product because it fails to make a cost/benefit analysis (Cierpicki, Wright & Sharp, 2009). That’s why one of the most important characteristics for a successful innovation is its amenability to modification (Robinson, 2009). If the reason the individual is resisting the innovation is because of lack of compatibility, the producer should modify its product. But if the reason is that the individual doesn’t have the proper knowledge to make inferences about the product, the manufacturer should modify the way it’s informing the consumer. It is this lack of information that causes perceived uncertainty or the individual (Shiu & Walsh, 2011). From this section we can conclude that the lack of information causes a feeling of uncertainty, which creates a resistance against innovation from the consumer’s side.
2.2 Consumer uncertainty
Uncertainty is recognized as an important concept within consumer behaviour. Consumer uncertainty can be defined as a condition in which the information available deviates from the consumer’s ideal information state (Castano, Sujan, Kacker, & Sujan, 2008). This is usually the case with RNP’s, since individuals fail to have any knowledge or prior product experience with the product. This means that RNP’s produce a high degree of uncertainty in the consumers’ eyes. Since the outcome of a choice can only be known in the future, this means that the consumer has
to deal with a high level of uncertainty (Taylor, 2014). A consumer purchasing intentions are linked to his perceived uncertainty level. If the individual is unable to draw inferences about the product he will be unable to make a cost/benefit analysis, and therefor undervalue the innovative product (Taylor, 2014).
Consumers with a low level of knowledge uncertainty are more likely to be confident in their ability to make an informed purchase decision; therefor this will increase their valuation of the product and the purchase decision. Conversely, consumers with high levels of knowledge uncertainty will feel less confident about their ability to make an informed purchasing decision, leading to a lower valuation of the product and lower purchasing intentions. If an individual feels high levels of perceived uncertainty, then he will tend to search for more information (Castano et al., 2008). By acquiring and processing this new information the individual will be able to make better inferences about the new product. This will help reduce the perceived risk of the
individual, because he will be able to make a better cost/benefit analysis about a possible purchasing. In other words acquiring more knowledge reduces the uncertainty of a purchase, because better inferences about beneficial outcomes can be made (Castano et al., 2008).
2.3 Willingness to pay
Individuals have a maximum price that they are willing to pay for a product. This maximum price that assures the purchase of the product is called a consumer’s willingness-to-pay. Willingness-to-pay is in this research defined as a consumer's maximum price at which he/she will buy a unit of the given product. In marketing a person's maximum willingness-to-pay is also referred to as (floor) reservation price (Mankiw 2012; Varian, 1992; Wang et al., 2008). This means the maximum price a consumer is willing to pay with a 100 per cent purchasing possibility. This economic concept is widely invoked and well-known (Varian,1992; Wang 2007). From this it follows that when a product or service is valued high by the consumer the maximum willingness-to-pay is higher than for a product that is valuated lower.
The consumer most often undervalues really new products. This results from a lack of knowledge and product experience with the RNP, which in turn gives the consumer a sense of uncertainty (Goel, 1997; Schmidt & Calantone, 1998). Uncertainty has been shown to have a negative influence on the consumers' willingness-to-pay (Mankiw, 2012; Varian, 1992; Taylor
2014). A consumer's valuation is based on a cost/benefit analysis of the product. The lack of information and product experience will reduce the consumer's capabilities to predict benefits from buying the product (Taylor, 2014; Gregan, 2002). Accordingly, for a high level of uncertainty a lower floor reservation corresponds than when low levels of uncertainty are perceived. This implies that by lowering the consumer's level of uncertainty the willingness-to-pay should increase (Mankiw, 2012).
2.4 Analogy and structure-mapping
In the field of cognitive psychology, analogy is described as a form of reasoning in which one thing is inferred to be similar to another thing in a certain respect, on the basis of known similarities in other respects. With respect to consumer-learning, this mean that analogical learning is transferring knowledge from a familiar domain (the base) to a novel domain (the target), in order to better conceptualize innovative products or ideas. It is this sharing of a system of interconnected relations that makes it possible for a consumer to use knowledge and/or
experience of a familiar domain in order to gain a better understanding of the novel and unfamiliar situation (Gentner, 1983; Gentner & Markman, 1997; Goel, 1997; Gregor-Paxton, Hibbard, Brunel, & Azal, 2002; Sujan, 1985; Vosniadou & Ortony, 1989).
Therefor in this paper analogy will be defined as follows: reasoning by analogy entails the identification of structural similarities between disparate domains and the transfer of additional information from the familiar base domain to the novel target domain in order to gain a better understanding of the novel and unfamiliar situation.
The most widely accepted theory for analogical learning is Gentner’s structure mapping
(Gentner, 1983, 1985; Vosniadou, 1989). Even though this model of structure mapping is widely accepted by researchers in the field, the number of suggested stages that exist differs among them. For example Gregor-Paxton et al. (2002) propose a three stages structure: (a) access, (b) mapping, and (c) inference. Holyoak (1985), although considered faulty by many, proposes a goal-drive pragmatic framework wherein he tries to replace structure with relevance.
Even though there are differing views on the stages of which the structure mapping consists, the most support can be found for the 4-stage structure mapping. Successful transfer of knowledge
requires (a) access: the retrieval of a potentially useful source, (b) mapping: finding a mapping, or set of appropriate correspondences between the elements of the base and target, (c) transfer: using the mapping, together with knowledge of the source, to construct inferences about the target, and (d) schema induction: evaluating and possibly adapting the inferences in light of what is actually known about the target (Holyoak & Spellman, 1992; Gerntner, 1983; Vosniadou & Ortony, 1989).
During the first stage (access), a base domain becomes active because information is retrieved from the long-term memory. The memory in the base domain is considered a potential source of relevant knowledge in order to create a better understanding of the novel domain (target domain). The more attributes the target domain seems to have in common with the base domain, the higher the amount of information that is used to make inferences about the novel domain.
During the mapping stage the accessed base domain is aligned with the novel target domain. In this stage the focus is at the idea that if one relational aspect of an object is the same, more inferences can be transferred from the base to the target. Eventually the main goal of the mapping stage is to uncover and the transferring of commonalities in the way the object in the domain relate to each other.
During the transfer stage individuals use the knowledge they’ve retrieved during the mapping stage from the base domain. The individuals process the information and use this information to create interferences about the target domain, based on a shared causally coherent system. Since the target and base domain share a set of shared characteristics, the individual can use the knowledge of the base domain in order to create new inferences about the target domain (Clement & Gentner, 1991; Gentner & Markman, 1997; Gentner, 1989).
During the final induction stage the newly acquired information is used to create a new schema. This new schema consists of the original mental representation of the target, altered and adjusted by the newly created knowledge from the transferring stage.
There is still a gap in the literature in how analogies exactly influence the consumers’ valuation of really new products. That’s why in this research we try to see how consumers’ willingness-to-pay changes with the use of different analogies. By doing so, we hope to find out how different analogies have different effects on consumers’ perception of a product.
From a marketing point of view analogical learning has recently become more interesting. Producers have found it difficult to educate the people about really new and innovative products. RNP’s are products that are so innovative that they defy existing product categories and that they create an entire new category. This means that due to the novel characteristics of the products they cannot be placed into existing categories. This makes traditional learning by categorization impossible. Since analogical learning works by accessing knowledge from a familiar, yet
disparate domain, it enables the individual to draw inferences about a RNP. Even though the RNP is categorized in a different category, the system of common characteristics makes it possible for the individual to draw inferences. Since this creation of new knowledge reduces the amount of uncertainty perceived by the individual it will increase it’s positive affection towards the product (Gregor-Paxton et al., 2002).
2.5 Diffusion of innovation
According to Roger’s Diffusion of Innovation theory there are different kinds of consumers with different kind of interest in innovative products. His Diffusion of Innovation theory tries to recognize differences between consumer segments and tries to identify different aspects of innovative products that appeal to certain segments. The Diffusion of Innovations offers three insights into the process of innovation diffusion: 1. Which product qualities make an innovation spread, 2. The importance of peer-peer conversation and peer networks, and 3. Understanding the different needs of segments. From a marketing perspective the Diffusion of Innovation is very important, because it gives insights in how and why innovations are accepted or rejected by consumer segments.
Importance of Peer-to-Peer
Advertising and media stories are good marketing methods to spread information about new innovations. Even though these channels are good for spreading information, it is peer-to-peer communication that really spreads the innovation. The success of an innovation not only depends on the first adoption, but also on how well it manages to meet the needs of more and more
demanding and risk-averse consumer segments. The process of keep evolving in order to target different segments is called Reinvention.
The early adoption of RNP’s involves the management of high to moderate levels of risk and uncertainty. The early adaptors of the innovation will therefor be people who can judge this uncertainty better than others. These so-called early adaptors are in general financially better supported, perceive less risk and are personally more confident in making decisions involving high to moderate levels of uncertainty and risk.
The five different consumer segments
Rogers (1983) differentiates between five different segments of consumers in his diffusion theory. It is important from an economic and marketing point of view to differentiate between these segments, because each of them play a different role in the adaption and diffusion of innovation. The five different segments are; Innovators, Early Adaptors, Early Majority, Late Majority, and Laggards. The first two segments consist of the Innovators, the people who come up with the innovations, and the early-adaptors, the first people to adopt these innovations. These first groups of people play one of the biggest roles in innovation adaption. Early adaptors tend to be more economically successful, well connected and informed and also more socially accepted. Early Adaptors deal with uncertainty and risk in a different way and are more likely to purchase innovative products if the benefits seem apparent. If the expectations of the new product are met then the early adaptor is very likely to spread the word to others. Since early adaptors tend to be an important hub of information within their social circle the opinion of the early adaptor is valued highly by many others. The other segments will be referred to from now on as followers. These people tend to need more information, hands on experience and opinions from early adaptors before purchasing innovative products (Robinson, 2009; Rogers, 1983). That’s why it is important for an innovative to succeed that they get adopted by the early adaptors and that they have positive feelings towards the product, so that positive worth-of-mouth can spread
3. Theoretical Framework
3.1 The influences of analogies on the consumers’ willingness-to-pay
Previous studies have shown that the use of analogies can be used to increase an individual’s knowledge and product evaluation of a RNP. Furthermore it has been shown that the use of analogies can be used as an effective tool to communicate the distinct benefits of a Really New Product to individuals. By using analogies in a correct way the consumers’ lack of knowledge about a product can be reduced. This increase in knowledge will change the consumer's perception toward the product due to a reduction in perceived uncertainty. The reduced
uncertainty will make the consumer feel more comfortable about the situation since the perceived value of the product becomes clearer. This increase in consumer product valuation should be translated into a higher willingness-to-pay, since willingness-to-pay is the monetary
representation of a consumer's floor reservation price (Wang, Venkatesh & Chatterjee, 2008). Taking these findings into account, it is hypothesized in this paper that the use of an analogy will increase an individuals willingness-to-pay for an innovative product. Also it is argued that the use of different analogies will influence the individuals willingness-to-pay. Since different analogies call upon information stored in different base domains of an individual, and these base domains differ in valuation, it is hypothesized that the use of an analogy with a higher monetary valuation will result in a higher willingness-to-pay. More formally, it is hypothesized that:
H1a. The use of a textual and visual analogy within the product description will increase the consumers' willingness-to-pay
H1b. The use of a textual and visual analogy in which an object is used that has a lower value than the product will lead to an increase in consumers' willingness-to-pay compared to a treatment without analogies, even though the perceived value of the object in the analogy is lower than the product
H1c. The use of an textual and visual analogy in which an object is used that has a higher value than the product will lead to an increase in consumer’s willingness-to-ay compared to a
H1d. The use of a textual and visual analogy in which an object is used that has a higher value than the product will lead to an increase in consumer’s willingness-to-pay compared to a treatment with an object with a lower value than the actual product.
3.2 The effect of analogies on different consumer segments
Roger (1983) has show with his model of diffusion theory that different consumer segments exist and that those different segments behave differently towards innovation. This would suggest that different segments react differently on the treatments used in this experiment. The early adaptor for example is characterized by his different attitude towards innovation. An early adaptor perceives less risk when little or no product information and/or product experience is available (Robinson, 2009). Since when dealing with really new products no product experience is available and knowledge is limited, the purchasing decision of these products are perceived as highly uncertain (Robinson, 2009; Roger, 1983). The use of analogies is expected to help educate the consumer about the new product. Since early adaptors due to their nature are more open for innovation, it is expected that the use of analogies would have less of an influence on their
willingness-to-pay than on followers. Since followers are more resistant against innovation due to lack of knowledge, the use of analogies will increase their willingness-to-pay because of the increase in information about the product. To summarize, it is expected that an early adaptor perceives less risk and therefor their willingness-to-pay will be significantly higher than for followers (Robinson, 2009; Roger, 1983). Therefor the author hypothesizes:
H2a. Early adaptors have a higher willingness-to-pay for innovative products than followers
Another characteristic of an early adaptor is that he is considered to be economically more successful than other consumers (Robinson, 2009). Due to his higher economical freedom an early adaptor is less restricted in purchasing decisions (Robinson, 2009). If an early adaptor buys a product that turns out to be not as beneficial as he expected it will most likely affect him less, since the monetary loss is lower compared to an individual with less economic freedom. In this paper it is therefore expected that the larger economic freedom of the early adaptor makes him
more likely to buy products with a relatively high-perceived uncertainty. It is expected that how higher the degree of being an early adaptor is, the higher the positive purchasing intentions towards the product are. Early adaptors are expected to value innovative product more in general and show this by a higher willingness-to-pay than consumers who show a low degree of being an early innovator.
H2b. The degree of being an early adaptor negatively moderates the effect of an analogy on willingness-to-pay
4. Methodology
This study tries to describe how the use of analogical learning influences the consumer’s willingness-to-pay for an innovative product. Furthermore it seeks to explain how the use of different kind of analogies influences the consumer’s value placing of the innovative product. It will also try to identify if certain analogical learning situations are more appealing to certain kind of consumers, in this case innovators/early adapters in particular. Meeting both objectives this will help answer the research questions: ‘What is the effect of analogy as a tool on consumers’ willingness-to-pay, and how does the valuation of the analogy affect the willingness-to-pay? ‘.
4.1 Research design
The research method chosen for this research is an experimental research setting. This research method is the most appropriate research setting, since we can alter independent variables in different treatments in order to see the effect on the dependent variable consumers’ willingness-to-pay.
The first part of the experiment will consist of the manipulation of the independent
variable, analogy, and measure its effect on the dependent variable willingness-to-pay. The use of an experimental research setting not only allows for a highly controlled context, but using a real auction will also help us to come close to a real-life situation, making the validity of the results higher. Since only one variable will be altered in the first part of the experiment this means that any deviations between the groups must be caused by the alteration of the analogy (considering a normal standard deviation), since all other factors will be held constant.
The second part of the experiment will consist of a post-test questionnaire. This short questionnaire will try to identify which type of consumer, according to the diffusion of innovation model, the participant is. Using this information it will be possible to see if certain analogical learning models are valued higher by certain groups of consumers. Since this questionnaire will be done after the initial experiment there is not chance of influencing the individuals initial decision making on the willingness-to-pay for the innovative product beforehand.
4.2 Method
The experiment will be conducted with the use of a second price sealed-bid Vickrey auction (Vickrey, 1961). This experiment will be carried out using the platform www.veylinx.com. This platform enables consumers to place a sealed bid on a product. The Vickrey auction method entails that the highest bidder wins the auction, but only has to pay the second-highest bid. Even though evidence has been found that a Vickrey auction may still lead to overbidding due to competitive bidding (Kagel, Harstadt, & Levin, 1987), Vickrey (1961) has showed with his auction method that the chances of truthful bidding are greater than that of under-/overbidding. By using this method we expect to yield the best-unbiased results.
During the experiment a RNP will be auctioned: Sugru. The individuals will be divided into different groups. Each group will see a different advertisement for the product, each with a different manipulation of the independent variable. One group will see the product without any analogy, while the other groups will see the product with an analogy. The analogy that will be included will differ per group. The analogies that will be used will all try to trigger characteristics of repair. By altering the analogy that is displayed we try to manipulate the base domain that is targeted by the participant when trying to make inferences about the really new product. By using different analogies the expectation is that different base domains will result in different inferences and valuations about the target domain. Meaning that if a certain analogy is linked to a high in value base domain, this will result in a high value inference of the target product, leading to a higher willingness-to-pay.
The first treatment is a situation in which only the product is featured together with a short list of its features. This will be the control situation. The second treatment consist of the same elements as the first treatment, but in addition it has both a textual and visual analogy. Added is the text: “This product restores things again. The way sowing back on a button restores a shirt.” and an imagine of a button being sown back on the shirt. The third treatment consists of the same elements as the first treatment with as addition the text: “This product restores things
again. Like changing a cracked screen restores a smartphone”. See Appendix B for the
advertisements. These three different treatments are chosen, because we believe that these different situations will yield different results in willingness-to-pay. In the first situation there is no analogy given, therefor the uncertainty about the product is high due to the lack of knowledge. This should decrease the consumer’s willingness-to-pay. The second treatment included the
analogy of a button being changed. This analogy gives the consumer more information about the product, therefor it should reduce the uncertainty and increase willingness-to-pay. The third treatment also includes an analogy, but the replacement of a screen is more expensive and higher valued than the replacement of a button. Therefor it is expected that in the third situation the consumers’ willingness-to-pay is higher than in the previous treatments.
Furthermore at the end of the experiment a short questionnaire has to be filled in by the participants. This questionnaire will consist of some control questions and questions about how the participants see themselves when it comes to innovation adaptation. The diffusion of innovation theory states that there are different kinds of people that hold different kinds of attitudes towards innovation. By controlling for a moderating effect at the end of the experiment we hope to be able to draw inferences about which analogy appeals most to certain kind of consumers. This information could then be used in order to market products in different ways to target different kinds of consumers.
4.3 Sample
In order to enhance the validity and generalisation of the research it was strived to obtain a sample that matches the Dutch population closely. In total 404 people participated in the
research, of which 202 male and 202 female. This comes close to the Dutch national average of 49.5% male and 50.5% female as shown in table 4.3a. Both the men and women in the sample tend to be a bit younger than the Dutch national average. Moreover, the sample is slightly skewed to the right, because the age group of 20 to 65 years comprises for 91.7% of the sample. This means that these are highly overrepresented in comparison with the Dutch national average. In addition, 24.3% of the participants indicate to be a full-time student. When looking at the educational level of the sample we see that a fairly large percentage of the sample consists of people with a higher education degree (10.4%) and an university degree (30.4%). The high number of people with a high education can be explained by the fact that a fairly large portion of the sample is obtained from students who recruited fellow students or former students for the sample (Table 4.3c). Table 4.3e shows that most of the people in the sample are familiar with online shopping and do this on a regular basis. 17.8% of the participants indicated that they own an Apple iPhone. The Apple iPhone with a broken screen is used as an analogy in the third
treatment, so a large part of the sample owns the product that is used in the analogy. This should be accounted for when analysing the data and making inferences about the variable influencing the willingness-to-pay (Table 4.3d).
4.4 Variables and measurements
The data on the two independent variables will be collected during the live Vickrey second price sealed-bid auction (Vickrey, 1961). In addition to the newly gathered data some other data such as demographics are gathered from the Veylinx database. Demographics such as gender and age are known from all participants, whereas other variables such as educational level, experience with online shopping, and smartphone software are not known from the total sample. Since these data are only used for descriptives this should not influence the results of this research.
Table 4.3a. Average age and gender distribution Dutch population compared to sample
Men Women
Average Age Dutch Population 39.9 41.7
Average Age Sample (N=404) 39.7 38.2
Gender Distribution Dutch Population 49.5% 50.5%
Gender Distribution Sample 50% 50%
Table 4.3b. Age distribution Dutch population compared to age distribution sample
< ,20 20 - 40 40 - 65 65 - 80 80, > Dutch Population 23.1% 24.6% 35.5% 12.6% 4.2
% Sample (N=404) 2.4% 55.6% 36.1% 5.4% 0.5%
Table 4.3c. Education level
Primary Education Secondary Education Vocational Education Higher Education University Degree Sample (N=404) 1.7% 4.7% 10.4% 10.4% 30.4%
Table 4.3d. Smartphone type
IOS Android Windows Other Non Unknown Sample
(N=404)
17.8% 25.5% 1.2% 2.0% 7.2% 46.3%
Table 4.3e. Online shopping frequency
Never Yearly Monthly Weekly Daily Unknown Sample
4.4.1 Willingness-to-pay
The participant's willingness-to-pay is measured by looking at their auction bid value. Since the bid is placed using a Vickrey auction, the willingness-to-pay responds closely to the consumer's floor reservation price (Wang et al., 2008; Vickrey, 1961).
4.4.2 Early adaptors
The goal of the post-experiment questionnaire is to identify early adaptors. According to the distribution of innovation theory only around 16% (2.5% innovators and 13.5% early adaptors) make up this part of the population. The questions are ranked on a 5-point likert scale, ranging from 1 for I totally agree to 5 I totally disagree. Meaning that someone who would totally agree on all four questions, and therefor be identified as an early adaptor, scores a total of 4 points. Looking at the 16th percentile (of the top 50 percentile of auction bid heights) shows that this cut-off point is 8 points. Therefor someone in this research will be identified as an early adaptor of the individual has a score between 4 to 8 points on the four questions. This means that from the 170 people who completed the questionnaire a total of 37 have been identified as early adaptor and 133 as follower.
5. Results
The following sections include the results of the data analysis. First, some of the descriptive statistics about the sample are given. Secondly, the Cronbach's Alpha is calculated in order to compute the reliability of the scale used in order to define an individual as an early adaptor. To conclude the ANOVA-tests, independent t-tests, and regression analysis are carried out in order to test the hypotheses.
5.1 Descriptive Statistics
This section will give descriptive statistics on the data collected on willingness-to-pay and on the identified early adaptors. We will focus on checking the parametric statistics, which will be used to find evidence for the hypotheses. In order to generalize the findings about this research, the assumption of normality and homogeneity of variance need to be explored. The central limit theorem states that if the sample data are approximately normal, then the sample data will most likely be normally distributed too (Field, 2009). For this to be the case, the sample has to be at least (N=30). This is the case in all situations. Therefor the sampling distributed can be
approximated by a normal distribution. Some visuals will be added to the quantitative data in order to assess its normality (Field, 2009)(Appendix A).
5.1.1 Auction bid data
In total there were three different treatments. The mean auction bid value for treatments 1,2, and 3 including zero bids are M=214.72, M=257.21, and M=378.09 respectively. With Standard Deviations of SD=24.369, SD=28.714, and SD=688.004 respectively. When looking at the frequency distribution histograms (Appendix A) it shows that for all the three treatments the frequency of the bid amount is highly positively skewed. Similarly, the box plots also show that for none of the treatments the distribution is normally distributed and that the bid amount is highly positively skewed. In addition they show that most bids are highly clustered at the low end, which is mainly caused by the amount of zero biddings that are included in these
descriptives.
For the first treatment the Skewness value is 1.789, with Standard Error of Skewness SE=.261, the second treatment has a Skewness value of 1.685 and SE=.209, and the third
Z=6.854, Z=8.062, and Z=18.505 respectively, which demonstrates significant (p<.001) positive Skewness for all three treatments. Similarly, the Kurtosis values are for the first treatment 3.616 with SE=.410, second treatment 2.710 with SE=.414, and lastly treatment three 19.585 with SE=.420. The calculated corresponding Z-scores of Kurtosis for treatments 1,2, and 3 are Z=8.820, Z=6.546, and Z=46.631, respectively, and are significant (p<.001) on all three treatments (Field, 2009). Concluding from these numbers it can be said that the untransformed auction bid data, including zero bid values is non-normally distributed.
Table 5.1.1a. Describtive Statistics Untransformed Auction Bid Data, including zero bid values (N=404)
Treatment Mean 95% Confidence Interval 95% Confidence Interval Std. Std. Skewness Std. Kurtosis Std. Lower Limit Upper
Limit Deviation Error Error Error
1. 214.72 166.96 262.48 24.369 24.369 1.798 .261 3.616 .410 2. 257.21 200.93 313.49 28.714 28.714 1.685 .209 2.710 .414 3. 378.09 260.27 495.91 688.004 60.111 3.923 .212 19.585 .420
In order to transform the dataset to come closer to a normal distribution, the top 50 percentile bids are selected in order to filter out zero and very low bid values. Also the natural logarithm is taken from the bid value. These transformations should help bring the distribution closer to a normal distribution, so that inferences can be made about the statistics. After deleting the bottom 50 percentile bid amount and transforming bid amount to the LN of bid amount the Mean and Standard Deviations for the treatments are M=5.831 with SD=.674, M=6.028 with SD=.646, and M=6.190 with SD=.884 respectively (see table 5.1.1b.). When looking at the Histograms and Box Plots (Appendix A) it is already clearly visible that the distribution is less skewed and comes closer to a normal distribution. With values for the treatment with Skewness .059 and SE=.293, -.709 with SE=.293, and .334 with SE=.302, respectively. Corresponding are the following Z-values: -.191, -.266, and 1.106, which show that in neither of the treatments there is a significant positive or negative Skewness. Looking at the Kurtosis value of the treatments shows that K=-.594 with SE .578, K=-.339 with SE=5.86, and K=0.88 with SE=5.95, respectively. Yielding Z-scores of Z=-1.028, Z=-.578 and Z=.148, respectively. These values are far below the values of the upper threshold of 3.29, therefor there is no significant Kurtosis. Running the Kolmogorov-Smirnova test for normality however shows that the difference is still significant in all three treatments (D(67)=0.149, D(65)=0.154, and D(63)=1.71, with (p<.001)). Therefor none of the
distributions are normally distributed. This has to be accounted for when interpreting the results and be taken into account when looking at the reliability.
Table 5.1.1b. Descriptive Statistics Transformed LN of Auction Bid Data, top 50 percentile (N=202)
Treatment Mean 95% Confidence Interval 95% Confidence Interval Std. Std. Skewness Std. Kurtosis Std. Lower Limit Upper
Limit Deviation Error Error Error
1. 5.831 5.667 5.995 .674 .082 -.056 .293 -.594 .578 2. 6.028 5.867 6.188 .646 .080 -.079 .297 -.339 .586 3. 6.190 5.867 6.413 .884 .111 .334 .302 .088 .595
Table 5.1.1c. Test for normality ln auction bid data, top 50 percentile
Kolmogorov-Smirnova Shapiro-Wilk
Treatment Statistic df Statistic df
1. .149*** 67 .951*** 67
2. .154*** 65 .950*** 65
3. .171*** 63 .967*** 63
The second assumption that will be tested for is homogeneity of variance. The homogeneity of variance can be tested by running Levene’s test (Field, 2009; Levene, 1960). Levene's test shows that the differences in variances between each condition of the ln auction bid data in the top 50 percentile of the bids are not-significant (F(2, 192)=1.796, ns. (Table 5.1.1d). To conclude, the data from each treatment seem to fail to meet the requirements of a normal distribution. It does however meet the requirements for homogeneity of variance. When running parametric tests such as regression analysis or ANOVA analysis this should be taken into account, because the lack of normal distribution may yield inaccurate results and distorted data (Field, 2009).
5.1.1d. Testing for homogeneity of variance, Levene’s Test
Levene Statistic Df1 Df2 Sig.
Top 50 percentile 1.796 2 193 .169
5.1.2 Early Adaptors
This study tries to identify early adaptors with the use of four different questions that were asked after their auction bid was placed. These questions have been constructed based on prior research. In order to assess the scale reliability it is necessary to compute Cronbach's Alpha (Field, 2009).
The data reveals that from the top 50 percentile 170 from the 195 people answered all the questions of the questionnaire. In order to compute Cronbach’s Alpha, the 25 people who have not filled in the questionnaire completely have been removed from the sample. In order for the scale to be internally reliable and consistent, Cronbach's Alpha should have a value r>.70. Cronbach's Alpha r calculated on all four items is .883. Since this is higher than 0.7 we can assume internal reliability and consistency of the scale. Maximum reliability is achieved by keeping all four items. Deleting any items from the scale would not increase Cronbach’s Alpha.
As discussed in section 5.1.1a., an individual is labelled as an early adaptor when its total score on the questionnaire is between 4 to 8 points. In total 37 individuals are labelled as an early adaptor, which makes up almost 22% of the sample, which gives us a large enough sample size to make inferences since (N>30)(Field, 2009). For the followers the M=532.05 with SD=570.042, while for early adaptors mean M=673.57 with SD=618.681. Looking at the histogram shows that both the mean of the early adaptors and followers are severely positively skewed (Appendix A). Skewness for early-adaptors is 4.763 with SE 0.210, and for the other group Skewness 31.129 with SE .417. Corresponding are Z-scores of Z=22.681 and Z=7.500, respectively, showing significant signs of Skewness (p<.001). Indicating a pointy and heavy tailed distribution. Corresponding Kurtosis values are K=31.129 with SE .417 and K=11.535 with SE=.759, with Z=74.650 and Z=15.198. Which show a significant amount of Kurtosis (p<0.001). Running a Kolmogorov-Smirnova normality confirms these findings, showing a highly significant deviation in both groups from a normal distribution (Table 5.2.2e). We can conclude from these findings that the untransformed bid amounts of the top 50 percentile of the total sample in both groups is non-normally distributed.
Table 5.1.2a. Descriptive Statistics Top 50 Percentile – Early Adaptors/followers (N=170) Consumer Type Mean 95% Confidence Interval 95% Confidence Interval Std. Std. Skewness Std. Kurtosis Std. Lower Limit Upper
Limit Deviation Error Error Error
Follower (N=37) 532.05 434.28 629.83 570.042 49.429 4.763 0.210 31.129 .417 Early Adaptor (N=133) 673.57 467.29 879.85 618.681 101.711 2.910 .388 11.535 .759
Table 5.1.2b. Test for normality auction bid data top 50 percentile early adaptors/followers
Kolmogorov-Smirnova Shapiro-Wilk
Statistic df Statistic df
Follower .312*** 133 .566*** 133
Early Adaptor .232*** 37 .713*** 37
In order to increase the normality of the distribution in order to be able to make right inferences about parametric tests, the data is transformed with the Natural Logarithm (ln). The descriptive statistics for the ln of the bid amount are described in the section below.
Table 5.1.2c Test for normality ln auction bid data top 50 percentile early adaptor/non early Kolmogorov-Smirnova Shapiro-Wilk Statistic df Statistic df Followers .164*** 133 .955*** 133 Early Adaptor .123 37 .975 37
The transformed data yields a mean M=5.9782 and SE=0.064 for the followers and M=6.218 with SE=5.127 for the early adaptors. Corresponding are Skewness .307 with SE .210 and Skewness .057 with SE=.388 respectively. Corresponding z-values of Z=1.462 and Z=0.147, meaning that in neither of the two groups signs of significant Skewness are found (p<.05). Kurtosis K=.665 and SE=.417, and K=.173 and SE=.759. Corresponding z-values of Z=1.595 and Z=.228. Also the Kurtosis z-values indicate no significant Kurstosis (p<.05). Running a
Kolmogorov-Smirnova normality test shows that for the followers the statistic is .164 on a (p<.001) significance level, meaning that according to this test this groups distribution differs significantly from a normal distribution. For the early-adaptors the statistic of .123 is non-significant, therefor we can assume a normal distribution. Since the tests on Skewness and Kurtosis however failed to significantly (p<.05) find proof for a very unusual distribution, we will continue this research with the transformed data set and assume that both groups are close enough to a normal distribution in order to run parametric tests. However, one should take into account when making inferences that the KS-test showed a significant deviation from a normal distribution. Testing for homogeneity of variance shows F(2, 168) = .015, ns, therefore equal variances have to be assumed.
Table 5.1.2d. Descriptive Statistics Top 50 Percentile – Early Adaptors/Followers (N=170)
Consumer Type Mean 95% Confidence Interval 95% Confidence Interval Std. Std. Skewness Std. Kurtosis Std. Lower Limit Upper
Limit Deviation Error Error Error
Early Adaptor (N=37) 5.982 5.856 6.108 .735 .064 .307 .210 .665 .417 Follower (N=133) 6.218 5.960 6.476 .774 .127 .057 .388 .173 .759
5.1.2e. Testing for homogeneity of variance, Levene’s Test, early adaptors and followers
Levene Statistic Df1 Df2 Sig.
Sample (N=170) .015 2 168 .903
5.2 Hypothesis
In this section the data will be analysed in order to draw conclusions about the hypotheses states earlier in the third section of the research.
5.2.1 Auction bid Data
The first part of the experiment tests the willingness-to-pay for an innovative product using three different treatments. H1a states that the use of a visual and textual analogy leads to a higher willingness to pay. As can be seen in the diagram in Appendix A the mean bid is higher in both
normally distributed. In order to come closer to a normal distribution we will look only at the top 50 percentile of the bid amounts. This will reduce our sample from N=404 to N=195, which is still large enough of a sample size. To come even closer to a normal distribution the bid amount will be transformed to the Natural Logarithm of the bid amount. Even though this brings the distribution closer to a distribution as can be seen in Appendix A, it is still not normally distributed. Levene's test for the transformed data is non-significant (F(2, 192)=1.796, ns, so equal variances have to be assumed.
Even though the sample is not normally distributed, an ANOVA test is still generally known to be a good test (Field, 2009). To see if the difference in the bid amount between groups is significant a one-way independent ANOVA test was carried out to compare the three means of the bid values. As you can see in table 5.2.1b F(2, 192 = 3.835, p < 0.05), meaning that there is a significant difference between the means of the ln bid amount of the three treatments. In addition to the combined trend, it shows that there is a significant linear trend F(1, 192 = 7.631, p <0,05), but for the quadratic trend F(1, 192) = 0.022, ns) no evidence is found.
5.2.1a Descriptive statistics of ln auction bid top 50 percentile data
Treatment N Mean Deviation Std. Std. Error 95% Confidence Interval for Mean
Lower Bound Upper Bound 1 67 5.834 .674 .082 5.667 5.996 2 65 6.028 .646 .080 5.867 6.188 3 63 6.190 .884 .111 5.967 6.413 Total 195 6.013 .750 .054 5.907 6.119
5.2.1b ANOVA of ln auction bid top 50 percentile
Sum of Squares df Mean Square F Sig. Between
Groups 3.467.220.947 2 1.733.610.473 5.741** 0.004 Within
Groups 57.977.302.048 192 301.965.115 Total 61.444.522.995 194
Next a linear regression analysis is carried out in order to see if there is a significant difference in willingness-to-pay between the base treatment without an analogy and the treatments with an
analogy. The regression analysis shows that there is a significant (p <.05) difference in willingness-to-pay when an analogy is added to the treatment, finding support for H1a.
When looking at the plot of the means (Appendix A), it is clearly visible that there is an increase in willingness-to-pay between treatment 1 and treatment 2, and treatment 2 and treatment 3. Running a multiple comparison LSD test shows that the mean difference between treatment 1 and treatment 2 (MD=-83.094 with SE=95.669) is non-significant. Meaning that there is no evidence found for H1b. The difference in willingness-to-pay between treatment 1 and 3 (MD=-316.527 with SE=96.437, p>0.001) is highly significant, resulting in support for H1c. For the difference in willingness-to-pay between treatment 2 and 3 (MD = -233.433 with SE = 97.153, p<0.05) there has been found significant evidence, finding support for H1d.
To conclude, evidence has been found for H2a, so the mean willingness-to-pay for treatments with a visual and textual analogy is higher than for the control treatment without. The experiment failed to find support for H1b, meaning that the use of a low-value analogy doesn’t significantly increases the willingness-to-pay. H1c and H1d have been confirmed, so from this research can be concluded that the use of a visual and textual analogy with a high value will significantly increase willingness to pay over a situation without- or with a lower valued analogy.
In addition to these findings the amount of zero-bids has been analysed from the complete sample (N=404). For treatment 1 this was 39.9%, for treatment 2 34.8% and for treatment 3 32.8%. These numbers show a linear trend, in which apparently less people tend to bid zero (are completely not interested in the product) when more information (higher value analogies) are apparent. Unfortunately no statistical inferences can be made about these numbers due to the high non normal distribution of the total sample and the presence of close to zero bids. It is however interesting to see that the analogies seem to have some affect, and this could be studies in future research.
Table 5.2.1c Linear regression analysis between base treatment and others
Unstandardized Coefficients Standardized
Coefficients T Sig.
Model B Std. Error Beta
(Constant) 422.045 67.959 6.210*** .000 Treatment2&3 197.986 83.880 .168 2.360* .019
5.2.1d Multiple Comparisons, LSD-Test LSD (I) Treatment (J) Treatment Mean Difference (I-J)
Std. Error Sig. 95% Confidence Interval Lower Bound Bound Upper 1 2 -83.094 95.669 .386 -271.79 105.60 3 -316.527*** 96.437 .001 -506.74 -126.32 2 1 83.094 95.669 .386 -105.60 271.79 3 -233.433* 97.153 .017 -426.06 -41.81 3 1 .316.527*** 96.437 .001 126.32 506.74 2 233.433* 97.153 .017 41.81 425.06 5.2.2 Early Adaptors
Table 5.2.2a shows that overall the willingness-to-pay of early adaptors is higher (6.218 vs. 5.982) of the followers. Comparing the means of the early adaptors with the followers shows however that this difference is not significant. Therefor failing to find support for H2a, meaning that early adaptors do not have a higher willingness-to-pay for innovative products than
followers.
5.2.2a Group Statistics Early Adaptors vs. Followers
Group N Mean Std. Deviation Std. Error Mean
Early adaptor 37 6.218 .774 .127
Follower 133 5.982 .735 .064
In order to examine if the degree of being an early adaptors negatively moderates the effect of an analogy on willingness-to-pay a linear regression is carried out. In order to check for the moderating effect the treatments are separated in two groups: one group without analogy and
5.2.2b Comparison of the mean early adaptors vs. followers on willingness-to-pay
95% Confidence Interval of the Difference T Df Sig. (2-tailed) Mean Difference Std. Error
Difference Lower Upper Equal
Variances
one group in which the analogy is present. The standardised value from this variable is then multiplied with the standardized value of the degree of being an early adaptor in order to create the moderator variable. A linear regression analysis including the moderator value shows that the Adjusted R-Square dropped from 0.024 to 0.020, making this model a less good of a fit.
Table 5.2.2c shows that there is a small negative moderating effect, but with its significance of .561 this effect is not significant. Therefor failing to find support for H2b.
Table 5.2.2c Coefficients
Unstandardized Coefficients
Standardized Coefficients
Model B Std. Error Beta T Sig.
(Constant) 5.844 .102 57.069 0.000
Early Adaptor .202 .139 .112 1.453 .148 Analogy .216 .122 .135 1.770 .079 Analogy*Early
6. Discussion
6.1 General Discussion
The purpose of this study was to describe and explain if the use of analogies could successfully be used as a consumer-learning strategy. Research has been conducted on the influence of analogies on the influence to pay, and also it tried to identify early adaptors and their different attitude towards innovative products. It was proposed that the use of a textual and visual would influence the consumer's willingness-to-pay. It was also proposed that different analogies would yield different results. In addition the author suggested that different consumers would react differently to the use of analogies.
The data analysis shows that the use of an analogy indeed makes an influence on a
consumer’s willingness-to-pay. It shows that during the treatments in which an analogy was used consumer willingness-to-pay was significantly higher than in the control situation without any analogy. In contrary to what the author expected, the use of a low value analogy does not have a significant positive influence on willingness-to-pay. The data however showed support for H1c and H2d, meaning that the use of a high value analogy has a significant positive influence on willingness-to-pay, compared to a situation without any analogy or one with a low value. This is in line with what would be expected from previous research (Shiu & Walsh, 2011; Feiereisen et al., 2008; Gregan-Paxton et al., 2002).
The product that was auctioned was a really new product and it was very unlikely that the consumer would have ever been in contact with this product before. Therefor the consumer would perceive a high amount of uncertainty towards the product, and thus this would have a negative influence on willingness-to-pay (Mitchell, 1999; Sheth, 1981; Shiu & Walsh, 2011; Taylor, 2014). By adding an analogy the individual was able to use the information from the base domain, in order to create inference about the target domain (Gentner, 1983; Gentner &
Markman 1997; Gregan-Paxton et al., 2002). This new knowledge helped the individual to create inferences about the product. This increased knowledge reduced the uncertainty and raised the individuals valuation of the product (Shiu & Walsh, 2011; Schmidt & Calantone, 1998).
The use of an analogy can be a valuable consumer-learning tool, but only when used in the right way. Even though the use of a low value analogy had a positive effect on
willingness-to-pay, the difference was not significant. The support for H1c and H1d show that the use of a high value analogy leads to a significant increase in willingness-to-pay.
The theory behind the use of analogies is that the consumer can take existing knowledge from a product with similar features, and transfer these to the new product. It goes without saying however that if the individual is not familiar with the analogy, than the analogy would not help in gaining more knowledge (Roehm & Sternthal, 2001; Getner, 1983). The analogy used in the second treatment, someone repairing pants, is most likely something that can be recognized by everyone. Therefor the lack of support found for H1b is most likely not caused by the lack of ability to transfer knowledge from the base domain to the target domain. The low value analogy most likely was recognized, knowledge was transferred, but the low value of repairing a pair of pants caused a low value inferences about the product. The increase in knowledge did reduce uncertainty, increasing willingness-to-pay, but the low value of the analogy failed to increase the bid amount in a significant way.
This in turn would also explain why during the third treatment, with a high value analysis of a smartphone screen being repaired, the increase was significantly higher. The consumer recognizes the event of replacing a smartphone screen and sees this as something that is more valuable. Since the value of the analogy is relatively high, the valuation of the product is also higher than in previous conditions.
From the existing literature it was hypothesised that early adaptors would react differently towards innovative products. Their specific characteristics and their attitude towards perceived uncertainty would make them more attracted to innovative products, and therefor have a higher willingness-to-pay (Robinson, 2009). Even though the research showed that overall early adaptors had a higher willingness-to-pay, this increase was not significant. Also no proof has been found for the moderating effect. Reason for not finding support for the two hypotheses might be that the scale used to identify early adaptors is not accurate enough. Therefor consumers who are not really early adaptors are still labelled as such. These people would according to the literature bid a lower amount of money on the product, which lowers the average willingness-to-pay of the early adaptors. This could in turn have led to the failure of finding support for these hypotheses.
To conclude, this research shows that the use of an analogy is a successful consumer-learning tool in order to increase consumer knowledge and consequently willingness-to-pay. It
also shows that the kind of analogy used affects the willingness-to-pay. Advertisers should take this into consideration when using analogies. They should use an analogy that is easily
recognized by everyone, and is highly valued so that some of this high valuation ‘rubs off’ on the inferences about product. It should however be noted that the use of too highly valuations might lead to not meeting consumer expectations, which in turn could lead to bad reviews, negative word of mouth and a failure of the product.
6.2 Theoretical Implications
The findings from this study seem to support the existing literature, which suggest that the use of analogies is a successful way of transferring knowledge from a known domain to an unknown domain. It seems that the perceived uncertainty of buying a product while lacking product knowledge and experience could be reduced with the use of analogies. Furthermore this research finds support for most of the hypotheses stated by the author about the effects different analogies have on the consumers’ willingness-to-pay. The research found support that different kinds of analogies yield different valuations by the individual; therefor influencing it’s inferences about the product and consequently the willingness-to-pay.
The research also adds to the existing literature on early adaptors. The results show that early adaptors don’t have a higher willingness-to-pay for innovative products than followers. Even though an increase in willingness-to-pay was visible, this difference failed to be significant. The assumptions that early adaptors tend to value innovative products more, and therefor are willing to pay a higher price compared to followers lacks to find support. The degree of being an early adaptor doesn’t seem to influence willingness-to-pay.
6.3 Managerial Implications
Since 40 to 90 per cent of RNP fail (Cierpicki et al., 2000; Breu et al., 2008), and R&D one of the companies’ biggest expenditures is, is it important that innovations catch on (Gielens, 2008). Early adaptors have proven to be of high importance in the diffusion of innovation (Robinson, 2009; Rogers, 1983; Surry & Farquhar, 1997). This research shows that the use of analogies is a valuable consumer-learning tool in order to reduce the perceived uncertainty towards a really new product among followers. This research is important for managers, because it shows that the right
use of analogies can lead to a significant increase in willingness-to-pay. By using an analogy that is fitted in the right way the willingness-to-pay can be highly increased. Not only does the use of analogies seem to raise the overall willingness-to-pay, it also lowers the zero-bids. This suggests that the analogies successfully educate people in what the product can do, therefor making them more interested in the product.
Managers, and especially marketers, should use these findings when creating their
advertisements. Most companies market their RNP’s already together with a textual and/or visual analogy, but this research shows that the kind of analogy used has significant implications on the consumers interest and willingness-to-pay. Using high value analogies seem to increase the consumers’ perceived valuation of the product.
One final note on this however is that marketers should be careful not to oversell their product. By using analogies that are valuated much higher by the consumer than the actual value that the actual product delivers, the consumer might feel unsatisfied after buying the product. An unsatisfied consumer could lead to bad reviews and word-of-mouth, harming the success of the really new product.
6.4 Implications and Future Research
The results of this study must be interpreted in light of its limitations. First of all there was a considerable amount of people who placed a zero-value, or close to zero-value bid. This caused the untransformed data to be significantly positively skewed. Although the sample has been modified by taking only the top 50 percentile of the sample, and in one of the cases taking the Natural Logarithm in order to come closer to a normal distribution, a normal distribution could not be assumed.
Roger’s (1983) diffusion of innovation theory states that the spread of innovators based on their relative timing of adaptation is considered to be approximately normally distributed. Meaning that in general the willingness-to-pay in the population should also be approximately normally distributed. Reason why the sample was positively skewed and non-normally
distributed might be caused due to the fact that the sample is too small. Even though we’ve assumed that findings from this study are not influenced too much by the non normal distribution, future researchers should try to increase sample size in order to come closer to a normal
Future research should also try to control for multiple additional variables, such as net income and product preferences, in order to make better inferences about the data obtained. This was however not possible within this research, in order to keep it feasible. It is worth mentioning that the amount of zero bids seems to be affected by the analogies. A decreasing linear trend seemed to be visible, which suggests that more information (analogies) and the higher the value of the analogy, the more interested consumers are in buying the product in general. To make statistical inferences about this a larger sample is however needed and a different research set-up would be recommended.
Another suggestion for future research is to try to identify early adaptors more carefully. Since unexpectedly support could be found for neither H2a. nor H2b., it seems that our selection procedure of the early adaptors might not be efficient enough. Future researchers should try to come up with a better model in order to identify early adaptors in a more careful manner. In order to keep this research feasible, it was not possible to do any more testing on this subject. Also the sample-size could have influenced the results on this part, since after the transformations the sample of Early Adaptors (N=37) and others (N=133) was not very large. A larger sample size for future research might yield results that are more useable.
The results from this research show that analogies can be used as an effective consumer-learning tool. The research succeeded in showing that different value analogies have a significant impact on willingness-to-pay, but which analogy should be used in which situation is still not known. More research should be conducted in order to find out which analogies work best with certain products. Also as already mentioned before, a survey should be held after the consumer bought the product, to find out if the analogy wasn’t misleading for the consumer. If the consumer feels the product doesn’t meet its expectations, then this could lead to unsatisfied customers. Future research should try to use a larger amount of different value analogies, and should also try to use different products from different categories. It might be possible that the analogies increase willingness-to-pay in one category, while it may fail to do so in another category.
