The Influence of User Generated Content

1

The Influence of User Generated Content

University of Groningen

Faculty of Economics and Business MSc Marketing Management Master Thesis

8 July 2013

Supervisor: prof. dr. L.M. Sloot 2nd _{supervisor: dr. J.E.M. van Nierop}

2

Management Summary

This research is directed toward assessing the effects of user generated content. In a world where consumers increasingly buy products online, consumer reviews can be an essential information resource. This might lead to consumers’ purchase intentions and product attitudes being strongly influenced by online consumer reviews. The goal of this research is to find out how online consumer reviews influence a consumer’s purchase intention and attitude toward the product, and therefore, if companies should encourage online consumer reviews or not. The main effect of online consumer reviews on purchase intention and product attitude is believed to be positively moderated by the type of good being utilitarian (functional products) rather than hedonic (fun products). Furthermore, it is believed that the type of website that shows the product and consumer reviews also has a positive moderating effect – reviews on an independent website (in this research ‘De Consumentenbond’) are more trusted by consumers, leading to a higher impact of these reviews compared to the website being dependent (retailer website). Finally, the level of consumer product knowledge is believed to be negatively moderating the effect of the level of positivity of online consumer reviews on purchase intention and product attitude – when a consumer has little knowledge of a product, he is more likely to rely on consumer reviews. This leads to the following research question:

What is the effect of online reviews on purchase intention and attitude toward the product, and to what extent is this relationship moderated by type of website, a consumer’s product knowledge and the type of good?

3

level of positivity of online consumer reviews on purchase intention is partly mediated by attitude toward the product.

4

Preface

The master thesis that lies before you is the final product of the four years I spent as a student at the University of Groningen. In 2009, I required little time to decide that I was going to follow the Bachelor of Science Business Administration. However, at that time I did not yet know what career I really wanted to pursue. In the three years that followed, the field of marketing grabbed my interest. I started my own small business combining my passion for graphical design with my interest in online marketing, and after obtaining my Bachelor’s degree, I required even less time to decide that I wanted to try and obtain my master’s degree in marketing.

Soon, I knew I made the right decision choosing for Marketing Management. Although I favoured some courses over others, overall I was interested in and liked every course I took. The thesis topic I chose is aligned with my interest in online marketing and my work as a small business owner, resulting in this end product. I greatly enjoyed my time as a student at the University of Groningen, however, I also feel now is the right time for a new challenge.

I would like to take this opportunity to thank some people that helped me along the road of writing my thesis. First and foremost, I would like to thank my supervisor prof. dr. Laurens Sloot. His guidance, support, and feedback throughout this research proved to be of great help. I would also like to thank my fellow thesis group members for providing feedback and help during this process. Finally, I wish to thank my closest family and friends for their feedback when needed and for their support during my studies.

5

Table of content

2.1 Online word of mouth ... 11

2.2 Online consumer reviews ... 12

2.2.1 Negative online reviews ... 13

2.2.2 Prior research on online consumer reviews ... 14

2.3 Predictors of online buying behaviour ... 16

2.3.1 Purchase intention... 16

2.3.2 Attitude toward the product ... 17

2.4 Type of website ... 18

2.5 Type of good ... 19

2.6 Consumer’s product knowledge ... 22

2.6.1 Findings on experience goods ... 22

2.6.2 Findings on product and consumer experience ... 23

3 Hypotheses and conceptual model ... 25

4 Methodology ... 28

4.1 Research design ... 28

4.2 Questionnaire and scenarios ... 30

4.2.1 Level of positivity ... 31

4.2.2 Type of website ... 31

4.3 Measurement of concepts ... 32

4.3.1 Measuring purchase intention ... 32

4.3.2 Measuring attitude toward the product ... 32

4.3.3 Measuring product (class) knowledge... 33

4.3.4 Control variables ... 34

4.4 Plan of analysis ... 37

4.4.1 Analysis of demographics ... 37

6

4.4.3 Testing conceptual model and alternative model ... 38

5 Results ... 40

5.1 Descriptive analysis ... 40

5.2 Reliability analysis... 43

5.3 Test of normality ... 44

5.4 Knowledge and involvement per product ... 45

5.5 Basic analysis ... 46

5.6 Testing hypotheses ... 47

5.6.1 Main effect level of positivity... 47

5.6.2 Moderating effect type of website... 49

5.6.3 Moderating effect type of good ... 50

5.6.4 Moderating effect consumer knowledge ... 51

5.6.5 Effects of demographic and other control variables ... 53

5.7 Testing alternative model with mediating effect... 54

6 Discussion... 57

6.1 Effect of main independent variable ... 58

6.2 Effects of moderating variables ... 59

6.3 Effects of other variables ... 61

7 Managerial implications ... 63

8 Limitations and further research ... 65

Literature ... 67

Appendix 1: Results pre-test ... 72

Appendix 2: Questionnaire ... 73

Appendix 3: Reliability analysis ... 87

Appendix 4: Kurtosis and Skewness statistics ... 88

Appendix 5: Results good types ... 89

Appendix 6: Results ANOVA purchase intention ... 90

Appendix 7: Results ANOVA attitude toward product ... 92

Appendix 8: Results regression purchase intention ... 94

7

1 Introduction

The introduction of the Internet and its worldwide use has not only led to a lot of new opportunities for companies, but also for consumers. Where firms are now able to build an online presence and sell their products to consumers, for example via online shops, consumers gained the opportunity to publish so-called user generated content on the web themselves. One of the ways in which this occurs is via online consumer reviews. This relatively new phenomenon creates new challenges and yet unanswered questions for firms.

Nowadays, people are frequently using online reviews before buying products or services either online or offline and online reviews are becoming more popular and important as an independent product-information resource (Chen and Xie, 2008). A recent market study of ChannelAdvisor (2010) shows that only 8% of online consumers does not access online reviews and comments before making a purchase decision. Furthermore, eMarketer (2009) expects that in 2013, nearly 155 million US web users will use some form of user generated content, from almost 116 million users in 2008, indicating that the popularity of online reviews will keep on rising in the near future.

The explanation of the sudden popularity of online consumer reviews is related to consumer trust. The presence of online consumer reviews enables people to read opinions of peers as opposed to the traditional product information that is provided by marketers. Research by Godes and Mayzlin (2004) shows that consumers are weary of traditional information channels such as television and radio, since they are dominated by these marketers. Moreover, research by eMarketer (2010) shows that online consumers put more trust in opinions of peers (for example in online consumer reviews) than in sources initiated by marketers. This has awaken the attention of online retailers who recognize that a consumer review can have serious impact on a consumer’s purchase decision and thus on sales.

8

product before buying, while in a traditional store, consumers can actually feel, smell and sometimes even test the products, such as with clothing. Marketers recognized that online reviews can overcome these shortcomings by providing an indirect experience of the product the consumer intends to buy. However, online stores are not the only source of online consumer reviews, since there are also independent websites such as Consumentenbond.nl, Kieskeurig.nl and Tweakers.net in The Netherlands, that enable users to publish reviews and comments on products they bought so that other consumers may benefit from them in their purchase decision. This adds another layer of complexity for online retailers, since it is more difficult to manage and to a certain extent control such user generated content.

The finding of eMarketer (2010) that consumers put more trust in consumer opinions than in opinions of the provider leads to the assumption that online reviews should have an impact on sales. In the past few years, extensive research has been subject to the question whether the use of online reviews leads to an actual payoff in terms of sales. Surprisingly, the outcomes are mixed. For example, Chevalier and Mayzlin (2006) found that customer purchase behaviour is affected by online amateur book ratings. Moreover, more recent research by Zhu and Zhang (2010) shows a “differential impact of consumer reviews across products in the same product category”, while Park, Lee and Han (2007) find a positive relationship between online consumer review quality and quantity on consumer purchasing intention. However, Duan, Gu, and Whinston (2008) researched the effect of online reviews with movie review data of 2003-2004. Their key finding was that box office revenues of movies were not significantly affected by online user review ratings. Furthermore, Chen, Wu and Yoon (2004) used sales data of books of Amazon.com over 2003 and found that consumer ratings are not correlated with sales.

Until now, research has primarily focused on actual sales data such as conversion rates or a behavioural consumer response such as purchase intention. Although actual sales data are most reliable, due to the difficulty of obtaining a large and representative set of sales data, some researchers rather use purchase intention data which is easier to obtain. The second variable that is used in this research is a consumer’s attitude toward the product. Knowing how a consumer’s product attitude is influenced by online consumer reviews can further enhance the marketer’s understanding of the effect online reviews have on a consumer’s thinking and actions.

9

variables are proposed in this research that might moderate the relationship between the use of online consumer reviews and purchase intention and attitude toward the product.

1. Type of website. Are the reviews posted on an independent website such as Tweakers.net and Kieskeurig.nl, or are they posted on a dependent website (in an actual online store). It is expected that consumer’s put less trust in a dependent website, resulting in a higher effect of online consumer reviews on an independent website. 2. Type of product. A distinction is made between utilitarian and hedonic products.

3. A consumer’s product knowledge. To what extent does the consumer have experience with the product. For example, does the consumer have extensive knowledge of the product category. Did she own such a product before, etcetera. It is expected that the higher the consumer’s product expertise, the less he or she would rely on reviews, and therefore the lower the effect of online consumer reviews would be.

Furthermore, the proposed product category for this research is consumer electronics. Since recent research has primarily focused on movies (U.S. Box Office results) and books (Amazon book sales), this research will focus on consumer electronics. This product category is frequently used for online consumer reviews and relevant because of its relatively large size as an online shopping category.

This leads to the following research question:

What is the effect of online reviews on purchase intention and attitude toward the product, and to what extent is this relationship moderated by type of website, a consumer’s product knowledge and the type of good?

The goal of this research is to find out how online consumer reviews influence a consumer’s purchase intention and attitude toward the product, and therefore, if companies should encourage online consumer reviews or not. The conclusions of this research will lead to relevant implications for managers. This research contributes to current literature by further enhancing academics’ knowledge of online consumer reviews and its effect on consumers and sales. Further, the inclusion of the three proposed moderators will lead to improved knowledge of the settings in which the use of online consumer reviews leads to an improvement in sales.

10

11

2 Literature review

To be able to research the effects of online consumer reviews on online purchase behaviour, a good theoretical understanding of the variables involved is needed. First, online word of mouth will be discussed since online reviews stem from word of mouth. For example, Godes and Mayzlin (2004) state that it is cost-effective and easy to measure word-of-mouth using online conversations. Subsequently, online consumer reviews and research on this subject will be discussed in detail.

In the third section, a theoretical foundation for the selected online purchase variables will be provided, followed by a typology of the websites that can be used for showing online consumer reviews (dependent or independent website) and its possible moderating effect on the influence of online consumer reviews on online purchase behaviour. The fifth and sixth section will continue with a typology of hedonic and utilitarian goods and a discussion of consumer knowledge respectively.

2.1 Online word of mouth

Besides an informing role, an online consumer review also has a recommending function. People who have already bought the product act as recommenders through their online consumer review. This recommending function can be seen as electronic word of mouth, also referred to as eWOM, which stems from word of mouth. Liu (2006) defines word of mouth as “informal communication among consumers about products and services.” It is important to note the words ‘informal’ and ‘among consumers’ in this definition. This clearly shows that word of mouth is seen as communication by consumers and for consumers, without the company itself playing a role in the communication process.

12

an active marketing tool, the recommending function of online consumer reviews proves to have a strong impact.

Woodside and Delozier (1976) state that word of mouth can help a consumer to make risky purchase decisions, such as the decision to buy a recently introduced product, more easily. They argue that for this effect to take place, consumers should first be informed of the benefits of the newly introduced product, for example by product attribute information provided by the seller. Then, in the case of word of mouth, consumers should have the ability to decrease the perceived risk by using recommendations of family and friends. Since the introduction of the Internet, the role of word of mouth behaviour has quickly changed. Where traditional word of mouth behaviour only extended to people connected to the information sharer, nowadays, consumers can share their opinions and thoughts not only with friends and family, but actually with every other Internet user in the world. This has greatly extended the range of word of mouth (Burton and Khammash, 2010), and therefore possibly the impact of online consumer reviews. Now, consumers can decrease the risk by assessing product reviews of other consumers.

2.2 Online consumer reviews

Usually, online retailers incorporate seller generated content, product information provided by the seller, and user generated content, in the form of online consumer reviews, on their product pages. Content created by the seller usually specifies the product and its attributes in more technical terms and uses technical standards to indicate the performance of the product. On the other hand, user generated content (e.g., an online consumer review) is oriented on the user and therefore provides attribute information from the situation of a user who actually uses or used the product in the past (Bickart and Schindler, 2001).

13

Another difference worth mentioning is that user generated content can range from providing very subjective, emotionally written product information related to a consumer’s usage situation, to very objective product information, similar to seller generated content. However, seller generated content is usually standardized and written in a standard form for every product the online retailer offers. Furthermore, the text length can greatly differ with user generated content, whereas seller generated content usually has a standard length, dependent on the standardized form that is used (Pan and Zhang, 2011).

Based on the previously discussed academic literature concerning word of mouth and online consumer reviews, it can be argued that a consumer review can help a consumer to make a better purchase decision by informing him more thoroughly about the product’s attributes and usage situations. Unsurprisingly, whether online consumer reviews actually have an effect on purchasing has been an extensively addressed topic in recent literature.

2.2.1 Negative online reviews

Clearly, online consumer reviews do not always provide positive product information, but can also be negative and therefore act more as a product discourager rather than a product recommender. Recently, research has focused on this negativity effect (or negativity bias) in a consumer’s purchase decision.

The negativity bias implies that a consumer attaches more value to negative information than to positive information in the process of evaluating a product (Ahluwalia, 2000; Herr et al., 1991; Maheswaran and Meyers-Levy, 1990). Ahluwalia (2000) analysed the negativity effect more thoroughly and states that the negativity bias consists of two elements. The first element is that a consumer pays more attention to negative information than to positive information, since it is more noticeable. The second element is trust-related. A consumer is said to put more trust in negative information, since negative information describes the product more symptomatic and in a relevant manner.

14

Therefore, one could argue that for hedonic products a positivity bias is present. Sen and Lerman (2007) argue that this might be because consumers possess stronger feelings about the hedonic product on beforehand, and therefore are not influenced by it. Both positive and negative online reviews are researched in this paper. The negativity effect of utilitarian products will be discussed further below, in section 2.5.

2.2.2 Prior research on online consumer reviews

The importance of online reviews is recognized by scholars, and recently many studies are focused on this concept. To fully understand where current research stands on online consumer reviews, a literature overview is presented below. Since the body of research on user generated content is quite broad, this overview is limited to research executed from 2004 and later.

Study Method Key Findings

Senecal and Nantel (2004)

Generalized estimating equations

Online product recommendations have a positive effect on online choices. This effect is moderated by recommendation source and type of product (search vs. experience).

Lin, Luarn and Huang (2005)

Focus group interviews

Positive online reviews have a positive effect and negative online reviews have a negative effect on the consumer’s purchase intention.

Chevalier and Mayzlin (2006)

Differences-in-differences

Online consumer book ratings are positively related to a consumer’s purchasing behaviour. They find empirical evidence for a negativity effect.

Liu (2006) Multiple regression

Word of mouth information offers significant explanatory power for both aggregate and weekly box office revenue, primarily originating from the volume of WOM and not its valence. Dellarocas, Zhang and Awad (2007) Diffusion model

Online consumers’ movie ratings provide a good proxy of early box office sales.

Park, Lee and Han (2007)

Elaboration Likelihood Model

15

Duan, Gu, and Whinston (2008)

Simultaneous system

The rating of online reviews has no significant impact on movies’ box office revenues, however, the volume of online posting does have an impact.

Mudambi and Schuff (2010)

Tobit regression

Review extremity and review depth are positively related to perceived helpfulness of the online review. This relationship is moderated by product type (search vs. experience).

Zhu and Zhang (2010)

Differences-in-differences

Online reviews are more influential for less popular games and games whose players have greater Internet experience.

Table 1 Literature overview on online consumer reviews

Many studies used actual sales numbers to measure the effectiveness of user generated content. For example, Chevalier and Mayzlin (2006) researched the effect of online consumer reviews on online bookstores Amazon and Barnes and Noble. They found a positive impact of consumer reviews on sales figures and showed that negative reviews have a larger impact than positive reviews. Furthermore, Dellarocas, Zhang and Awad (2007) found a similar effect of online amateur movie ratings and Liu (2006) showed that word of mouth information, either positive or negative, has a significant impact on weekly and aggregate box office revenues, primarily originating from word of mouth volume.

16

reviews matter for less popular games, while previous research only found evidence for an impact of some of these aspects, such as the volume of reviews (Godes and Mayzlin, 2004). A returning variable that supposedly affects the impact of online reviews is the type of product (search versus experience goods). For example, Senecal and Nantel (2004) incorporated the type of product and recommendation source in their study and found a moderating effect of both variables on the relationship between online product reviews and online choices. Furthermore, Mudambi and Schuff (2010) also found this moderating effect of product type, although their research focused on the impact of online consumer reviews on perceived helpfulness of the review. They also found that review depth and review extremity influence this relationship. To conclude, although the current body of research on online product reviews is already quite broad, there are still further research avenues that require exploration.

2.3 Predictors of online buying behaviour

In this research, variables related to a consumer’s buying behaviour are central when assessing the effects of online consumer reviews: a consumer’s purchase intention and attitude toward the product. These variables are discussed below.

2.3.1 Purchase intention

Purchase intention is commonly used in academic literature to measure effectiveness and to predict purchase behaviour. Purchase intentions can be defined as “decisions to act, or psychological states which represent the individual’s perception to engage in a particular behaviour” (Fishbein and Ajzen, 1975). Others define purchase intention in a brand setting as “the likelihood of buying the brand or of switching to another brand” (Keller, 2013).

17

Based on previously described literature on online consumer reviews and online word of mouth, it is expected that the level of positivity of online consumer reviews is positively related to a consumer’s purchase intention. This results in the following hypothesis:

H1: The level of positivity of a consumer review is positively related to a consumer’s purchase

intention.

2.3.2 Attitude toward the product

Attitude is seen as one of the most central and distinctive concepts in social psychology. A frequently cited definition of attitude is of Fishbein and Ajzen (1975). They defined attitude as “a learned predisposition to respond in a consistently favourable or unfavourable manner with respect to a given object”. Furthermore, they argue that an attitude toward an object (for example a product) occurs when someone follows upon beliefs about the product.

Based on the definition by Fishbein and Ajzen (1975) it can be concluded that attitude is quite an important subject in a consumer’s product consideration process. Even more so, it can be argued that a consumer’s purchase decisions are very much based on the attitude he or she has toward that specific object. Fishbein and Ajzen (1975) have adopted such a perspective and state that attitude yields behavioural intentions and eventually leads to behaviour itself. Therefore, adopting Fishbein and Ajzen’s (1975) perspective leads to the assumption that a consumer’s purchase intention is strongly related to his or her attitude toward the product. Furthermore, it can be argued that attitude toward the product is mediating (either partly or fully) the relationship between the level of positivity of online consumer reviews and purchase intention. Expectations are that a consumer’s product attitude is positively influenced by the level of positivity of online consumer reviews. Hence, when the level of positivity is high, a positive attitude toward that product is expected. Correspondingly, when the level of positivity is low, a negative attitude is expected. This leads to the expectation that the level of positivity of online consumer reviews is positively related to a consumer’s attitude toward the product. This results in the following hypothesis:

H2: The level of positivity of a consumer review is positively related to a consumer’s attitude

18

2.4 Type of website

Online consumer reviews can be presented on different websites. Most straightforward is the presence of user generated content on the retailer’s own website. However, there are also more independent websites that present online consumer reviews. It is expected that the type of website (dependent or independent) is related to the impact of the online consumer review on online purchasing behaviour.

Based on previous research by Senecal and Nantel (2002) the following three classifications of websites can be distinguished:

 “Seller”: The retailer’s own website, which makes this source ‘dependent’.

 “Commercially linked third parties”: Web shops that allow comparing different offers.

 “Non-commercially linked third parties”: Totally independent websites that test and

review products and retailers.

This research is executed in The Netherlands, where the ‘Consumentenbond’ is one of the largest non-commercially linked third party websites. However, this website is also built like a comparison web shop, and therefore can be classified as a non-commercial comparison web shop. Therefore, this research will be limited to only two of the three classifications identified from previously discussed literature:

1. A retailer’s own website (dependent website); and

2. An independent comparison web shop (independent website).

It is expected that the nature of the website (dependent or independent) has an impact on the effect of online consumer reviews on online purchasing behaviour. More specifically, based on prior literature on search costs and search effort it is expected that the independent website decreases a consumer’s search costs by showing multiple product offerings from different sellers simultaneously. Therefore, a consumer would favour an independent website because it has a higher usefulness for the consumer (Alba et al., 1997; Bakos, 1997; Lynch and Ariely, 2000). Moreover, it is expected that a consumer is more inclined to perceive an independent website as a provider of objective product information compared to a dependent website. Therefore, a consumer should value user generated content on a non-commercial independent website more than on a retailer’s website (dependent).

19

principle as proposed by Kelley (1973). According to this theory, when a consumer suspects that a review is written because of personal or situational determinants, the reviewer will be marked as biased. It can be argued that for positive reviews on dependent websites, consumers would be more inclined to believe that the reviewer receives some form of reward for his or her contribution. Therefore, the review would have a smaller impact on the consumer in comparison with an equally positive review on an independent website.

Senecal and Nantel (2004) researched the effect of type of website in an online product recommendations setting. Although they did not find any prove for an impact of the type of website on online choices, it can be argued that this would be different for purchase intention and product attitude. Senecal and Nantel (2004) used a dichotomous variable to measure online choice; a consumer would either pick the product or not. However, it can be argued that when a consumer is faced with a choice to pick the product or not, it is possible that he would still pick it although he is in fact influenced by the type of website. Since this research uses purchase intention and product attitude as dependent variables, it is expected that, based on previously discussed literature, the type of website does have a measurable impact on the proposed variables concerning online purchase behaviour.

To summarize, it is expected that online consumer reviews have a larger impact on online purchase behaviour when displayed on an independent website than on a dependent website. This results in the following hypothesis:

H3a: The positive impact of online consumer reviews on purchase intention is larger when the

reviews are shown on an independent website than on a retailer website.

H3b: The positive impact of online consumer reviews on attitude toward the product is larger

when the reviews are shown on an independent website than on a retailer website.

2.5 Type of good

20

product offers. This is the distinction between hedonic and utilitarian goods, which is used in this research.

 Utilitarian goods can be characterized as goods with primarily utilitarian benefits,

which refer to the “functional, instrumental, and practical benefits of consumption offerings” (Batra and Ahtola, 1990; Chitturi, Raghunathan, and Mahajan, 2007; Dhar and Wertenbroch, 2000). A personal computer and a microwave are examples of utilitarian goods.

 On the other hand, hedonic goods provide hedonic benefits, referred to as “aesthetic, experiential and enjoyment-related benefits” (Batra and Ahtola, 1990; Chitturi, Raghunathan, and Mahajan, 2007; Dhar and Wertenbroch, 2000). A sports car and a luxury watch are examples of hedonic goods, since they provide pleasure, fun and excitement (Hirschman and Holbrook 1982; Strahilevitz and Myers 1998).

Whether the good can be classified as a hedonic good or a utilitarian good may have a moderating influence, since one could argue that consumers will attach more value to consumer reviews of a utilitarian good than of a hedonic good. This might be because utilitarian benefits are more objective and not so much related to personal opinions and feelings as hedonic benefits. Therefore, it might be that consumer reviews for a utilitarian product are more influential than reviews for a hedonic good. Moreover, one could argue that a consumer looking for a hedonic good is paying less attention to consumer reviews in general, because he is looking for a fun and enjoying product and is therefore less willing to use negative reviews in his search. Pan and Zhang (2011) suspected and found this moderating effect of the type of product on the relationship between review valence and consumer review helpfulness. They argue that reviews in the utilitarian product category “are likely to be factual and objective, reflective of the functionality-driven consumer experiences”, where reviews in the hedonic product category “are likely to be subjective and emotionally laden, reflective of the underlying consumption experiences that involve, to a great extent, personal taste.” Therefore, they argue that the moderating effect of review valence on helpfulness is positive, since people put less value to a negative review in the hedonic goods category because it is based on subjective criteria, while product information in the utilitarian category is more objective.

21

goods. When a consumer evaluates a review of a hedonic good, the motivations of the reviewer are more likely to be attributed to factors related to the reviewer rather than the product. However, when consumer reviews for a utilitarian product are negative, the negative aspects are attributed to the product, resulting in a higher impact of negative reviews (Sen and Lerman, 2007).

Although the previously described literature relates the product type only to a review’s helpfulness, it is expected that when a review is more helpful, its recommending function will be stronger as well, resulting in a higher purchase intention and attitude toward the product. Therefore, it is expected that the type of good has a positive moderating influence on online purchase behaviour, although this effect is based on review valence and not directly on the type of good. More specifically, in case of an hedonic good, reviews are written in a more subjective and non-factual manner, based on personal

feelings and therefore having a smaller impact compared to utilitarian goods, which lead to objective reviews reflective of the functionalities of the product. Furthermore, based on findings on the negativity effect by Sen and Lerman (2007) it is argued that negative online consumer reviews for utilitarian products have a larger impact than negative reviews for hedonic products. The interaction effect expected between level of positivity and product type is graphically shown in figure 1. The resulting hypotheses are the following:

H4a: The positive impact of online consumer reviews on purchase intention is larger for

utilitarian than for hedonic products.

H4b: The positive impact of online consumer reviews on a consumer’s attitude toward the

product is larger for utilitarian than for hedonic products.

H4c: Negative online consumer reviews for utilitarian products have a larger impact than

negative online consumer reviews for hedonic products.

Figure 1 Interaction effect between level of

22

2.6 Consumer’s product knowledge

In addition to the type of website and type of good, it is expected that a consumer’s knowledge plays a role in the relationship between online consumer reviews and online purchasing behaviour. More specifically, does the consumer have extensive knowledge of the product category or did he perhaps own a similar product before? If so, it is expected that the consumer would have more knowledge of the product’s attributes and therefore has a lower need to use and rely on consumer reviews. In contrast to the type of website and type of good, a consumer’s product knowledge will not act as an experimental variable.

A consumer’s product knowledge can be defined as “the memories and knowledge that are in people’s minds” (Brucks, 1985). Alba and Hutchinson (1987) proposed a framework for consumer knowledge and distinguished two components:

 Familiarity: can be defined as “the number of product-related experiences that have

been accumulated by the consumer.” Product related experiences include “exposure to advertising, interactions with salespersons, choice and decision making, purchasing, and product usage in various situations.”

 Expertise: can be defined as “the ability to perform product-related tasks successfully.”

These include “beliefs about the product attributes” and “decision rules for acting on those beliefs.”

According to Alba and Hutchinson (1987), product familiarity is positively related to consumer expertise, indicating that the more product related experiences a consumer has, the higher his expertise and thus his “ability to perform product-related tasks successfully.” Based on this framework, it can be argued that the higher a consumer’s product familiarity, the better his ability to evaluate a product by himself. Consequently, a consumer’s need to assess online reviews would be lower, since online reviews can be classified as (indirect) product-related experiences.

2.6.1 Findings on experience goods

23

product reviews can be related to a consumer’s product expertise, since with an experience good, every consumer’s product expertise would be relatively low, unless the consumer has had previously used a similar product or a product in a similar product category.

Senecal and Nantel’s (2004) findings point in such a direction. They show that for experience goods, consumer rely significantly more on recommendations in comparison with other product types. Hence, when a consumer is looking for a product he does not have experienced himself in the past, online consumer reviews play an important role. Based on these findings, it can be argued that when a consumer’s product experience is low, the impact of online consumer reviews is larger than when a consumer’s product experience is high. Until now, research has primarily focused on product type when assessing a consumer’s product expertise. However, it can be argued that even within the hedonic product category, a consumer’s product expertise may differ between customers, for example because of prior experience with a similar product or product category. More specifically, when a consumer wants to buy a music cd from Bob Marley, a consumer who has previously listened to another Bob Marley cd or to similar reggae artists would have a higher product expertise, therefore a lower need to assess online product reviews which might result in a lower impact of online reviews.

2.6.2 Findings on product and consumer experience

Zhao et al. (2013) incorporated the concept product knowledge in their research. They focus on the effect of online reviews on individual purchase behaviour of experience goods. They assume that “consumers are uncertain about the true product quality or the extent to which a product matches their preference or usage condition.” This leads to consumers trying to eliminate or reduce the uncertainty about product quality by reading online reviews and trying to learn from prior usage experiences of other customers. In addition to indirect experiences from others, consumers also use their own direct experiences from products in a similar product category in their decision making process. However, their most remarkable finding is that consumers learn more from indirect experiences from others than from their own experiences. Despite of this, consumers who have a high product expertise might still be less inclined to use online reviews, therefore relying more on their own usage experiences resulting in a lower impact of online consumer reviews.

24

information, since they know which product attributes are important for evaluating the product. In contrast, others found the opposite effect; high product expertise leads to more information search, since they can more easily learn new information and know what (attributes) to search for (Johnson and Russo, 1984; Brucks, 1985). Applied on online consumer reviews, it is expected that when a consumer has high product experience, the consumer is less inclined to rely on consumer reviews since high experience users use none or other, more objective product attribute information.

To conclude, it is expected that product experience has a moderating influence on the relationship between online consumer reviews and online purchasing behaviour. More specifically, it is expected that when a consumer has a high product experience, consumer reviews have a lower impact compared to when a consumer has high product experience. This results in the following hypotheses:

H5a: The higher a consumer’s product knowledge, the lower the impact of using online reviews

on purchase intention will be.

H5b: The higher a consumer’s product knowledge, the lower the impact of using online reviews

25

3 Hypotheses and conceptual model

The theoretical framework in the previous section provided a literature body for this research and subsequently several hypotheses.

The first two hypotheses relate to the main effect of online consumer reviews:

H1: The level of positivity of a consumer review is positively related to a consumer’s purchase

intention.

H2: The level of positivity of a consumer review is positively related to a consumer’s attitude

toward the product.

The second set of hypotheses relate to the expected moderating effect of the type of website. Based on previous literature it is expected that reviews on an independent website have a stronger impact on online purchase behaviour than a dependent website, resulting in the following hypotheses:

H3a: The positive impact of online consumer reviews on purchase intention is larger when the

reviews are shown on an independent website than on a retailer website.

H3b: The positive impact of online consumer reviews on attitude toward the product is larger

when the reviews are shown on an independent website than on a retailer website.

The third set of hypotheses relate to the expected moderating variable product type. It is expected that the impact of online reviews for a product that primarily satisfies utilitarian benefits (utilitarian good) on online purchase behaviour is stronger than for a product that primarily satisfies hedonic benefits (hedonic good). Furthermore, based on the existing body of research, it is expected that the negativity effect will be stronger for utilitarian than for hedonic goods.

H4a: The positive impact of online consumer reviews on purchase intention is larger for

utilitarian than for hedonic products.

H4b: The positive impact of online consumer reviews on a consumer’s attitude toward the

26

H4c: Negative online consumer reviews for utilitarian products have a larger impact than

negative online consumer reviews for hedonic products.

The last set of hypotheses relate to the consumer’s product knowledge. It is expected that a consumer’s product knowledge negatively moderates the effect of online consumer reviews on online purchase behaviour.

H5a: The higher a consumer’s product knowledge, the lower the positive impact of online

reviews on purchase intention will be.

H5b: The higher a consumer’s product knowledge, the lower the positive impact of online

reviews on attitude toward the product will be.

Based on the concepts described in the literature review and the resulting hypotheses described above, the following conceptual model can be presented (see figure 2). Following the previously described constructs, it is assumed that the level of positivity of an online consumer review is positively related to purchase intention and attitude toward the product (represented by H1 and H2). However, it is expected that this effect is moderated by the following variables:

 Whether the website is either dependent (marketer controlled) or independent (non-marketer controlled);

 Whether the type of good provides either hedonic or utilitarian benefits (hedonic versus utilitarian goods); and

 The level of consumer knowledge of the product.

27

Alternative model with mediating effect

In the theoretical framework it was concluded that attitude toward the product might have a mediating effect on purchase intention. For example, Fishbein and Ajzen (1975) claim that attitude toward an object is very much related to behavioural intentions. Following this reasoning, an alternative model can be designed which expects a mediating effect (either partly or fully) of attitude toward the product on purchase intention.

Figure 3 Alternative model with mediating effect

28

4 Methodology

The previous chapter summarized the hypotheses and concluded with a conceptual model. To test the conceptual model, an empirical research is executed. This chapter covers the methodology of this research. First, the research design and method of data collection are discussed, followed by questionnaire design and a description of scenarios. Subsequently, the measurement of concepts is discussed. This chapter concludes with a plan of analysis.

4.1 Research design

To provide an answer to the research question formulated in the introduction of this research, a quantitative research will be executed in the form of an experiment. More specifically, a 3 (level of positivity) x 2 (product type) x 2 (type of website) factorial design with a questionnaire is used. The questionnaire will be distributed among Dutch people aged 18 to 65 years old, since it is expected that this group is making decisions independently an can be accounted for those decisions. Furthermore, expectations are that this group has at least some knowledge of the consumer electronics product category and is known with shopping online.

The questionnaire will be posted online, to make the collection of data faster and more flexible than with a traditional questionnaire on paper. Malhotra (2006) states that an online questionnaire has several benefits, such as its flexibility, the speed of data collection and the fact that the data is digital. This leads to easier processing of the questionnaire data and results. Furthermore, respondents benefit from an online questionnaire in comparison with a traditional questionnaire. For example, respondents now have the convenience of finishing the questionnaire when they like. The web link to the online questionnaire is distributed via social media such as Facebook, personal mailing lists, and by asking others to forward it to colleagues, friends and other people they know.

29

and purchase intention. The second part of the questionnaire consists of questions regarding the respondent’s knowledge of and involvement with the products. The questionnaire is concluded by some questions concerning control variables.

Category and products

This research is focusing on the product category of consumer electronics. Until now, research on online consumer reviews has primarily focused on movies and books, since products in these categories are reviewed excessively online. However, nowadays consumer electronics are sold online as well, and this category is also popular with online reviewers. For example, Kieskeurig.nl, probably one of the largest Dutch comparison websites, is full of reviews for all types of products in the consumer electronics category.

In the literature framework, the type of product was thoroughly discussed. Two product types were identified: hedonic and utilitarian products. To be able to test whether or not the type of product acts as a moderator in the proposed conceptual model, at least one hedonic and one utilitarian good need to be selected. However, to improve the representativeness of the consumer electronics category, two hedonic and two utilitarian products are used in this research. To select two hedonic and two utilitarian products, a pre-test was executed. First, four products per product type were selected which were expected to be either hedonic or utilitarian and are approximately equally high priced:

 Utilitarian goods: Washing machine, vacuum cleaner, printer, computer screen.

 Hedonic goods: Smartphone, luxury HiFi audio set, SLR (Single Lens Reflex) camera,

electronic tablet.

The pre-test was based on work by Voss, Spangenberg and Grohmann (2003) who identified several dimensions that could be classified as either hedonic (primarily related to fun, enjoyment and excitement) or utilitarian (primarily related to the functionality and effectiveness of the product). In total, eight respondents were asked to rate the eight selected products on a semantic differential with these hedonic and utilitarian dimensions.

30

score of 28.125 on hedonic dimensions and 21.250 on utilitarian dimensions. The two goods most clearly classified as utilitarian goods are the washing machine and vacuum cleaner. The washing machine showed a score of 10.125 on hedonic dimensions and 32.375 on utilitarian dimensions. Furthermore, the vacuum cleaner showed a score of 9.875 on hedonic dimensions and 30.625 on utilitarian dimensions.

4.2 Questionnaire and scenarios

The questionnaire can be divided into three parts. First, the scenarios are introduced, each starting with a short introductory text to place the respondent in a setting, followed by the product page and questions regarding the respondent’s purchase intention and product attitude based on that scenario. Moreover, a question is added to measure the control variable attitude toward the price. Every scenario can differ in type of product, level of positivity of the online consumer review, and the type of website. Second, a section is dedicated to measure the respondent’s knowledge of the four different products. Furthermore, to be able to control for the expected effect of a consumer’s involvement in the stated product class, several questions will be used to measure how involved the consumer is with the product class. Finally, the consumer is asked to answer several questions related to demographics and other control variables. The full questionnaire can be found in Appendix 2.

The respondent is shown a scenario with a screenshot of a product page similar to a normal web shop. The name, price, some specifications by the manufacturer and an image of the product are shown. Moreover, a section is shown below the product’s specifications which states that “x consumers rated this product with:”, where ‘x’ represents the number of reviewers. Right below this sentence, the number of stars are shown representing the rating, which can be 2.5 stars, 3.5 stars of 4.5 stars (the level of positivity of the online consumer reviews). The number of consumers that reviewed the product is controlled for by using the number fifteen in every scenario. The price of the product is based on the average price of the product, as can be found on the comparison website ‘Tweakers.net’.

31

situation (type of website and level of positivity) varies per product. This leads to developing six different questionnaires. All six questionnaires contain four scenarios, one for every product. Furthermore, every questionnaire contains two scenarios with an independent website and two with a dependent website. Finally, all questionnaires feature at least one scenario for each level of positivity. Respondents are randomly assigned to one of the six questionnaires.

4.2.1 Level of positivity

The level of positivity of online consumer reviews is indicated by stars. The online consumer review score can range from half a star to 5 stars. In this research, three levels of positivity are used: 2.5 stars (low), 3.5 stars (moderate) and 4.5 stars (high). The lowest level of 2.5 stars is chosen since one of the formulated hypotheses concerning type of product (H4c) states that negative online consumer reviews are expected to have a larger impact for utilitarian products than for hedonic products. Therefore, 2.5 stars represent a negative online consumer review. Furthermore, the level of positivity of 3.5 stars is chosen since it is clearly positive and therefore able to test all other formulated hypotheses. However, 3.5 stars is probably still not that extremely convincing. Therefore, a third level of positivity of 4.5 stars is incorporated in this research. A level of positivity of 4.5 stars is very high, which should probably lead to a high purchase intention and positive product attitude. Furthermore, 4.5 stars is still believable by consumers, while a level of positivity of 5 stars might lead to scepticism.

4.2.2 Type of website

In the theoretical framework it was concluded that two types of website can be distinguished: a dependent (retailer website) and an independent type of website. This research uses ‘De Consumentenbond’ as independent website. ‘De Consumentenbond’ is known in The Netherlands as an institution which tries to provide consumers with honest and righteous information to help them in their purchase decisions. Therefore, based on the theoretical framework, it can be classified as a non-commercially linked third-party website. However, ‘De Consumentenbond’ also is a comparison web shop, which would classify it as a non-commercially linked third party comparison shop.

32

4.3 Measurement of concepts

This section will discuss the ways of measuring the constructs as formulated in the theoretical framework and conceptual model, based on existing research and earlier reported scales. Most variables are measured using a Likert scale, which has the advantage that analysis is easier because of the ability to interpret the data as interval. All Likert scales use seven items, since Malhotra (2006) states that this leads to easier understanding of the questions by respondents. Moreover, it reduces the chance that a consumer becomes confused (Malhotra, 2006).

4.3.1 Measuring purchase intention

Since purchase intention is such a frequently researched concept, theoretical work on scales measuring the construct is quite extensive. For example, a frequently used scale to measure purchase intention originates from work of Baker and Churchill (1977). They intended to measure a person’s conative attitude toward an advertisement. However, they actually measured a consumer’s behavioural intention toward a product. Since then, various versions of the scale have been reported, ranging from two to four items, frequently using seven point Likert-type items. For example, Neese and Taylor (1994), and Kilbourne, Painton and Ridley (1985) used versions of the scale and found Cronbach’s Alpha’s of 0.81 and 0.91 respectively. This indicates that the scale is reliable in measuring the construct purchase intention.

For this research, the originally created scale by Baker and Churchill (1977) is used, although the third item is slightly modified so it is more fitting for an online setting. The scale consists of three items, which should be rated on a 7-point Likert scale with 1 is ‘No, definitely not, and 7 is ‘Yes, definitely’.

1. Would you like to try this product?

2. Would you buy this product if you happened to see it in a (web)store? 3. Would you have the intention to buy this product?

4.3.2 Measuring attitude toward the product

33

of this research. Furthermore, high internal consistency was reported by Muehling, Laczniak, and Stoltman (1991) with a Cronbach’s Alpha of 0.93.

To conclude, the following scale is used to measure attitude toward the product. These items should be rated on a 7-point Likert scale:

1. Good/Bad

2. Favourable/Unfavourable 3. Positive/Negative

4.3.3 Measuring product (class) knowledge

In the theoretical framework, it was concluded that product knowledge consists of two components (Alba and Hutchinson, 1987): Familiarity and expertise. Therefore, a reliable scale is needed which measures product class knowledge correctly and addresses both the familiarity and expertise component. Several scales for measuring product knowledge were developed in the past, such as the three item scale by Park, Mothersbaugh, and Feick (1994). Although this scale showed a Cronbach’s Alpha of 0.91 and is therefore sufficiently reliable in measuring the construct, the questions do not clearly distinguish between familiarity and expertise.

Beatty and Talpade (1994) also reported a scale to measure product class knowledge. It measures product class knowledge on three five-point Likert-type items. The scale supposedly originates from work of Beatty and Smith (1987), although that scale only reports three items instead of four. Beatty and Talpade (1994) reported a Cronbach’s Alpha of 0.86, which indicates that the scale is sufficiently reliable. Moreover, this scale clearly distinguishes product experience and product familiarity as two separate components of product class knowledge. Although Beatty and Talpade (1994) reported a slightly lower Cronbach’s Alpha than Park, Mothersbaugh and Feick (1994), the clear distinction between product experience and familiarity leads to favouring Beatty and Talpade’s scale for this research.

Beatty and Talpade (1994) measured product knowledge after the purchase was executed. However, this research measures a consumer’s product knowledge before a purchase has been executed, leading to a slight alteration of the items formulated by Beatty and Talpade (1994). More specifically, where the items were originally formulated in past tense, the items used in this research are formulated in present tense. Furthermore, the scale is changed from a five point Likert-type to a seven-point Likert-type. This leads to the following scale:

34

2. As compared to the average person, I would say that I am highly knowledgeable about this product category.

3. I describe myself as being very familiar with this product category. To summarize, table 2 shows a how the different variables are measured.

Variable Measurement scale Level of

positivity (L)

Level of positivity (L) is an ordinal variable with three levels: L = low (2.5 stars), L = medium (3.5 stars) or L = high (4.5 stars), which varies per scenario.

Type of good (G)

Type of good (G) is a pre-determined binary variable with 0 = type of good is hedonic and 1 = type of good is utilitarian.

Type of website (W)

Type of website (W) is a pre-determined binary variable with 0 = type of website is dependent and 1 = type of website is independent.

Product knowledge (PK)

Product knowledge (PK) is measured by using a 7-point Likert scale with 3 items, from which variable PK will be created using the average of the 3 questions. Therefore, PK can be interpreted as an interval variable. A respondent’s product knowledge is asked for in every scenario, right after introducing the product.

Attitude toward the product (A)

Attitude toward the product (A) is measured by asking respondents to rate the product on a 7-point Likert type semantic differential with the items ‘good/bad’, ‘favourable/unfavourable’, and ‘positive/negative’ after the scenario is shown. Variable A will be created from the average of the questions. Since A is measured with a Likert scale, it can be interpreted as an interval variable.

Purchase intention (PI)

Purchase intention (PI) is measured after the scenario is shown by using a 7-point Likert scale with three items. Variable PI will be created from the average of the questions. Since PI is measured with a Likert scale, it can be interpreted as an interval variable.

Table 2 Summary of measurement scales of concepts

4.3.4 Control variables

35

following demographic questions are asked at the end of the questionnaire. It is expected that respondents are more committed at the end of the survey than at the beginning. Since some questions such as asking a respondent for his or her income can be deemed as intrusive, the demographic questions are placed at the end of the questionnaire.

 What is your gender?

This is a nominal variable with answer possibilities ‘Male’ and ‘Female’.

 What is your age?

Open question, resulting in interval data.

 What is your monthly net income?

Nominal variable with a 9-point scale. The answer possibilities are < €500, €501 - €1000, €1001 - €1500, €1501 - €2000, €2001 - €2500, €2501 - €3000, €3001 - €3500, €3501 - €4000, > €4000.

 What is your level of education?

Nominal variable with 7 answer possibilities: Lager middelbaar onderwijs (VMBO), Hoger middelbaar onderwijs (HAVO/VWO), MBO, HBO, WO and Other.

Besides questions regarding demographics, several questions are used to assess the experience the respondent has with Internet, purchasing products online and using online consumer reviews when purchasing products online. Hence, the following questions are used:

 How experienced would you name yourself in using Internet for finding product information?

This question can be answered on a 7-point Likert scale ranging from ‘very inexperienced’ to ‘very experienced’.

 How frequently do you purchase products online?

This question can be answered on a 7-point Likert scale ranging from ‘never’ to ‘very often’.

 How frequently do you use online consumer reviews in your online purchase decisions? This question can be answered on a 7-point Likert scale ranging from ‘never’ to ‘very often’.

Since ‘De Consumentenbond’ is used as the independent type of website in this research, respondents should be asked for their attitude towards ‘De Consumentenbond’ and whether they are a member of ‘De Consumentenbond’. It might be that consumers who have a membership have a more favourable attitude toward the independent type of website on beforehand. Therefore, the following questions are used:

36

Nominal variable which can be answered by clicking either ‘Yes’ or ‘No’.

 What is your attitude toward ‘De Consumentenbond’?

This question can be answered on a 7-point Likert scale, using three of the earlier identified attitude items on a semantic differential: ‘Like/Dislike’, ‘Good/Bad’ and ‘Positive/Negative’. Furthermore, a fourth item ‘Independent/Not independent’ is added.

Product involvement

A respondent’s product involvement with the proposed products might influence the effect of consumer reviews as well. For example, a consumer who is highly involved will probably put more effort into assessing the product’s pro’s en cons. However, a consumer who is not very involved might not care that much about the product category. Therefore, he might not be willing to put much effort into making the right purchase decisions. Previous research by Park, Lee and Han (2008) has shown that involvement can have a moderating impact on the effect of online reviews. Therefore, product involvement is incorporated in this research as a control variable.

To measure a consumer’s involvement a four-item scale by Beatty and Talpade (1994) is used. Although this scale is based on an earlier reported scale by Mittal and Lee (1988, 1989), Beatty and Talpade added a fourth item to the scale. The first three questions mainly focus on a consumer’s interest in a specified product category, while the fourth item is focusing more on personal relevance. The reported scale of Beatty and Talpade (1994) showed a Cronbach’s Alpha of 0.74, 0.80 and 0.93, indicating that the scale is reliable in measuring the construct involvement.

Although others such as Van Trijp, Hoyer and Inman (1996) proposed a three item scale measuring a consumer’s involvement in a specified product class, the reported Cronbach’s Alpha is lower (0.69) than Beatty and Talpade’s scale. Therefore, this research uses the four Likert-scale items proposed by Beatty and Talpade (1994). The fourth item is reversed, which leads to recoding the question when analysing the data:

1. In general I have a strong interest in this product category 2. This product category is very important to me

3. This product category matters a lot to me

37

Price attitude

Since the four products in this research all have different prices, one could argue that a consumer´s attitude toward the price has an influence on purchase intention and attitude toward the product as well. Although the price of the products is selected based on a large Dutch comparison website, different people might still have different attitudes toward these prices. Therefore, a consumer´s attitude toward the price is added as a control variable in this research. The respondent is asked to rate the selling price of the product on a 7-point Likert scale, ranging from 1 (Very low) to 7 (Very high).

4.4 Plan of analysis

Three parts can be distinguished for the plan of analysis. First, analysis of the demographics of the respondents is executed to describe the sample and to see whether the respondent’s characteristics correspond with the intended target group. A second analysis will focus on gaining some basic insights about the data, such as distribution and data plotting. The third and final part will focus on testing the hypotheses formulated in the theoretical framework and on testing the alternative mediating model.

4.4.1 Analysis of demographics

The first step in analysing the results of this research is describing the demographic characteristics of the respondents who participated in the survey. Since the questions regarding demographics are all nominal variables, respondents are divided into groups based on their answers. Therefore, descriptive analysis will provide insight into the demographic characteristics of the respondents by counting frequencies. The description of the sample will be analysed to see whether it corresponds with the intended target group. Furthermore, the sample will be compared to CBS statistics (Central Bureau of Statistics in The Netherlands) of the Dutch population. Based on this, a decision will be made whether or not to weight the data on one of the demographic variables.

4.4.2 Gaining basic insights about the data