PURCHASE INTENTION

(1)

C AN RECOMMENDATIONS IN ADVERTISEMENTS BACKFIRE ?

T HE EFFECT OF

RECOMMENDATIONS ON

PURCHASE INTENTION

(2)

The effect of

recommendations on purchase intention

Can recommendations in advertisements backfire?

by Eliza Komen

University of Groningen Faculty of Economics and Business

Master thesis, MSc Marketing

August, 2012

Supervisor 1: Dr. Jia Liu Supervisor 2: Stefanie Salmon

Bataviastraat 46a 9715 KP Groningen 0625494975/0508538580

e_komen@hotmail.com S1816527

(3)

Management summary

In this research the influence of recommendations on purchase intention is investigated. The general findings in literature say that recommendations enhance purchase intention because it lowers search costs and reduces information asymmetry. But nowadays, consumers have a greater urge to differentiate themselves and they want the freedom to choose for themselves what is good or not.

This is called the need for uniqueness (Hoyer& MacInnis, 2008). Could therefore recommendations lower purchase intention? Existing research on this topics stem from Leibenstein (1950), he indicated the snob-effect which means that recommendations do lower purchase intention. The author limited his findings to scarcity. Therefore the aim of this study was to investigate if purchase intention is indeed enhanced by recommendations, and whether this main relationship could become negative - counter effect - due to the moderating influence of NFU and personal relevance of the product.

The research was performed via a questionnaire with in total 159 participants. The results support the literature findings on the positive influence of recommendations on purchase intention.

Furthermore, recommendations have a negative effect on purchase intention due to NFU, which is in line with literature. However, the results did not show a significant influence of personal relevance of the product on the relation between recommendation and purchase intention.

The main implications for this study are that although it seems like a good idea at first to incorporate recommendations in your advertisements, marketers should be aware of the possible counter effect.

The results show that a high NFU is a common personality trait among consumers, hence, recommendations could therefore lower the purchase intention. Leibenstein (1950) stated that for most commodities the motivation for exclusiveness is not that great. Therefore, marketers should thoroughly investigate whether the product they want to advertise could be used by consumers to differentiate themselves. Because when the need for exclusiveness is low, marketers could include recommendation to enhance purchase intention. Whereas for certain products marketers should leave the recommendation out, or make it less visible to diminish the counter effect of recommendations on purchase intention. The main academic implication is that this research contributes to current literature findings because it investigated the moderating effect of NFU and the personal relevance of the product on the main relation between recommendations and purchase intention.

(4)

Preface

After studying for four years in Utrecht at TIO, university of applied science, I felt the urge to develop myself more on an academic level. Therefore I decided to continue studying, and I have chosen Business administration at the University of Groningen. Since I have come in contact with marketing in my second year at TIO, I have always been very interested in the topic. Therefore it was clear to me that I was going to specialize myself in marketing.

After 3,5 years of studying in Groningen, it was time for me to start writing my master thesis. I did not had a quite clear idea about the research topic I was interested in. But then I heard of the possibility to write your thesis in a small group and I became interested. When I was starting writing my thesis, I was well aware of the challenge I was putting myself into. Writing a master thesis is a time consuming and intensive process, accompanied with ups and downs. Because I wanted to finish in this academic year, I put a lot of pressure on myself and my supervisor, dr. Jia Liu. There were times when my motivation hit rock bottom, especially due to the fact that I still had three courses to finish besides my thesis, and I had a time consuming job next to being a student. But looking back on the past six months, I could say that I have learned a lot while writing this thesis. Actually, I also enjoyed the process a lot. This is mainly because it was very motivating to write the thesis in a small group. I have received a lot of constructive feedback from miss Liu, my second supervisor miss Salmon and my fellow students Suzanne Legtenberg and Victorine Marchesini. Furthermore, Suzanne and Victorine were very helpful in certain phases like analyzing the data. With each other’s help, we managed to finish our thesis in time and keep the process of writing a thesis fun.

I should furthermore thank all the people who took the effort to fill in my questionnaire. Especially the people who took the effort of distributing my questionnaire further to their friends and family.

Without them I would have not managed to acquire enough response.

Now that I have finished my master, I am going to travel upcoming February. After eight years of studying it is now time to do absolutely nothing for a few months. Needless to say that I am very much looking forward to that. When I return I hope to get a marketing related job soon, so I can put my obtained knowledge into practice.

Groningen, 27 July 2012

Eliza Komen

(5)

List of tables

Table Page number

Table 1: Research design 26

Table 2: Sample description 31

Table 3: Familiarity with Supermarket Wine handbook and Omfietswijn-logo 31

Table 4: Sample description per condition 32

Table 5: Cronbach’s alpha scores 33

Table 6: Significant figures of regression 35

List of figures

Figure Page number

1: Conceptual model 25

2: Omfietswijn-logo 27

List of abbreviations

The following abbreviations were used throughout this report (in alphabetical order):

e.g. For example

et al. And others

etc. Etcetera

H Hypothesis

i.e. That means

NFU Need for uniqueness

SPSS Statistical Package for the Social Sciences

WOM Word of mouth

(8)

1. Introduction

Recommendations play a big role in marketing nowadays. Consumers can easily obtain information of products on the Internet. Therefore they could compare and choose from a lot of different products in their purchase decision. The general conclusion is that recommendations enhance the purchase intention of consumers, because it lowers search costs and reduces information asymmetry, which exist when there is a lot of different information that contradict each other (Laffey

& Gandy, 2009). However, consumers also have the need to differentiate themselves. Having status is getting more important. This can be obtained by acquiring unique products to impress others.

Furthermore, consumers want to have the freedom to choose for themselves what is good or not instead of letting others tell you that, this is called the need for uniqueness (Hoyer & MacInnis, 2008). Could recommendations therefore have a counter effect on purchase intention? To answer this question, this research answers the following problem statement;

“How does included recommendations in advertisements influence purchase intention, and could this influence be turned due to the moderating effect of personal relevance of the product and

need for uniqueness on this relation?”

The following research questions are answered; how does included recommendations influence purchase intention? How does personal relevance of the product affects the relation between recommendations and purchase intention? How does the need for uniqueness affects the relation between recommendations and purchase intention? To answer the problem statement I undertook a quantitative research - questionnaire - among 159 people.

Based on existing discussion and publication we may conclude that recommendations influence purchase intention because research shows that authority sells (Jones, 2011), that people are influenced by word of mouth in their purchase decisions (Chen, Wang & Xie, 2011), and that the demand for a commodity is increased due to the fact that others are also consuming the same commodity - bandwagon effect - (Leibenstein, 1950). Therefore it could be concluded that products which are recommended will enhance consumers’ purchase intention.

However, a lot of research is focused on how purchase intention is enhanced by positive word of mouth, authority and the bandwagon effect. Little research is focused on the topic if recommendations could have a counter effect whereby consumers’ purchase intention will decrease

(9)

the snob-effect which means that recommendation lowers purchase intention. However, prior research concerning the snob-effect is mostly based on scarcity (Van Herpen, Pieters & Zeelenberg, 2005). Furthermore, Leibenstein (1950) stated in his research that for most commodities and most buyers the motivation for exclusiveness is not that great. Is this also true for experience goods? With these goods it is hard to assess the quality before the purchase. This implies that consumers turn to various sources of information for experience goods (Nelson, 1970). Therefore, recommendations could have an effect on purchase intention for experience goods because it can be seen as an information source consumers use in their purchase decisions.

This study builds on the research of Leibenstein (1950). My main contributions are that I investigate whether recommendations negatively influence purchase intention due to the personal relevance of the product. When a product is highly relevant, it is consistent with people’s values, needs, goals and emotions and will have a higher outcome risk (Hoyer & MacInnis, 2008). When people perceive something as highly relevant the product influences people in their purchase decision due to risk of the outcome and the route of persuasion taken. If a product is highly relevant to you, people are persuaded via central processing, therefore recommendations could lower purchase intention due to counterarguments. Whenever a product has low relevance on consumers life, the accompanied outcome risk is lower and consumers are less involved in acquiring information on beforehand which implies that they could be easier influenced by simple heuristics like recommendations (peripheral route). Second, I include NFU to check whether the lowered purchase intention is owing to a high NFU. Third, it helps marketers decide whether to advertise with statements like “recommended by x people”, or “best tested according to Consumentenbond, January 2012”, this depends on the fact whether the product is of high or low relevance to the consumer. Because at a first glance, including recommendations in advertisements seems like a good idea. People could use the rule of thumb that if a lot of people have the product, or it is been recommended by either friends or an authority figure, it must be good. However, for some goods people want the freedom to judge for themselves whether the product is good or not (Hoyer & MacInnis, 2008). Furthermore, consumers want to differentiate themselves from others to give them some sort of status. Therefore it could be that you should, as a marketer, do not include recommendations in advertisements for certain products due to a possible counter effect.

I organize this paper in five chapters. I will continue with the literature review, which will be investigated in chapter two. The research design and plan of analysis are stated in chapter three. The results will be presented and discussed in chapter four. Finally, chapter five consist out of conclusion,

(10)

2. Literature review

In this chapter the findings of the current literature are discussed. It provides a theoretical background from which the hypothesis are drawn up. This chapter starts with the independent variable - recommendations -, than the dependent variable - purchase intention - will be explained and the relationship between recommendations and purchase intention is described. Lastly, personal relevance of the product and NFU are introduced as two possible moderators that influence the main effect.

2.1 Recommendations

Recommendations occur when a customer will refer a seller or product positively to another potential customer (Palmatier, Dant, Grewal & Evans, 2006). It reduces search efforts and could increase sales because recommendations deliver relevant information to consumers, but this depends on how trustworthy the source is.

There are several types of recommendations. Below is an overview.

2.1.1 Interpersonal influence

Interpersonal influence has two main types; informative and normative influence (Deutsch &

Gerrard, 1955). Informative influence refers to the tendency to accept information from others as evidence of reality. For example, opinion leaders directly influence other consumers by giving them advice and verbal directions about their search for, purchase of, and use of a product (Flynn, Goldsmith & Eastman, 1994). Normative influence on the other hand entails the tendency to conform to the expectations of others (Burnkrant & Cousineau, 1975). Hence, normative opinion leaders exert social pressure and social support and thereby influence decision making processes of the influenced consumer (Glock & Nicosia, 1964). Since people aim to create and maintain meaningful social relationships, they often engage in behaviors approved by others such as adopting a product to appeal to other product adopters (Cialdini & Goldstein, 2004). Grewal, Mehta & Kardes (2000) say that the product and situation determine which type of influence is more important.

Privately consumed goods prioritize the informative influence, whereas for publicly consumer goods both types of influences are critical. Two forms of interpersonal influences are WOM and salespersons, they are described in the subparagraphs below.

2.1.1.1 Word of mouth

Word of mouth (hereafter WOM) occurs when consumers inform one another about (un)favorable

(11)

ownership, usage, or characteristics of products, services, and sellers (Westbrook, 1987). In addition to exchanging information, consumers may also influence others simply by visibly using a product (Gilbert, Jager, Deffuant & Adjali, 2007). Hoyer & MacInnis (2008) define WOM as a nonmarketing source that is delivered personally. Several authors like Silverman (1997) suggest that WOM has the most important influence in the consumer decision making process. However, high mass media usage is necessary to increase the speed of information and therefore make consumers become aware of the product. WOM effectively encourages people to start using a product (Herr, Kardes &

Kim, 1991), and it is more likely to activate people to act upon received advice than mass media (Gelb & Johnson, 1995).

Research shows that positive WOM is more common than negative WOM (East, Hammond & Wright, 2007) and it is more persuasive than written information (Herr et al., 1991). Online forums, blogs, websites and e-mail can potentially magnify the effect of WOM (Hoyer & MacInnis, 2008). This is because review sites on the Internet have also become more influential. Furthermore, producers discovered blogs as a possible channel to stimulate the social diffusion of their products. By sending free samples to popular bloggers, they hope to get positive product reviews that generate WOM recommendations (Gilbert et al., 2007). An example of this is Nivea which uses readers of a certain magazine that fits the target group as consumers that give recommendations to other consumers.

They send samples of new product to readers, after that they stated in the magazine that “86% of the readers is convinced and would recommend product X to their friends”.

WOM can come from a person’s reference group (in-group) or from opinion leaders like market mavens which is a consumer on whom others rely for information about the marketplace in general.

A market maven seems to know all about the best products, good sales and the best stores (Hoyer &

MacInnis, 2008).

Not every person has the same amount of influence on other consumers. This depends on the status that people have. People of high status may have a disproportionate influence on other consumers, which is one of the reasons why many producers use famous people to endorse their products (Gilbert et al., 2007). Different types of influential consumers possess varying characteristics, which implies their varying influence on the consumer around them. van Eck, Jager & Leeflang (2011) made a typology of influential consumers;

 Innovators/early adopters (Engel, Kegerreis & Blackwell, 1969) which are consumers who influence other consumers through their innovative behavior and knowledge about a specific

(12)

 Market mavens (Feick & Price, 1987) are consumers who may not have knowledge about a specific product category, but rather about markets in general.

 Opinion leaders (Katz & Lazarsfeld, 1955) who are consumers that represent a combination of innovative behavior and market knowledge.

Both opinion leaders and early adopters reveal similar characteristic, which makes it likely that many opinion leaders are early adopters and vice versa (van Eck et al., 2011). By sharing their expert evaluations, opinion leaders ‘translate’ marketing messages into WOM, which recipients perceive as more reliable than an advertisement (Nielsen, 2007).

With taking all of the above into account, it seems that providing WOM may be attractive for several reasons. However, consumers who promote a product through WOM may decrease the uniqueness of their possessions. Thus, positive WOM may hurt consumers who have a high NFU (Cheema &

Kaikati 2010). The NFU is later in this report described.

2.1.1.2 Salespersons

Recommendations could also stem from salespersons at the point of purchase. Zeng and Reinartz (2003) stated that consumers rely on salespersons for successfully choosing among product alternatives. This is because it requires certain levels of consumer expertise to choose successfully, which is the understanding of the attributes in a product and knowledge about how various alternatives stack up on these attributes. Clearly, information does not mean expertise.

Hoyer & MacInnis (2008) define salespersons as a marketing source delivered personally. Consumers may use the knowledge and assistance of salespersons to further their personal goals. Salespeople are sometimes questioned for their credibility because they work for the company and could therefore be not objective.

2.1.1.3 Credibility

Recommendations stem from different sources. But which one is the best for consumers? This depends on how credible consumers perceive the recommendation. Consumers tend to perceive information delivered through marketing sources as being less credible, more biased, and manipulative. In contrast, nonmarketing sources appear more credible because we do not believe that they have a personal stake in our purchase, consumption, or disposition decisions (Hoyer &

MacInnis, 2008).

(13)

Because nonmarketing sources are credible, they tend to have more influence on consumer decisions than marketing sources do (Lazarsfeld, Berelson & Gaudet, 1948). We tend to believe information that we hear from people with whom we have close relationships, in part because their similarity to us (and our values and preferences) makes their opinions credible (Duhan, Johnson, Wilcox & Harrell, 1997). Certain people are also regarded as more credible than others because they are experts or are generally recognized as having unbiased opinions. Similarly, certain media have higher credibility because they base their opinions on carefully acquired and trustworthy information (Hoyer & MacInnis, 2008).

2.1.2 Automatic recommendation systems

Bodapati (2008) distinguish two kinds of recommendation systems based on the underlying technology. Recommender systems can be broadly categorized as content-based and collaborative.

The main difference between the two systems are that content-based systems match customer interests with information about the products, while collaborative systems utilize preference ratings from the other customers (Cheung, Kwok, Law & Tsui, 2003).

Content-based systems provide recommendations to a customer by automatically matching his interests with product contents, and have been reported in the literature on recommending web pages, newsgroup messages and news items. Notice that recommendations are made without relying on information provided by the other customers. Typically, they use an intelligent engine to mine the customer’s ratings records and then create predictive user models for product recommendation (Bodapati, 2008). These customer ratings may either be acquired explicitly by form- filling or implicitly via an intelligent agent (Cooley, Mobasher, Srivastava, 1999). Purchase records can also be good indicators of customer preferences (Bodapati, 2008). This is in line with a cornerstone idea in customer relationship management whereby a firm should emphasize selling more products to existing consumers rather than merely acquiring more consumers. To achieve add-on selling, many firms use automatic recommendation systems (Bodapati, 2008). The author further discussed that companies should not only base their recommendation decisions on the raw purchase probability but also on the purchase probability conditional of the recommendation; thus, it is important to consider the sensitivity of the purchase to the recommendation.

It has been intuitively assumed that recommendations increase sales by providing high-quality, useful information to customers. However, customers may not trust recommendations whose rationale is not properly explained (Wang & Benbasat, 2007).

(14)

Cheung, Kwok, Law & Tsui (2003) call automatic recommendations a form of personalized direct marketing, whereby direct marketing is defined as a promotion process which motivates customers to place orders through various channels (McDonald, 1998). Due to the increasing popularity of Internet commerce, a wealth of information about the customers can now be readily acquired online.

One consequence of the giant growth of Internet is that a tremendous amount of product information can now be made available to the consumers at a very low cost. However, consumers, who were used to having only a limited range of product choices due to physical and/or time constraints, are now facing the problem of information overload. An effective way to increase customer satisfaction and consequently customer loyalty should be one that helps the customers identify products according to their interests. This calls for the provision of personalized product recommendations (Cheung et al., 2003).

However, Pathak, Garfinkel, Gopal, Venkatesan & Yin (2010) argue that recommendations itself add to information overload and could therefore become lost in the clutter. This is because a typical product web page of an online retailer includes information such as product features, images, expert reviews, customer reviews, and ratings. Adding recommendations could enhance the information overload.

2.1.2.1 Influence of automatic recommendation systems

Automatic recommendation systems attempt to analyze a customer’s purchase history and identify products the customer may buy if the firm were to bring these products to the customer’s attention (Bodapati, 2008). An example of this is YouTube, amazon.com (“recommended because you purchased…”/“Consumers who bought this item also bought….”) or hm.com whereby the site automatically come up with the same kind of clothes that you might also like when you are shopping online. The aim of recommender systems is to help users navigate through an universe of products in an online store by providing products that the users may like (Choi, Lee & Kim, 2011). Automatic recommendation systems contribute to better customer experience and enhance success in meeting customer needs (Liang, Lai & Ku, 2006). These systems reduce search effort exertion by customers and increase cross-selling by providing relevant product recommendations (Choi et al., 2011).

2.1.3 Independent websites

Lastly, another kind of recommendations are those which stem from independent comparison websites like consumentenbond.nl, kiesbeter.nl or independer.com. Comparison websites enable

(15)

consumers to evaluate and choose products from a range of providers (Laffey & Gandy, 2009).

Moreover, the authors stated that comparison websites lower search costs and reduce information asymmetry by offering extensive product information to support consumers in making a product choice. Zeng and Reinartz (2003) call these website expertise providers. Their mission is to facilitate the development of consumer expertise and help consumers make better decisions. They generate their own proprietary information that is very valuable for consumer decision making. Most comparison sites are based on expensive products whereby the risk of purchasing a product is more present than for convenience goods. However, there are also comparison sites for convenience goods. An example is omfietswijn.nl. This site ranks wine that different supermarkets are selling.

It can be argued that independent website are some kind of WOM. However, the difference between WOM and independent website is that the latter contains information and recommendations which stem from comparing and evaluating different brands. The purpose is to help consumers in their purchase decision by recommending the best products in certain product categories by comparing and evaluating different products. An intensive research has been set up to compare the different products. In the case of WOM, this stems normally from family and friends who purchase a certain product and are satisfied without an extensive investigating beforehand or trying different products.

Furthermore we can say that websites like kiesbeter.nl and consumentebond.nl are more independent - the sites do not gain anything by their recommendations - and more knowledgeable than regular WOM.

These independent comparison sites are becoming more popular. Furthermore, most product providers cannot simply ignore the impact of comparison websites; they are a too important distribution channel and are encouraging customers to regularly switch (Laffey & Gandy, 2009).

2.1.3.1 Credibility

Zeng and Reinartz (2003) stated that independence and trust are key ingredients for success in this domain. Independence is necessary to guarantee that the advice that consumers are going to get is unbiased. Consumers also need to build up trust towards the expertise providers, trust in its capabilities of providing deep and sound advice and trust in its independence. Because these sources are independent - a nonmarketing source - they are viewed by consumers as high in credibility. Hoyer

& MacInnis (2008) say that many consumers choose movies based on films critics’ recommendations, make dining decisions based on restaurant reviews, make buying decisions based on Consumer Reports articles, and choose hotels based on the American Automobile Association’s rating. This

(16)

shows that people are influenced in their purchase decision through an entity outside a marketing organization.

In short we could say that independent websites influence consumers because they are seen as a trustworthy - nonmarketing - source to obtain information. They will enhance the purchase intention because people are influenced in their purchase decision through an entity outside a marketing organization. Therefore I focus on independent website in this research because I do not want the answers to be biased by credibility issues. This means that recommendations that stem from marketing sources (e.g. salespersons) will not be investigated. Besides, Internet is a very popular source to obtain information. About 95% of Dutch people who is active on Internet use it to obtain information (cbs.nl).

2.2 Purchase intention

Purchase intention is the dependent variable. This research investigates whether recommendations (the independent variable) has influence on purchase intention (dependent variable) and how this relationship is influenced by two moderators.

Chintagunta & Lee (2012) say that the intention to purchase a particular product precedes the actual purchase. It reflect consumers’ likelihood of purchasing a product.

2.2.1 Actual and ideal state

Normally someone buys a product because they experience a gap between their ideal state and their actual state. Hoyer & MacInnis (2008) define an ideal state as a way that consumers would like a situation to be. We form our ideal state by relying on simple expectations, usually based on past experience. Furthermore, it can be a function of our future goals or aspirations. Both expectations and aspirations are often stimulated by our own personal motivations - what we want to be, based on our self-image - and by aspects of our own culture. Reference groups also play a critical role because we strive to be accepted by others and because reference groups serve as a guide to our behavior. Finally, major changes in personal circumstances can instigate new ideal states (Hoyer &

MacInnis, 2008).

The actual state is the real situation as consumers perceive it now. It is mostly influenced by simple physical factors like malfunction of running out of a product. Needs also play a critical role, if your friends make fun of your clothes, your actual state would not be acceptable. Problem recognition occurs if consumers become aware of a discrepancy between the actual state and the ideal state

(17)

process. This is not for every situation and consumer true, whether a consumer is likely to act depends on the amount of discrepancy, the level of motivation, ability, and opportunity. The higher these levels, the more likely that the consumer will act (Hoyer & MacInnis, 2008). If problem recognition is stimulated, consumers will start the decision process to solve this problem. This starts with internal search, which is the process of recalling stored information from memory (Hoyer &

MacInnis, 2008). If the consumer’s decision cannot be entirely based on information from memory, they engage in external search, the process of collecting information from outside sources like magazines, dealers and advertisements (Hoyer & MacInnis, 2008).

2.3 Influence of recommendations on purchase intention

In short we could say that WOM has the most important influence in the consumer decision making process (Silverman, 1997). Furthermore, Zeng and Reinartz (2003) say that consumers rely on salespersons for successfully choosing among product alternatives. Automatic recommendation systems contribute to better customer experience and enhance success in meeting customer needs (Liang et al., 2006) which will enhance the purchase intention. Lastly, independent websites lower search costs and reduce information asymmetry (Laffey & Gandy, 2009).

Therefore we could say that recommendation have an influence on purchase intention because they assist in consumers’ purchase decision by reducing search efforts and they could increase sales because recommendations deliver relevant information to consumers, but this depends on how trustworthy the source is. Furthermore, based on existing discussion and publication we may conclude that recommendations influence purchase intention because research shows that authority sells (Jones, 2011), that people are influenced by WOM in their purchase decisions (Chen, Wang &

Xie, 2011), and that the demand for a commodity is increased due to the fact that others are also consuming the same commodity - bandwagon effect - (Leibenstein, 1950). Therefore it could be concluded that products which are recommended will enhance consumers’ purchase intention.

Therefore I come up with the following hypothesis:

H1: Recommendations, compared with no recommendations, will increase purchase intention.

2.4 Need for uniqueness

NFU is defined as a characteristic trait of consumers who pursue novelty through the purchase, use and disposition of goods and services (Tepper Tian, Bearden & Hunter, 2001). It covers three

(18)

behavioral dimensions: creative choice counter conformity, unpopular choice counter conformity and avoidance of similarity. In the case of creative choice counter conformity the consumer’s choice reflects social distinctiveness, yet the choice is one that others will approve of. Unpopular choice counter conformity means choosing products and brands that do not conform to establish distinctiveness despite possible social disapproval. Avoidance of similarity reflects losing interest in possessions that become commonplace to avoid the norm and hence re-establish distinctiveness (Hoyer & MacInnis, 2008).

2.4.1 Consequences of need for uniqueness

People who score high on the personality trait NFU may dispose clothing that has become too popular in favor of emerging fashion trends, seek out handcrafted or personalized items, and customize products to their own specifications (Hoyer & MacInnis, 2008).

Furthermore, they may show more reactance to recommendations (Hoyer & MacInnis, 2008). This because they do not want to be said what is good and what not. They are doing the opposite of what the individual or groups wants them to do (Hoyer & MacInnis, 2008). This is because people feel that their freedom is being threatened and therefore engage in reactance (Hoyer & MacInnis, 2008).

Another consequence of a high NFU is the desire to possess unique products (Simonson & Nowlis, 2000), which provide differentiation from other people. For example, high uniqueness consumers are likely to prefer distinct product designs (Bloch, 1995) with attributes that “define the person as different from members of his or her reference group” (Snyder, 1992). High uniqueness consumers are more drawn to scarce products than low uniqueness consumers, exert more effort to own innovative products (Lynn 1992; Snyder 1992), and are more likely to choose options that others do not choose (Worchel, Lee, & Adewole, 1975).

2.4.2 Topics related to need for uniqueness

NFU is related to self-monitoring behavior whereby people look to others for cues on how to behave.

If you score low on self-monitoring behavior you are guided more by your own preferences and desires and you are less influenced by normative expectations (Becherer & Richard, 1978). Normative influence represent how others influence our behavior through social pressure, it is the collective decisions about what constitutes appropriate behavior. Normative influence can also affect conformity, the tendency for an individual to behave as the group behaves (Hoyer & MacInnis, 2008).

Stafford (1966) say that conformity is related to brand-choice congruence because you might conform buying the same brands as others in your group do.

(19)

Besides, NFU is related to the personality trait competitiveness which is de desire to outdo others through conspicuous consumption of material items like gadgets (Hoyer & MacInnis, 2008). People who express NFU are somewhat rebellious. This means that they are doing the opposite of what an individual or group wants you to do. Algesheimer, Dholakia and Hermann (2005) found that when a member feels too much pressure to perform certain rituals or assume certain roles, the desire to participate in the community or buy the brand in the future may be lowered.

Furthermore, NFU is somewhat the same as the snob effect. This is the extent to which the demand for a consumer good is decreased owing to the fact that others are also consuming the same commodity (or that others are increasing their consumption of that commodity). This represents the desire of people to be exclusive; to be different; to dissociate themselves from the "common herd"

(Leibenstein, 1950). Therefore, including recommendations could have a counter effect because if a product is recommended, the overall demand could increase. This implies a decreasing demand for people who show the snob-effect because they want to be exclusive. This is in line with this research because the NFU has similarities with the snob-effect. The author further said that for most commodities and most buyers, the motivation for exclusiveness is not that great.

In short we could say that NFU influence consumers because they are guided by the behavior of others. If a lot of people show the same behavior, high NFU consumers react by showing a devious behavior simply to differentiate themselves. This will moderate the effect of recommendations on purchase intention because if you score high on NFU, recommendations could lower your purchase intention because you want to differentiate yourself from others by not buying the product, and because you want to determine for yourself what is good or not. For people who score low on NFU, it is not likely that recommendations will decrease their purchase intention. Those consumers do not have a strong urge to differentiate themselves, furthermore, they look at other consumers for guidelines on how to behave. Which implies that recommendations could be seen as an advice on what is (dis)approved of by other consumers.

H2: Recommendations, compared with no recommendations, will decrease purchase intention for people who score high on need for uniqueness, compared with people who score

low on need for uniqueness.

(20)

2.5 Personal relevance of the product.

Hoyer & MacInnis (2008) define personal relevance as something that has a direct bearing on the self and has potentially significant consequences or implications for our lives. Normally, products which will have a higher risk in terms of outcomes - e.g. loss of money or damage for your image when the quality is bad -, will have a higher personal relevance. The authors stated that people perceive something as personally relevant when it is consistent with their values, needs, goals and emotions.

Trampe, Stapel, Siero & Mulder (2010) stated that personal relevance is regarded as the extent to which an advocacy has intrinsic importance, or personal meaning. Consumers often buy products and brands for what they mean, rather than solely for what they can do (Belk 1988; Berger & Heath 2007). Furthermore, the consumers’ perception of the personal relevance of a product itself is being influenced by product attributes (Zhu, Wang, Yan & Wu, 2009).

Zeng and Reinartz (2003) made a distinction between products and services which are bought primarily for their physical performance, and products and services which are bought for their social image or for their sensory enjoyment. The former are called functional goods (for example, detergents, insurance, and appliances), the latter are called value-expressive goods (for example, clothing, jewelry, and office location). If a product is dominated by functional concerns, it will often be evaluated along a sober list of various product attributes, which makes comparisons quite easy. In contrast, value expressive goods are not chosen on attribute information but based on judgments that are holistic and difficult to articulate. Issues such as the relation of the product to one's self, nonverbal cues, and emotional experiences become overriding aspects of the purchase.

Moreover, there is the emblematic function of products whereby products are being used to symbolize membership in social groups. We consciously or unconsciously use brands and products to symbolize the groups to which we belong or want to belong (Edson Escalas & Bettman, 2003). A watch or a car may symbolize our social status. This is especially true for products whereby the consumer is highly involved and when the products are consumed in public. Therefore we can roughly divide products into ones that are used in a private matter and products which are used publicly. When a product is used to symbolize our social status we want to purchase something that a certain group of consumers have (Edson Escalas & Bettman, 2003). Therefore, recommendation may lower the purchase intention because if a lot of people recommend/purchase the product, and we want to distance ourselves from certain groups, we will not buy the recommended product.

(21)

2.5.1 Elaboration Likelihood Model and Heuristic Systematic Processing Model

In marketing it is widely known that when under conditions of high personal relevance people chose to spend more time thinking about the advertisement than individuals exposed to the same advertisement under low personal relevance conditions (Haugtvedt & Strathman, 1990). This is linked to two dual process theories, namely; the concept of the Elaboration Likelihood Model (Petty

& Cacioppo, 1984) and the Heuristic Systematic Processing Model (Chaiken, 1980). Since there is a great deal of overlap between these two theories, they will be integrated into one theoretical framework.

Fennis & Stroebe (2010) say that dual process theories distinguish two routes to persuasion, which form the endpoints of a continuum of processing intensity. The dual process theories of persuasion consider two modes of information processing, systematic and non-systematic (i.e. peripheral or heuristic processing). Modes differ in the extent to which individuals engage in message-relevant thought in order to decide on whether to accept message arguments. The mode used depends on processing ability and processing motivation. Processing motivation is important because unless an issue is relevant to recipients, they will not expend much effort in thinking about arguments for or against the issue. Personal relevance (i.e. the importance of an outcome for the individual) is the major variable affecting processing motivation. Processing ability is important because in order to judge the validity of the arguments contained in a communication a person needs knowledge, time and peace of mind - i.e. absence of distraction - (Fennis & Stroebe, 2010).

The central route to persuasion is taken when recipients carefully and thoughtfully consider the arguments presented in support of a position, - systematic processing -. Consumers are influenced via the quality of the arguments (Petty & Cacioppo, 1984). The second route reflects the fact that people often change their attitudes without thinking about the arguments contained in a communication, for example, because an expert or a trusted friend has made a recommendation or because the issue is unimportant. Petty & Cacioppo (1986) called this mode of attitude change the peripheral route to persuasion. Consumers are then influenced via rules of thumbs (Petty &

Cacioppo, 1984).

Originally it was assumed that peripheral (heuristic)- and systematic processing modes are compensatory: the more individuals relied on systematic processing, the less they would use heuristic processing (Petty & Cacioppo, 1986). More recently it has been suggested that the two modes of processing can co-occur if systematic processing of arguments does not allow one to arrive

(22)

at a clear-cut conclusion, for example, because the arguments contained in a communication are ambiguous (Bohner, Moskowitz, Chaiken, 1995; Chaiken & Maheswaran, 1994).

Therefore we can say that when products are highly relevant to the consumer, the advertisement with recommendation will be more evaluated through central-route processing because people are more motivated to elaborate the advertisement thoroughly. This means that consumers may come up with counterarguments because they analyze the true merits or central issues of the message carefully and effortful (Hoyer & MacInnis, 2008). Petty & Cacioppo (1984) stated that under central processing cognitive responses occur that deeply scrutinize the depicted claims in advertisements.

This could result in agreeing with the claims - support arguments -, or questioning them and come up with counterarguments. However, when people are taking the peripheral route because the product is of low personal relevance, a recommendation can function as a heuristic. This is a simple decision rule to help consumers making purchase decisions (Hoyer & MacInnis, 2008). Therefore, both routes of persuasion are applicable for this research. For example (Trampe, 2012), if a product advertise with the claim “sold more than 11.000.000 times this year” consumers in the central processing could come up with the counter argument “Do I like everything that the majority of consumers like?”. Consumers in the peripheral route could be thinking “11.000.000 people cannot be wrong”.

This indicates that via the central route a recommendation could not have the intended effect because people scrutinize the message more and could question the depicted recommendation.

They could feel that it is not trustworthy enough, or they like to determine for themselves what is good or not.

This research focuses on experience goods, which are goods whereby it is difficult to assess the quality prior to purchase and usage, for example wine. Nelson (1970) stated that when making purchasing decisions for experience goods, consumers usually turn to various sources of quality information on the product in the absence of any pre-purchase quality assessment.

When the personal relevance of the product is high, people engage more in acquiring information on beforehand. Zeng and Reinartz (2003) stated that when information search is an important factor in consumer decision making, Internet has several advantages for information search because it has greatly improved the effectiveness of information search. Furthermore, they stated that it depends on the perceived risk, frequency of purchase and functionality versus value expressive whether Internet is being used in the information search.

(23)

In short we could say that personal relevance of the product influence people in their purchase decision due to risk of the outcome, the emblematic function of products and the route to persuasion taken. This will moderate the effect of recommendations on purchase intention because if a product is highly relevant to you, recommendations could lower your purchase intention. Under central processing consumers come up with counterarguments whereby the recommendations lower the purchase intention. Whenever a product has low relevance on consumers life, the accompanied outcome risk is lower. Hence, consumers are less involved in acquiring information on beforehand.

This implies that they could be easier influenced by simple heuristics like recommendations (peripheral route). Therefore, recommendations will not decrease purchase intention for low personal relevant products.

H3: Recommendations, compared with no recommendations, will decrease purchase intention for high personal relevant products, compared with low personal relevant products.

2.6 Conceptual model

The conceptual model gives an overview of all of the above mentioned hypotheses. For the main effect I expect a positive relationship, it is logical that the purchase intention will increase due to recommendations. People could use the rule of thumb that if a lot of people have the product, or it is been recommended by either friends or an authority figure, it must be good. However, could this main effect be turned due to the need for uniqueness and personal relevance of the product.

The conceptual model for this research is depicted below.

(24)

Figure 1: Conceptual model

H2 H3

H1 Purchase intention

Recommendation

Need for uniqueness Personal relevance

of the product

(25)

3. Methodology

In this chapter the used methodology is described to investigate the research question and the formulated hypotheses. Furthermore, it explains how the independent variable, the two moderating variables and the dependent variable were measured.

3.1 Research design

To answer the problem statement;

“How does included recommendations in advertisements influence purchase intention, and could this influence be turned due to the moderating effect of personal

relevance of the product and need for uniqueness on this relation?”

I used a 2 (recommendation versus no recommendation) × 2 (low versus high personal relevance) × 2 (low versus high NFU) design. The design is depicted below in table 1.

Low personal relevance High personal relevance

Low NFU High NFU Low NFU High NFU

Recommendation in advertisement

Condition 1:

Wine, situation 1

Condition 2:

Wine, situation 1

Condition 3: Wine, situation 2

Condition 4: Wine, situation 2 No recommendation

in advertisement

Condition 6:

Wine, situation 1

I designed four different advertisements (please see appendix 1). To gain representative measure results, the minimum number of respondents was set at 160, 20 for each condition.

3.1.1 Choice of research

I used a questionnaire to investigate the problem statement. Participants were randomly assigned to one of the four advertisements. I distributed the questionnaire via internet (Qualtrics) to accelerate the coding of data and to minimize possible faults by respondents (like not filling in a question, I forced response for the questions). The link to the questionnaire was posted on Facebook and emailed to friends and family. In both cases I requested to further share or email the link to the recipients’ friends and family, to create a snowball sampling. After 1,5 week, I still did not had enough respondents. I emailed Liane Voerman to ask whether she will post my link on the Master Table 1: Research design

(26)

Marketing community on Nestor. Furthermore, I printed out my questionnaire and put it at my work - Julia’s at Groningen railway station -. After 2,5 week I had collected the number of required respondents.

In the questionnaire I mostly used 7-point Likert scales, except for the manipulation check questions.

For a complete overview of the questionnaire, please see appendix 1.

3.1.2 Choice of sample

This research had no restrictions whatsoever when it comes to participation. Everyone is allowed to fill in the questionnaire which creates a broader sample. This is because the main effect is applicable for the two sexes, all age groups and all education levels. This could enhance the generalizability of the results. The only thing that could bias the data is the nationality of the respondent, because the used recommendation - see 3.2.1 - is applicable for the Netherlands. However, this recommendation is being explained in the questionnaire, therefore people understand the recommendation and could be influenced by it.

3.2 Measures of the major variables

This part describes how the major variables - independent variable, moderating variables and the dependent variable - were measured.

3.2.1 Recommendation – independent variable

I wanted to use an existing recommendation expression to make the recommendation more persuasive and credible. Because I have chosen for wine, the recommendation

“Omfietswijn” was used, which is related to the Supermarket wine handbook.

This handbook is well known in the Netherlands. It is a guide for good supermarket wines. The Omfietswijn concept is given to wines in the guide that have a high value for money. The Omfietswijn-logo has also an own app where the wines are ranked. Because it is not a marketing source, we can see this as an independent review site - the kind of recommendation I focus on in this study -.

Two advertisements include this recommendation, the other two are without the recommendation.

Because this logo is only used in the Netherlands, I asked at the end of the survey if people are Dutch. If not, this could explain the non familiarity with the logo of certain participants. However, I also included a brief description of the meaning of the logo stated; “Recommended by Supermarket

Figure 2: Omfietswijn-logo

(27)

wine handbook 2012 as ‘Omfietswijn’. Supermarkets could have such special and delicious wine that you are happy to bike to the other side of the city for it. You can recognize real good supermarket wines by the Omfietswijn-logo” to make sure that people who are not familiar with it, still understand its purpose. This should control for the unfamiliarity with the logo.

3.2.2 Personal relevance scenario – moderating variable

I have chosen for the product wine whereby I manipulate the using occasions to create a low or high personal relevance of the product situation. One occasion is whereby you need to pick a bottle of wine for the weekly dinner with your friends (situation 1). The other occasion is a dinner whereby you will meet your parents in law for the first time. To make a good impression you bring along a bottle of wine (situation 2). Please see appendix 1 for a complete overview of the manipulated occasions.

In situation 2 the outcome is more risky, e.g. you could make a bad impression on your parents in law - high personal relevance -. But bringing a bad wine to your weekly dinner with friends will not harm you that much - low personal relevance -.

People should base their purchase intention on the described scenario – hence, personal relevance - and not on a particular wine brand. Therefore I used a non existing brand to exclude the possible brand attitude bias.

In my theoretical framework I indicated that I want to focus on experience products whereby it is hard to assess the quality before purchasing, so you probably acquire more information on beforehand, or you will be more influenced by rules of thumb like recommendations. This condition is applicable for wine because you could not easily determine the quality of the wine before using it.

3.2.3 Need for uniqueness – moderating variable

According to previous research, NFU can be measured with the scale of Snyder & Fromkin (1977).

The authors developed a scale that consist out of 32 statements. In this research I used the 10 statements which were most applicable. Participants were asked to indicate how much they agreed with each statement.

(28)

3.2.4 Purchase intention – dependent variable

To measure the dependent variable, purchase intention, I asked the likeliness of purchasing and whether people are willing to buy. These questions were based on Dodds, Monroe & Grewal, 1991 and Bone & Ellen, 1992.

3.3 Manipulation check and control variables

This part describes how the manipulations that were performed are checked. Furthermore, the controlling variables which could influence the results of this research are explained.

3.3.1 Random assignment

As said before, I used manipulation to assign the participants to one of the four conditions. To check whether participants were roughly equally divided between the four conditions, I performed a frequency test.

3.3.2 Recommendation

For the independent variable, I checked whether people saw the recommendation, hence, if the recommendation manipulation worked out correctly. People whereby the manipulation failed would be excluded for further analysis by removing them from the data set. Furthermore, I wanted to check whether people were familiar with the Omfietswijn-logo. Because if this is not the case, than the recommendation could be less persuasive.

3.3.2 Personal relevance

I instructed participants that they were buying wine for the weekly dinner with their friends, or for a dinner whereby they meet their parents in law for the first time. The first one is the low personal relevance scenario, the latter is the high personal relevance scenario. To check whether this manipulation worked out correctly I asked whether the product was important and meaningful to them (based on Mano & Oliver, 1993; Zaichkowsky, 1985).

3.3.3 Wine consumption

Even though participants were asked to imagine they are shopping for wine, and the scenarios did not explicitly stated that the wine was for own consumption, I still wanted to check whether the purchase intention could have been influenced by liking wine in general. Therefore I asked if people like red wine, if they buy it often and how knowledgeable they are. I control for these three variables in the regression. The reasons why these variables have an possible influence on purchase intention could be that if you like red wine, you are more willing to buy wine anyway. If you like wine you have

(29)

a positive attitude towards the product and you are therefore more likely to buy it. Moreover, if you buy red wine often, you probably have a preference for certain brands of wine. The wine in the advertisement was of a fictitious brand, indicating that you rather go for your familiar brand instead of this unfamiliar brand. Lastly, people who are knowledgeable are less influenced by the advertisement. They determine for themselves, based on their knowledge, whether it is a good wine.

An advertisement with or without recommendation has little influence on this decision.

(30)

4. Results

In this chapter are the most relevant results of this research described. First of all the sample characteristics and the manipulation checks are explained. After that, the reliability analysis is depicted and the main results are explained. Lastly, further results and a summary of the major findings are depicted.

4.1 Sample characteristics

This part gives an overview of the descriptives of the sample.

4.1.1 Definitive number of participants

A total of 187 people filled in the questionnaire. However, 26 of these respondents did not completed the questionnaire. All of them stopped immediately at question one. This could indicate that people were deterred of the level of English. I deleted all 26 incomplete questionnaires because they did not give me any answers whatsoever. This means that 161 questionnaires remained. After that, I checked whether the independent variable manipulation worked our correctly. Only two participants indicated the wrong answer, stating that the wine was recommended when it was not.

Because only two respondents did not indicate the correct answer, I excluded them from the data set and continued the analysis with 159 participants in total.

Because there were four different advertisement, a largely equally number of respondents per condition will give the most reliable results. The distributions of the respondents per advertisement were 39 for advertisement 1, 40 for advertisement 2, 40 for advertisement 3, and 40 for advertisement 4.

4.1.2 Description of the sample

A description of the total sample (n = 159) is depicted below in table 2:

Gender Age Nationality Education level

Female 100 (62.9%) <20 10 (6.3%) Dutch 155 (97.5%) Secondary school 1 (0.6%) Male 59 (37.1%) 21-40 131 (82.4%) Other EC 3 (1.9%) MBO 12 (7.5%)

41-60 13 (8.2%) Other 1 (0.6%) HBO 60 (37.7%)

>60 5 (3.1 %) University 86 (54.1%)

Table 2: Sample description

(31)

As can be retrieved from the table, the questionnaire was mostly filled in by Dutch women in the age of 21-40 years with an Academic education level.

Even though almost 98% of the participants were Dutch, their familiarity with the Supermarket Wine handbook and the Omfietswijn-logo is lower than expected, see table 3.

Familiar Frequency Percentage

Yes 91 57,2%

No 68 42,8%

Please see appendix 2 for the associated graphs.

4.1.3 Representativeness

Of the 159 respondents, 25.8% indicated that they do not like red wine. This could bias the results of purchase intention. Nevertheless, both personal relevance scenarios stated not specifically that the wine was for own consumption. So even though people did not like red wine, they could still buy it as a gift (scenario 2 - parents in law -). Therefore the answers could still be representative, and are included in the sample. However, I used it as a control variable in the regression.

Furthermore, I wanted to check whether there is no significant difference between age and gender of the participants in the four conditions. I performed an Univariate ANOVA on age to check for significance, this was not possible for gender because this is a dummy variable. Age (F (3, 87) = .173, P >.05, Sig. = .915) is not significant, indicating that there is no significant age difference between the participants groups of the four conditions and I do not need to control for this. Besides, I checked whether NFU is significantly different between the four conditions. Even though this personality trait is not possible to manipulate, I still wanted to check for validity reasons if there is no significant difference among respondents in the conditions. There is no significant difference (F (47, 87) = 1.317, p >.05, Sig.= .134) between the four conditions. Please see appendix 3 for the all the significant figures.

Because gender could not be checked on significance with Univariate ANOVA, I also calculated frequencies per condition. An overview of the sample characteristics per condition can be found below in table 4. Furthermore, appendix 4 depict an graphical overview of the sample characteristics per condition. As can be retrieved from table 4, the distribution is somewhat skew. Therefore I control for gender in the regression.

Table 3: Familiarity with Supermarket Wine handbook and Omfietswijn-logo

PURCHASE INTENTION

C AN RECOMMENDATIONS IN ADVERTISEMENTS BACKFIRE ?

T HE EFFECT OF

RECOMMENDATIONS ON

PURCHASE INTENTION

The effect of

recommendations on purchase intention

Can recommendations in advertisements backfire?

Management summary

Preface

Table of contents

List of tables

List of figures

List of abbreviations

1. Introduction

2. Literature review

2.1 Recommendations

2.2 Purchase intention

2.3 Influence of recommendations on purchase intention

2.4 Need for uniqueness

2.5 Personal relevance of the product.

2.6 Conceptual model

3. Methodology

3.1 Research design

3.2 Measures of the major variables

3.3 Manipulation check and control variables

4. Results

4.1 Sample characteristics