RATE THIS STORE!

(1)

RATE THIS STORE!

The impact of e-WOM on consumers’ online

retailer choice

(2)

[2]

UNIVERSITY OF GRONINGEN

RATE THIS STORE!

The impact of e-WOM on consumers’ online

retailer choice

Hil-Chi Wai Oktober 2010

(3)

[3]

Management Summary

The purpose of this study was to determine if eWOM influences the choice of a webshop. Previous studies have had a focus on the impact of eWOM on product choice, but none of the studies determined if those results could be generalized to webshop choice. Previous studies found that trust is one of the most important factors in e-commerce. Considering the great amount of various webshops available to the general public, trust is especially important, as consumers cannot be expected to be familiar with all the webshops. This study investigates the impact of eWOM on the choice of a webshop. It has been hypothesized that negative eWOM about a webshop has a stronger impact on the purchase intention than positive eWOM. Additionally, it has been hypothesized that negative eWOM has a stronger impact on trust compared to positive eWOM. Furthermore, the difference in impact of eWOM about niche webshops and generic webshops on purchase intention and trust has been studied.

A 3x2 factorial design was employed to research the issues. eWOM (negative eWOM, positive eWOM and no eWOM) and the type of webshop (niche webshop and generic webshop) were the independent variables. The dependent variables were purchase intention and trust. In the resulting questionnaire, which has been filled in by 162 participants, the independent variables were manipulated.

It can be concluded that eWOM influences the purchase intention and also the trust in a webshop. The division of the influence of positive eWOM and negative eWOM is asymmetric. Compared to positive eWOM, negative eWOM has the strongest impact on both the purchase intention of a consumer as well as the trust in a webshop. The impact of eWOM does not differ per type of webshop. In other words, webshop reviews have the same effect on webshops, regardless of the type of webshop, so niche webshops and generic webshops are equally affected by webshop reviews. Furthermore, it has been found that trust in a webshop and purchase intention are correlated.

(4)

[4]

Preface

As you may already have noticed, this is my Master thesis, written during my university education at the Rijksuniversiteit Groningen. I sincerely hope you will enjoy reading this thesis, and most importantly be able to gain some knowledge from it.

Before starting with my Master thesis, I thought that it should not be all that difficult to write a Master thesis. Even though I had heard from many people that it was a struggle, I thought that if I were to schedule everything, it would be a piece of cake. Well, I have experienced that it is not as easy as I thought it would be. Starting this thesis I thought that as long as I would have an interesting topic, it would be easy to research that certain topic. Looking back I can say it is partly true, but unfortunately not completely true. There were weeks in which I was totally overloaded with ideas and all those ideas could not fit within one framework. I have experienced that an interesting topic only made it more interesting to work on my Master thesis, but not easier to finish it. On the other hand, even with an interesting topic, it was not always interesting to read all the scientific articles which could add some value to my literature review. But after all, I am very glad with this end result.

I would like to thank my family and friends for their support and their encouragement. They made the time when I was working on my thesis more pleasant and were able to help me when I got overloaded with ideas again. Moreover, I would also like to thank all the 162 participants who filled in my questionnaire. At last but not at least, I would like to thank my supervisor Dr. Wander Jager for his support and patience. Thanks to his supervision, I was able to put my ideas in a study that was realizable.

Kind regards,

(5)

[5]

1. Introduction

1.1 Background problem

More and more customers make purchases via the Internet. A study concerning Dutch Internet-usage behavior, showed that between the year 2003 and the year 2005 an increasing number of consumers made online purchases (marketing-online.nl). Research shows that 71% of the Dutch consumers occasionally make online purchases. It is expected that this number will increase in the future (Starkenburg, 2010).

When little is known about a product and the risk of purchasing is experienced as high by the customer, customers often look for experiences of others, who have already purchased and used the product. It is not uncommon for customers to research how the product is rated by other customers and if the product meets the expectations of other customers. Many consumers use a comparator (e.g. dealtime.com and bizrate.com) to compare several products based on their characteristics and sales prices. But what if the webshop itself is unknown to the customer? With the arising number of webshops, the customer cannot be familiar with all the existing webshops. In the Netherlands only, there were 2,200 webshops in 2004, in 2007 there were 8,700 webshops and in 2009 there were over 20,000 webshops (John Kokkelmans, 2009). Based on the increasing number of webshops, it is difficult for consumers to know if the online company can be trusted or not. About 7% of the Dutch consumers worry about possible problems with online transactions (Starkenburg, 2010).

(9)

brick-[9]

and-mortar stores. The same can be said about the choice of the webshop. It can be risky for a consumer to make purchases at a certain webshop, because the company is unknown to the consumer. There is no physical store that they can visit in case of a problem. Several problems can arise by making online purchases, e.g. mistreatment of personal information or problems concerning shipment. The purchase judgment of consumers is often based on the information presented on the website (Park, Lee & Han, 2007). Therefore, consumers look on the Internet to gather more information. It is found that 25% of Internet users access online reviews before paying for a service delivered offline (ComScore 2007; cited by Zhu & Zhang 2010). Other surveys also show the importance of reviews before making a purchase (e.g. Simmons, 1994 and Riller, 1999; cited by Chen & Xie, 2005).

On the other hand, it is difficult for companies to maintain their reputation. Consumers can easily and interactively communicate about products, brands and companies before making a purchase. The search costs are low and there is almost no regard for distance or time. The Internet has opened up new options for consumers to gather unbiased product information from other consumers. Furthermore, the Internet also enables consumers to offer their own consumption-related advice by engaging in eWOM (Hennig-Thurau et al., 2004). Research shows that when consumers are not certain about purchasing a specific product, they are likely to use search mechanisms that provide value-added information (Koufaris, Kambil & LaBarbera, 2002). Consumers want to know which products suit them best and if the retailer is trustworthy. Every consumer with access to the Internet can find the same information, worldwide. Due to the Internet, rumour does not have to reach the news before the whole world knows about it. This means that it is important for a retailer to get positive eWOM. It is found that negative eWOM has a stronger impact on product choice, compared to positive eWOM (Park & Lee, 2009). This asymmetric impact is important for companies to take into consideration. Also the impact of eWOM can differ per type of webshop. With the arising number of webshops, many of them are not well-established. For example, a consumer can perceive it as risky to purchase at a small and unknown webshop, but does not perceive this risk when he or she wants to purchase at a well-established webshop. Because of that, depending on the type of a webshop, eWOM might have a different impact.

(10)

[10]

1.2 Problem statement

Based on previous findings, the following problem statement has been formulated:

What is the impact of negative and positive e-WOM on the choice of a webshop?

Other relevant findings that have been derived from this problem statement are discussed in the second chapter, the theoretical framework.

1.3 Research questions

To answer the problem statement, the following research questions are important:

 What is the impact of eWOM about a webshop on the purchase intention?

 Is there a difference in impact of negative and positive eWOM about a webshop on the purchase intention?

 Does the impact of eWOM differ per type of webshop on purchase intention and how does this differ?

 What is the impact of eWOM about a webshop on trust in a webshop?

 Is there a difference in impact of negative and positive eWOM about a webshop on trust in a webshop?

 Does the impact of eWOM differ per type of webshop on trust and how does this differ?

These questions will form the base of the theoretical framework, which will be discussed in the next two chapters.

1.4 Theoretical and social relevance

(11)

[11]

make or break the webshop, because this means that webshops with negative reviews can hardly attract new customers. It is important for companies to take this into account. Porter (2001) showed how the Internet influences the industry structure and how the bargaining power of customers increases. With the increasing popularity of the Internet as an information source, it is important to understand the possible impact of eWOM on the company.

1.5 Structure of the thesis

(12)

[12]

2. Theory and hypotheses

Results of previous researches will be discussed in this chapter. In the first part of this chapter, eWOM will be discussed, in the second chapter webshops will be discussed. The role of trust within this research will be discussed in the third part of this chapter. Finally, in the fourth part, the conceptual model of this study will be given.

2.1 Electronic Word-of-Mouth

The subject electronic Word-of-Mouth (eWOM) will be discussed in this section. First, the definition is specified and the difference between eWOM and traditional Word-of-Mouth (WOM) is explained. Secondly, the consumer intentions of providing eWOM will be discussed. And finally, the effects of eWOM and the difference in impact of positive and negative eWOM will be discussed as well in this section.

2.1.1 Definition electronic Word-of-Mouth

The basic idea behind Word-of-Mouth (WOM) is that information about products, brands, companies, services etcetera can spread from one consumer to another. WOM communication includes any information about a target object (e.g. product) transferred from one individual to another (Brown et al., 2005). WOM communication transferred on the Internet is called electronic Word-of-Mouth (eWOM). Hennig-Thurau et al. (2004) defined eWOM communication as ‘any positive or negative statement made by potential, actual, or former customers about a product, brand or company, which is made available to a multitude of people and institutions via the Internet’.

2.1.2 Difference between e-WOM and traditional WOM

(13)

[13]

had experiences with the product, brand or company on the Internet (Duhan et al., 2007; cited by Lee & Youn, 2009). eWOM also eliminated the restrictions on time and location. People can access the discussions whenever and wherever they want, which makes eWOM attractive to Internet users (Cheung et al., 2009). Different from traditional WOM, eWOM is measurable because comments are written and are available on the websites (Godes & Mayzlin, 2004). Some types of eWOM are controllable, because marketers can decide whether or not to allow consumer reviews to be shown. And if they are shown, marketers can offer a specific review format in order to guide the way of commenting. Thus, marketers can apply marketing strategies for eWOM more strategically than for traditional WOM (Park & Kim, 2008).

Gruen et al. (2006) state that eWOM effects on consumers may be very similar to traditional WOM effects and Hennig-Thurau (2004) found that eWOM participants exhibit a similar set of motivations as participants of traditional WOM. In that manner, some research in traditional WOM can be generalized to eWOM.

2.1.3 Consumer motivations to provide eWOM

As stated in section 2.1.2; difference between eWOM and traditional WOM, Hennig-Thurau (2004) found that eWOM participants exhibit a similar set of motivations as participants of traditional WOM. Thus, literature about traditional WOM intentions as well literature about eWOM intentions will be used in this section.

Hennig-Thurau (2004) segmented four groups with different eWOM intentions. These are:

 Self-interested helpers, who appear to be strongly driven by economic incentives. Economic incentives can be that the eWOM providers receive rewards for posting their opinions. E.g. Agoda.com, a hotel booking website asks their customers to write a review about the visited hotel, customers that do this are rewarded with points that can be redeemed for discounts on future bookings. This segment is the largest of the four segments.

 Multiple-motive consumers, who are motivated by a large number of factors. This segment has high scores on all factors, these are: platform assistance, venting negative feelings, concern for other consumers, extraversion/ positive self-enhancement, social benefits, economic incentives, helping the company and advice seeking.

(14)

[14]

give others the opportunity to buy the right product. The last motivation is based on products, but can be generalized to the right webshop.

 True altruists, who want to help consumers as well as companies. This is the second largest segment of the four segments and has the lowest score on economic incentives. These consumers provide eWOM because they are satisfied with a company and want to help the company to be successful, in addition they try to help other consumers to make the right decisions.

The contributions of true altruists and consumer advocates are small compared to the other two segments and have the lowest platform visiting frequency. Hennig-Thurau et al. (2004) shows that social benefits, economic incentives, concern for others, and extraversion/self-enhancement are the primary reasons for consumers to publish their opinions and experiences on web-based platforms. Based on these findings, platform operators can develop programs that appeal to the basic motives that drive eWOM behavior. Brown et al. (2010) found that satisfaction, commitment, and identification have a significant influence on positive WOM intentions and behaviors. These findings fit the best with the true altruist segment of Hennig-Thurau, as discussed above.

Contributors have positive WOM intentions when they are satisfied about a certain company and want that company to be successful. Brown et al. (2010) also found that the effect of satisfaction on positive WOM is partially mediated by consumer commitment to the retailer. The overall level of satisfaction has less influence on positive WOM when consumers have a high level of commitment. Thus, when the consumer is strongly committed to the company, the consumer will provide positive WOM even if the consumer has not experienced satisfaction with the retailer. Based on that finding, consumer commitment can be important for the choice of a webshop. This finding argues that commitment predominates satisfaction and commitment can be a reason for consumers to choose for a certain webshop before making a purchase and provide positive WOM.

2.1.4 Effects of eWOM

(15)

[15]

higher response elasticities. This is one of the reasons why word-of-mouth-marketing (WOMM) becomes more and more popular (Kozinets et al., 2010). WOMM is the intentional influencing of consumer-to-consumer communications by marketing techniques (Kozinets et al., 2010). Similar to traditional WOM, it is shown that eWOM might have higher credibility, relevance and empathy to customers than marketer-created sources of information on Internet (Bickart & Schindler, 2001; cited by Gruen et al., 2006). It can be concluded that eWOM is interesting for companies because of its effectiveness and for consumers as an information source. For this study, it is important to know what the effect of eWOM about a webshop is on the purchase intention.

It is found that purchase intention is a major consequence of pre-purchase satisfaction (Bai et al., 2008; cited by Chen, Hsu and Lin, 2010), which makes it more interesting to determine the impact of eWOM on the purchase intention of the consumer. Because it is expected that eWOM influences the pre-purchase satisfaction of consumers. In other words, it is expected that positive eWOM leads to a positive purchase intention. The desire of consumers to make a purchase in a webshop is reflected by online purchase intention (Chen, Hsu and Lin, 2010).

Also, much research has been done on the different impacts of positive and negative eWOM on purchase intentions. It has been found that the division is asymmetric. Park and Lee (2009) found that negative eWOM has a greater effect than positive eWOM. They also found that this effect is especially strong when the eWOM is about experience goods rather than about search goods. A reason for the difference between the effect of positive and negative eWOM, is that negative information is more attention grabbing in general and receives more detailed examination than positive information (Homer & Yoon, 1992; cited by Park & Lee, 2009).

Because of these findings, it is important to include the difference in effect of positive and negative eWOM in the conceptual model. If there are differences in the impact, the webshop can include these findings in their strategy. E.g. if the result will be that negative eWOM has a stronger impact on the purchase intention compared to positive eWOM, the company should try to avoid this.

(16)

[16]

H1a Positive eWOM about a webshop leads to a positive purchase intention and negative eWOM about a webshop leads to a negative purchase intention

H1b Compared to positive eWOM, negative eWOM has a stronger impact on purchase intention than positive eWOM

2.2 Webshops

The focus of this section is the subject webshops. First, relevant variables which should be considered when choosing a webshop will be discussed. The influence of the nature of products sold by the webshop will be considered in the second part. After that, the different webshop segments will be given and the use of reviews when choosing a webshop. Finally, the impact of eWOM per webshop segment will be discussed in the last part. In this section some main points will result in one or more hypotheses, which will be given at the end of this section.

2.2.1 Webshop choice

With the rising number of webshops, the consumer can choose from a growing number of webshops. According to Su (2008), consumers face three types of search costs which are:

 Costs of locating the right seller

 Costs of obtaining price information

 Costs of obtaining product information

Consumers are more willing to spend time on choosing the right webshop for the right product as long as the search costs are relatively low. Consumers are using the growing amount of digital information about shops and products to make the right purchase decision (Su, 2006). A recent study of IBM’s Institute for Business Value showed that more than 40 percent of the consumers will base their choice of retailer on digital information. At the same time, these consumers have higher expectations of the retailers.

(17)

[17]

unplanned store choice decision. In the Netherlands, an audience award is awarded to the best home shopping retailer of the year. This can also be a reason for consumers to purchase from this retailer and not from another retailer.

The focus of this study will lay on the influence of eWOM on the choice of a webshop. Other variables that can influence the purchase decision will not be taken into account.

2.2.2 Nature of products

The nature of the products sold by the online retailers can have an influence on the impact of eWOM. For some products the touch-and-feel aspect can be less important for a consumer, for example books and CD’s compared to shoes and clothes. The nature of products can be categorized in search goods and in experience goods. Search goods are defined as products that are characterized by product attributes where complete information about the goods can be acquired prior to purchase (Nelson, 1974; cited by Park & Lee, 2009). Experience goods are defined as products that are characterized by attributes that cannot be known until the purchase and after use of the product. Goods can also be labeled as experience goods when the search for information is more costly or difficult than direct product experience (Klein, 1998; cited by Park & Lee, 2009).

It is found that the type of product has an effect on which information sources consumers use when they are making consumer choices (Bearden & Etzel, 1982, Childers & Rao, 1992, King & Balasubramanian, 1994; cited by Senecal & Nantel, 2004). Recommendations for experience goods are found to have more influence than recommendations for search goods (Senecal & Nantel, 2004, Park & Lee, 2009). Based on this finding, it can be stated that webshops that only offer experience goods, are more easily affected by positive as well as negative eWOM, compared to webshops who only offer search goods. Thus, the impact of eWOM has a greater effect on experience goods than it has on search goods. This finding initiates that eWOM can have a different impact on various webshops based on the nature of products they sell.

2.2.3 Webshop segments

(18)

[18]

 Well-established webshops, e.g. Amazon.com, Dell.com and Bol.com. These are often webshops with a broad assortment, in breadth as well as in depth. Well-established webshops usually advertise online and offline.

 Generic webshops, e.g. Wilbro.nl. These are relatively small webshops with general merchandise. The product offering is not unique and mass-produced products dominate their assortments, usually their assortments are not as broad and deep as well-established webshops and generic webshops more often than not have an overall higher price level, because well established webshops often have a higher bargaining power and that generally leads to lower prices. An advantage of purchasing from generic webshops can be the more personal service.

 Niche webshops, webshops with a longtail focus (Anderson, 2006). Products that cannot be sold profitably in brick-and-mortar stores but can profitably be sold on Internet, e.g. onestopbuttonshop.com who offers many different buttons. The target group is relatively small compared to the other two types of webshops and has a deep assortment.

2.2.4 The use of reviews when choosing a webshop

The Internet marketing research company ComScore (2007; cited by Zhu & Zhang 2010), found that 25% of Internet users access online reviews before paying for a service delivered offline. Understanding how online reviews affect consumers’ purchase decision is important to firms that rely on eWOM to spread information about their products and their services. The impact of online consumer reviews on product sales depends on product and consumer characteristics (Zhu & Zhang, 2010). For this study, it is important to study what kind of impact the use of online reviews has. According to Zhu and Zhang (2010) the informational role of reviews becomes more salient in an environment where alternative means of information acquisition are relatively scarce. It is found by the same authors that consumers do not search for reviews at all, when they think that reviews are rare and difficult to find for less popular products. Because of that finding, it can be interesting to see if there is a difference in the use of recommendation source when purchasing at the different webshops.

2.2.5 Impact of eWOM per webshop segment

(19)

[19]

products toward a huge number of niches with the advent of the Internet. Consequently, the number of niche webshops and generic webshops is increasing. Zhu and Zhang (2010) found that online reviews are more influential when consumers have a relatively good amount of Internet experience, because of the relatively low search costs. These consumers also have a greater confidence in the use of Internet and this seems to dominate the concerns about the credibility of eWOM. The authors also found that for videogames, online reviews are more influential for less popular online games. This is an important finding for this research, because there is a possibility that this is the same case for webshops, that online reviews can have a stronger impact on less popular/niche webshops. Which means that eWOM can have a interaction effect with the type of webshop on the purchase intention. Amazon.com is a well-known webshop that operates worldwide and it is expected that eWOM has a small impact on the choice of these kinds of webshops and that eWOM is more relevant for the product choice. Presumably because the company already has a stable position in the market and a strong brand name. In this case consumers are less susceptible for eWOM concerning the Amazon webshop and are likely to have a stronger focus on product choice. Although there is the possibility that consumers can choose for another webshop because of price advantages or other reasons, it is assumed that trust in the webshop will not be strongly influenced by eWOM. Based on the growing number of niche and generic webshops, this study will have a focus on these types of webshops. Thus, the influence of eWOM on well-established webshops will be excluded.

The impact of eWOM on websites like eBay.com is considered as more important, because many sellers are clustered on those websites. Consequently, eWOM on the individual sellers can be important. But because this study is about webshops and individual sellers on eBay are not considered as webshops, these sellers will not be taken into account. E.g. a large part of the sellers on eBay are reselling their own (used) products and are not doing business in the way that companies do.

For now, the finding of a stronger impact of eWOM on niche games (Zhu & Zhang, 2010) can be generalized to other products besides games, but not yet to niche webshops, which is important for this study. The following hypotheses are formulated:

H2a Negative eWOM about a niche webshop has a stronger impact on purchase intention compared to negative eWOM about a generic webshop

(20)

[20]

2.3 The role of trust

One of the most frequently cited reasons for consumers not to purchase from webshops is the lack of trust (Lee & Turban, 2001). Trust is defined as ‘the perceived credibility and benevolence of a target trust’ (Doney & Cannon, 1997; cited by Bütner & Göritz, 2008). Trust is crucial in electronic commerce for turning visitors into customers. Jarvenpaa, Tractinsky and Vitale (2000) differentiated online shopping from offline shopping and suggested that trust is crucial for online transaction. Purchasing on the Internet involves more uncertainty and risk than traditional shopping, partly because webshops are not as well known to consumers (Lee & Turban, 2001).

In this study, two parts of trust play a role: consumer-to-consumer trust and business-to-consumer trust. Consumer-to-business-to-consumer trust is the credibility of eWOM and business-to-business-to-consumer trust the trust in the webshop. In the first part of this section, the difference of these two types of trust will be explained. After that, both types of trust will be discussed. At the end of this section, the hypotheses which belong to this subject will be given.

2.3.1 Difference between credibility of eWOM and trust in the webshop

As stated before, there is a difference between the credibility of eWOM and trust in the webshop. Credibility of eWOM relates to consumers assessing whether the source of eWOM is credible and true. Because it is shown that some companies manipulates its eWOM.

Trust in the webshop relates to whether the consumer thinks the webshop can uphold its promises. Credibility of eWOM and trust in a webshop can be related. For example when a consumer has never made a purchase at a certain webshop, and before making a purchase searches for reviews about the webshop. This consumer would presumably search for reviews about the webshop, to find out if the company can be trusted. But even if there are only positive reviews about the webshop, the consumer can still mistrust the reviews. If the consumer does not trust the reviews, this may have an effect on the trust in the company, because he or she can think that the company is manipulating the eWOM. On the other hand, when the consumer does trust the webshop, it is less likely that he or she will be influenced by eWOM or mistrust the eWOM.

2.3.2 Credibility of eWOM

(21)

[21]

increasing skepticism surrounding the motives behind eWOM. Studies show that firms not only regularly post their product information on online forums, but also proactively induce consumers to spread the word about their products online (Godes & Mayzlin, 2004). In addition, there are even companies who strategically manipulate online reviews in an effort to influence the purchase decision of the consumer (Dellarocas, 2006, Harmon, 2006; cited by Zhu & Zhang, 2010). Because of this, credibility is becoming a major concern for eWOM receivers (Cheung et al., 2009). If the consumer does not trust the review, it might lead to losing trust in the company, which will result in a purchase at another webshop. McKnight and Kacmar (2006) demonstrated that credibility is one of the most important antecedents of eWOM adoption. Information credibility is a vital predictor of an online consumer’s further action (McKnight & Kacmar, 2006). As argued before, consumers can decide not to choose for a certain webshop because they do not trust the review. Consumers have difficulty in using source similarity to determine the credibility of the information because the recommendations are typically from unknown users (Chatterjee 2001). As a result ,the consumer cannot know beforehand if the reviewer can be trusted or not. There is the possibility that the eWOM provider profits from a positive review, e.g. by receiving money or other rewards from the company. If the company really is manipulating the online reviews and this is known by the public, it can harm the company. Current customers can decide to purchase somewhere else, but it will especially cause the potential customers to choose another webshop.

2.3.3 When eWOM is seen as credible information by the consumer

(22)

[22]

service, rewards customers with points for discount by providing its experiences of a visited hotel. In this case, it is important that the customer get rewarded even if the review is negative for the hotel. Otherwise, by only rewarding consumers who give positive reviews, the reviews are indirectly influenced by the company. Apart from the involvement of consumers, it is pointed out that consumers who choose an unfamiliar retailer because of a lower price will seek more negative WOM information, and are more likely to believe that the problems may occur compared to consumers patronizing a firm they are familiar with (Chatterjee, 2001). Shamdasani et al. (2001) also state that informational factors (e.g., source, message, and receiver) are the major elements that affect a consumer’s information evaluation. These results have implications for customer service and positioning strategies of online retailers. Firms with a low price strategy are more susceptible to the effects of negative eWOM because, not only are consumers more likely to search for negative eWOM, it is also more likely they will review the negative eWOM as credible and likely to recur. Research shows this is especially true for pure online retailers compared to click-and-mortar firms (Chatterjee, 2001).

2.3.4 Types of recommendation websites

The nature of the website can influence the impact of a given recommendation. According to Senecal and Nantel (2002; cited by Senecal & Nantel, 2004), recommendation sources that can be used and promoted, can be categorized in three types of websites. These are:

 Sellers (e.g. Amazon.com)

 Commercially linked third parties (e.g. Bizrate.com)

 Non-commercially linked third parties (e.g. Consumerreports.org)

This categorization of recommendation sources is based on product reviews and is not based on webshop reviews, however, it can be generalized to webshop reviews. For example consumerreports.org is a website about brands as well as online retailers and bizrate.com rates the different online retailers. Often sellers have product reviews instead of reviews about the seller, but e.g. eBay provides ratings about the seller too.

(23)

[23]

possibility that consumers will choose the webshop that is recommended by others, which is the base of this research.

Consumers discredit recommendations from endorsers if they suspect that the eWOM provider has incentives to recommend the product. Senecal and Nantel (2004) found that the type of website did not have an effect on the perceived trustworthiness and did not influence consumer’s propensity to follow a product recommendation. This is an interesting result, because this means that the credibility of a review does not depend on the type of website. So the recommendation on a manufacturer’s site (seller) has the same influence as a recommendation on a non-commercially linked third party. On the other hand, Chen and Xie (2005) found that third-party product reviews have a significant effect on the success or failure of products. In their research, seller reviews are not taken into account. Both findings are related to product recommendations. It could be interesting to analyze if this result can be generalized to webshop choice as well.

2.3.5 Trust in the webshop

The perceived risks associated with online shopping by the consumer, influences attitudes toward online purchases (Koufaris, Kambil & LaBarbera, 2002). The main determinants of trust are previous purchases, the security and privacy policies of the webshop and service quality (Martin & Camarero, 2008, Mukherjee & Nath, 2007). Customers have to rely on the promises given by the online retailer when engaging in online transactions (Bütner & Göritz, 2008). Besides that, their personal information given to the company should also be handled in a correct manner, and should not be sold to third parties who use the information improperly, and also that the creditcard information is safeguarded (Jarvenpaa et al., 2000, Belanger et al., 2002, Garbarino & Strahilevitz, 2004; cited by Bütner & Göritz, 2008). A recent study of Networking4all (2010), an Internet security company, found that 25% of the webshops which are connected to the Dutch Home shopping Organisation are unsafe. 61% of the webshops can improve the security of the webshop. This shows that, even when a webshop is allowed to publish the Dutch Home shopping Organisation logo, buying in a webshop carrying that logo does not guarantee safe payment. Given the possible risks, customers want to know if the webshop is trustworthy or not before purchasing. Because of that, trustworthiness has been identified as a key driver for the success of electronic commerce (Schlosser et al, 2006).

2.3.6 Factors influencing trust in the webshop

(24)

[24]

Besides the visual appeal, Koufaris, Kambil and Labarbera (2002) found that customer’s product involvement can have a significant impact on their online experience, however, for new customers involvement is less important. Thus, the visual appeal can be more important for involved customers.

2.3.7 Impact of trust on purchase intention

Due to low barriers of entry and lack of gatekeepers on the Web, well-established and reputable websites have been more readily accepted by consumers than unknown sites (Shamdasani et al., 2001). This finding is in line with the findings of Kwon and Lennon (2009). These authors found that the existing offline brand image of a multi-channel retailer has a significant effect on the formation of its online brand image, perceived risk, and online customer loyalty intention. It is easier for a consumer to accept an existing retailer, than an unknown retailer that only serves consumers on the Web. The consumer does not know if the webshop can be trusted and can experience it as a risky purchase, because the online retailer is unknown. This is supported by the finding of Li, Browne and Wetherbe (2006), who argue that trust and commitment are the key mediating variables of stickiness with a certain website and that is in line with a finding of Chatterjee (2001). Chatterjee (2001) found that consumers, patronizing a familiar retailer, are less receptive to negative WOM information and seek less information. Thus, trust in a webshop can create customer loyalty. On the other hand, Kwon and Lennon (2009) found that online perceived risk has no significant effect on online customer loyalty. Consequently, even if the consumer experiences high risks by purchasing at a certain webshop, the customer can be loyal to this webshop. Also, actual online performance has a significant effect on online brand image, but the effect of websites is not strong enough to influence online perceived risk or loyalty of the consumer. Although the online brand image perceived by the consumer has a significant effect on perceived risk and online loyalty. Bütner and Göritz (2008) found that perceived trustworthiness has influence on the intention to buy and the financial risk to take. Thus, the more trustworthy an online retailer is perceived by a customer, the stronger the customer’s intention will be to buy from this retailer and the more likely the customer is willing to take a financial risk toward the retailer. For this study, it is important to research if online reviews have an impact on trust in the webshop and besides that, if the influence of eWOM on trust differ per type of webshop. Also the asymmetric division of the impact of positive and negative eWOM will be taken into account as well.

H3a Negative eWOM has a negative impact on trust in a webshop

(25)

[25]

H3c Negative eWOM about a niche webshop has a stronger impact on trust compared to negative eWOM about a generic webshop.

2.4 Conceptual model

The conceptual model is drawn based on the hypotheses. As can be seen in figure 1, the independent variables are: eWOM, positive, negative and no eWOM, and the type of webshop, generic and niche. The dependent variables are: purchase intention and trust.

(26)

[26]

3. Research design

The research design of this study will be given in this chapter. First, the research method will be discussed, after that, the sampling design and procedures will be discussed. Finally, the plan of analysis will be given. In appendix A, one can find the questionnaire.

3.1 Research method

For this study, it has been chosen to use a single cross sectional design, which means that only one sample of respondents is drawn from the target population, and from this sample, information is only obtained once (Malhotra, 2007). A survey is used as a method and will be provided electronically to gather quantitative data. The questionnaire will be put on Internet and the sample group will receive an email to fill in the questionnaire. Advantages of this method are the high speed, low costs and the perceived anonymity by the respondent. Also, with this method, the results will be free of potential interviewer bias, which is important when measuring the attitude of the consumer. On the other hand, the response rate could be low and the sample control is low to moderate (Malhotra, 2007).

As argued by Park and Lee (2009), the impact of positive and negative eWOM is asymmetric. To measure this difference in the purchase intention and trust, a control condition is included. By this, the difference in impact between positive and negative eWOM can be compared. There are two independent variables on different levels: eWOM and type of webshop. eWOM consists of three levels, which are positive, negative and no eWOM. The type of webshop consists of two levels, which are niche webshops and generic webshops. Thus, this is a 3x2 design. The two dependent variables are purchase intention and trust. In total there are six conditions. See table 1 for the factorial design of this study.

Niche webshop Generic webshop Negative eWOM Group 1 Group 4

Positive eWOM Group 2 Group 5

No eWOM Group 3 Group 6

TABLE 1FACTORIAL DESIGN

3.2 Sampling design and procedures

(27)

[27]

2007). There are six conditions, consequently, six groups are needed to fill in the questionnaire. The participants will be randomly assigned to one of the six conditions, with the same number of participants for each condition. Thus, the six conditions will be linked to one URL and when a participant clicks on the URL, the participant will automatically be assigned one of the six conditions. The minimum required number of respondents is 120 to make this design valid, with a minimum of 20 respondents for each condition (Rosnow and Rosenthal, 1991). Only Dutch consumers who purchase on the Internet will be taken into account in this research, it is important that the respondents make purchases on the Internet, otherwise the results will only be an indication of what consumers possibly do when they want to purchase on the Internet for the first time, and not what they usually do. Firstly, the consumers will be asked how many times per year they purchase on Internet, if this is zero, they will not be taken into account. By doing so, the results will be representative towards the target population. The questionnaire will be pre-tested by ten participants before the questionnaire will be sent to the other participants. This to be sure if every question in the questionnaire is clear and understandable.

All participants will receive an email with an URL to the questionnaire. This email is sent to 155 Dutch consumers, men and women in the age from 18 years to 55 years. In the email, people will be asked to fill in the questionnaire to help with a master thesis and they are also asked to forward the e-mail to their friends and family. Recipients who have not responded within one week time will be reminded to fill in the questionnaire.

3.3 Questionnaire design

(28)

[28]

3.4 Measurement

The manipulation of the independent variables will be discussed in this section. Also an overview of the questions in the questionnaire and to which variable the questions belong will be given in this section. In appendix A-2, one can find the questionnaire.

3.4.1 Manipulation

(29)

[29]

TABLE 2MANIPULATION

The participant will be asked about their attitude about the webshop based on the given review or information, to determine if the eWOM is actually perceived as negative, positive or neutral. The participant has to judge, to what extent they perceive the review as positive or negative, bad or good, and favorable or unfavorable on the 7-point Likert-scale. These items are implemented by multiple studies, e.g. by Holbrook and Batra (1987). Two questions are asked in addition to measure

Condition Manipulation Niche webshop: groups one,

two and three

You would like to purchase a product at the Internet and you can only find one webshop who offers the product. After looking around in the webshop, you search for reviews of previous customers. In the guestbook of the webshop you read the following:

Generic webshop: groups four, five and six

You are looking around in a webshop which is relatively small, but does have a wide assortment. At this webshop you find a product which you would like to purchase. After looking around in the webshop, you search for reviews of previous customers. In the guestbook of the webshop you read the following:

Negative eWOM, group one and group four

I’m not totally satisfied. It took long before I received my product and it was difficult to contact them about my order. I’m not sure if I will purchase at this store again.

Positive eWOM, group two and group five

I’m quite positive about this webshop. The order has been handled properly and I received the product quite soon. I will purchase at this webshop for sure in the future.

Niche webshop and no eWOM: group three (control condition)

Welcome to this webshop. You will possibly find some products in this webshop which you have not seen before, because this is not an average webshop, it is a specialty store who offers quality above quantity. Please look around and when you have any questions or remarks, don’t hesitate to contact us.

Generic webshop and no eWOM: group six (control condition)

(30)

[30]

the perceived information: ‘this information encourages me to purchase from this webshop’, and ‘this information recommends that I should purchase from this webshop’ (Kuan & Bock, 2007). Both questions are asked on a 7-point Likert-scale (totally agree – totally disagree). By this, it can be measured if the manipulation is successful or not.

As mentioned before, in total there are six conditions and two of them are the control conditions. With the control conditions it can be measured what the difference is in impact between positive and negative eWOM. The control conditions are also needed for the manipulation check. If the attitude towards the negative review is the same as the attitude towards the general information about the webshop, the manipulation is not successful. This means that no conclusions can be drawn, because the participant does not perceive the review as negative. The same counts for positive reviews.

3.4.2 Dependent measures

As can be seen in the conceptual model, figure 1, there are two dependent variables: purchase intention and trust. All the questions will be asked on a 7-point Likert scale. To measure the purchase intention of the participant after reading the review, several questions will be asked. First, it will be asked how likely the participant will start purchasing from the webshop in the future (Kuan & Bock, 2007). The second question is how likely one would continue using the webshop in the future (Li, Browne & Wetherbe, 2006). By asking this question, it can be measured if the review can also have a long-term impact or just impact on the purchase at that moment. One other question is asked to measure the purchase intention, which is: ‘I would like to increase my share of purchase from this webshop’ (Mukherjee & Nath, 2007). The fourth question is especially important for niche webshops: ‘I will look for a substitute product’ (7-point Likert-scale, very likely – not likely at all), because it is given that the product is unique and cannot be bought somewhere else. For generic webshops, the question: ‘an alternative webshop is attractive to me’ (Li, Browne & Wetherbe, 2006) on a 7-point Likert-scale, totally agree – totally disagree, will be more important. Because with this product the consumer has the possibility to purchase the product somewhere else.

(31)

[31]

thinks that the information provided on the webshop is true en the other question is till what extent one thinks that the information provided on the webshop is honest (Martin & Camerero, 2008).

Participants are asked if they always search for reviews about a webshop when they have the intention to purchase on a webshop, and if they always search for reviews about a webshop where they have not purchased before. Besides those two questions, the participants will also be asked when they search for reviews about a webshop, when they purchase at a well-established webshop, generic webshop or niche webshop. With these questions, the dependent variable ‘search for reviews’ can be measured.

3.5 Plan of analysis

To examine the reliability and validity of the measures, Cronbach’s alpha will be computed for all scales that will be combined (summated scale). This is needed to indicate the degree to which a set of items measures a single construct. This means that the individual items of the scale are all measuring the same construct and thus be highly correlated (Hair et al., 2010).

Cronbach’s alpha has to be at least 0.6 before it can be argued that the test scores are reliable (Malhotra, 2007). When Cronbach’s alpha is above 0.6, the single items will be combined to several variables, so the mean of these items can be used as a replacement variable. This is called summated scales (Hair et al., 2010). This provides the benefits that it reduces measurement error by using multiple variables to reduce the reliance on a single response and it represents multiple aspects of a concept in a single measure (Hair et al., 2010).

(32)

[32]

4. Results

Data gathered by the questionnaire will be analyzed and discussed in this chapter. The purpose of this study is to determine the impact of eWOM about a webshop on the choice of a webshop.

4.1 Participants

In total, there were 162 respondents. Around 132 respondents have answered all the questions. Each question in each condition is answered at least 20 times. It is not required to have an equal number of respondents in each condition (Malhotra, 2007) so all the respondents have been taken together. As argued in the previous chapter, only participants who usually make purchases on the Internet have been taken into account for the representativeness of this study. Although it might have been interesting to include participants who usually do not purchase on the Internet, the results would have been less relevant to webshop-owners. Also participants who only make purchases on websites like eBay.com and not in webshops have been excluded. In total, only 11 participants have been excluded for further analysis. Out of the 151 participants who make purchases in webshops, 124 respondents have filled in all the questions. The number of respondents per condition varies from 22 to 30. The number of participants per condition are described in table 3.

Condition Number of participants

Number of participants excluding non-buyers Negative eWOM- niche webshop 25 22

Positive eWOM- niche webshop 26 25

No eWOM- niche webshop 29 27

Negative eWOM- generic webshop 32 30

Positive eWOM- generic webshop 25 22

No eWOM- generic webshop 25 25

Total 162 151

TABLE 3NUMBERS OF PARTICIPANTS PER CONDITION

(33)

[33]

4.2 Missing variables

The first two questions about the frequency of purchases on the Internet have not always been filled in with a number, some respondents filled in ‘often’ or ‘sometimes’. These are marked as missing data, because respondents could perceive these terms differently. E.g. one respondent purchased about 100 times per year on the Internet, while the most respondents purchased between one to ten times per year. This large difference made it impossible to allocate what could be perceived as often. From the 162 respondents, 132 of them have filled in all the questions. The control condition without reviews about the generic webshop has the highest drop-out. From the total of 32 respondents in that questionnaire, some questions have only been filled in 21 times. Also one participant remarked that it was not possible for her to answer the statement about her attitude towards the webshop with the scale favorable - unfavorable. But the drop-out on that question is not higher compared to other questions. Although the possibility exists that the error in the questionnaire caused a part of the drop-out in the questionnaire. Every question which has not been filled in has been marked as a missing variable as well.

4.3 Recoded variables

Most of the questions were asked on a 7-point Likert scale, which were ranked from positive to negative. The results are recoded in the way that 1 shows the negative situation and 7 shows the positive situation. This means that for example a question is asked on a 7-point Likert scale from totally agree to totally disagree, has been recoded to totally disagree-totally agree. Thus, if the result is 1, it means that the participants totally disagree, while before recoding, 1 stands for totally agree. This is especially important when exact numbers of the results will be given.

The following variables are not recoded:

 I will look for a substitute product

 An alternative webshop is attractive to me

 I believe there is negligible risk that something may go wrong with using this website for purchase

 I never purchase at well-established webshops

 I never purchase at generic webshops

 I never purchase at niche webshops

 I never purchase at webshops where I have not purchased before

(34)

[34]

This is needed to have all the items measuring the same direction. E.g. the question ‘I will purchase at this webshop in the future’ is asked on a 7-point Likert scale (very likely-very unlikely). The question ‘I will look for a substitute product’ is asked on the same scale as well, but when someone marks unlikely for the first question, and likely for this question, the two questions cannot be taken together. They are both measuring the contrary, so recoding to the opposite is needed. Questions which have not been asked on a 7-point Likert scale, e.g. the purchases per year, age and income, have not been recoded.

4.4 Scale reliability

To measure the strength of each construct, Cronbrach’s alpha has been computed. Cronbach’s alpha had to be above 0.6, before it could be stated that the scale was reliable. Cronbach’s alpha has been calculated for every construct. All the conditions have been taken together to measure the strength of each construct. In table 4, one can see which questions in the questionnaire have been taken together to compute Cronbach’s alpha and it also shows the Cronbach’s alpha of the variables.

Variable Cronbach’s alpha Items Source Attitude towards webshop

0.924  What is your feeling towards this webshop based on the information given above? o Positive-negative o Bad-good o Favorable-unfavorable Holbrook and Batra, 1987 Attitude towards review

0.748  The review recommends me to purchase at this webshop

 The review encourages me to purchase at this webshop

Purchase intention

0.843  I like to place an order at this webshop directly

 I am likely to start purchasing from this webshop in the future (very likely-not likely at all)

 I intent to continue using the webshop in the future (very likely-not likely at all)

Kuan and Bock, 2007

(35)

[35]

 I would like to increase my share of purchase from this webshop(totally agree-totally disagree)

 I will look for a substitute product (very likely-not likely at all)

 An alternative webshop is attractive to me (totally agree-totally disagree)

Mukherjee and Nath, 2007

Li, Browne and Wetherbe, 2006

Trust webshop

0.652  I trust this webshp as a safe medium for transaction and purchase (very likely-not likely at all)

 I believe there is negligible risk that something may go wrong with using this website for purchase (very likely-not likely at all) Mukherjee and Nath, 2007 Mukherjee and Nath, 2007 Trust review

0.960  I think the information provided on this website is true (totally agree-totally disagree)

 I think the information provided on this website is honest (totally agree-totally disagree)

Martin and Camerero, 2008

TABLE 4SCALE ITEMS

As can be seen in table 4, all the constructs have a Cronbach’s alpha above 0.6, which means that the constructs are reliable. Thus, the means of the items have been taken together to create the new variables.

4.5 Manipulation check

Before the possible impact of eWOM on the purchase intention and trust could be measured, it was necessary to verify if the manipulation of the independent variables were actually perceived as positive, negative or neutral and the attitude towards the webshop after reading the review. Although the participants who pre-tested the questionnaire actually perceived the reviews are positive, negative or neutral, it could not be assumed that this would count for the actual sample group as well.

(36)

[36]

measured if the review was perceived as positive, negative or neutral. With the variable ‘attitude towards webshop’ it could be measured how the webshop was perceived after reading the review or general information. This was needed to determine if the attitude was consistent with the manipulation.

To measure the attitude, ANOVA has been conducted for both variables. The independent variable was eWOM and the dependent variable was in the first ANOVA ‘attitude towards webshop’. The result showed a significant effect (F (2, 128) =47.97, p<0.001). This means that the participants, who received the condition with a positive review about the webshop, indeed had a positive attitude towards the webshop. Participants, who had read a negative review, did have a negative attitude towards the review. In table 5, one can see the means of the attitude towards the webshop. As can be seen, the mean of the conditions who received some general information about the webshop, is between the mean of negative eWOM and positive eWOM.

eWOM Mean Std. Deviation Negative 2,57 1,010

Positive 5,00 1,428

No 4,56 1,185

TABLE 5ATTITUDES TOWARDS THE WEBSHOP

To measure if the review was perceived as positive, negative or neutral, another ANOVA is conducted. The dependent variable that has been used is ‘attitude towards review’ and the independent variable is ‘eWOM’. The result shows a significant effect (F (2,128) =29.12, p<0.001). Table 6 shows the means of the attitudes towards the review. It can be concluded that negative eWOM is indeed perceived as negative eWOM, positive eWOM is perceived as positive eWOM and the general information as no eWOM. The mean of the control condition is between the means of negative and positive eWOM. So, the general information is not considered as positive or negative.

eWOM Mean Std. Deviation Negative 2,25 0,951

Positive 4,07 1,408

No 3,94 1,282

TABLE 6ATTITUDES TOWARDS REVIEW

(37)

[37]

4.6 Descriptive results

Several questions were asked in the questionnaire to measure the usage of reviews and which recommendation source is often used. The participants have also been asked about their purchases on the Internet. The results will be presented in this section.

4.6.1 Reviews about a webshop

Zhu and Zhang (2010) found that consumers search less for reviews when they want to purchase a niche product. This because the consumer assumes that there will not be any information found on the Internet about the product. It is interesting to see if the finding for Zhu and Zhang can be generalized to webshops as well. In the questionnaire, the consumer has also been asked whether they would search for reviews. Figure 2 shows how often consumers search for reviews when purchasing at a well-established webshop, a generic webshop and a niche webshop, assuming that they have not purchased before in the certain webshop.

FIGURE 2SEARCH FOR REVIEWS TYPES OF WEBSHOP

(38)

[38]

negative, which means that they never-sometimes search for reviews when they want to purchase at a well-established webshop, and around 41 percent answers always-regularly. On the contrary, while purchasing at a niche webshop, 57.4 percent search for reviews and 22.8 percent does not or seldom search for reviews. With generic webshops, 50 percent of the participants search for reviews while 23.4 percent does not search for reviews. Thus, when consumers want to purchase at a niche webshop, they are more likely to search for reviews compared to purchasing at a well-established webshop. This means that the finding of Zhu and Zhang (2010) cannot be generalized to webshops.

FIGURE 3SEARCH FOR REVIEWS NEW WEBSHOPS

It is also tested how often the participants search for reviews when purchasing at a webshop where they have not purchased before and webshops where they have purchased before. Also for these variables, Chi-square statistic has been conducted. The results do not show differences (χ2= 7.509, df= 6, p= 0.276). This means that the division of the categories does not differ. When the participants have not purchased at a webshop before, consumers will not search for reviews more often (M=4.73, SD=1.68) than when the participants already have experience with the webshop (M=4.32, SD=1.80). It can be stated that even if the consumers already have experience with a certain webshop, reviews about the webshop can still be important (figure 3).

(39)

[39]

FIGURE 4IMPORTANCE REVIEWS

4.6.2 Use recommendation source

The participants have been asked how often they use the three types of recommendation sources. This has been done to analyze if some recommendation sources are perceived as more important than other recommendation sources. As given in chapter 2, the three recommendation sources are: sellers, commercially linked third parties and non-commercially linked third parties. To test if there are significant differences between the division of the categories, Chi-square statistic is used.

FIGURE 5USE OF RECOMMENDATION SOURCES

The results show significant differences when the use of non-commercially linked third parties recommendation source is compared to the use of sellers as a recommendation source (χ2=15.774, df=7, p<0.05). When comparing commercially linked third parties with sellers, the results are not significant on a p<0.05 level (χ2=12.145, df= 7, p=0.096). The same counts for commercially linked parties compared to non-commercially linked third parties (χ2=13.374, df=7, p= 0.063). This means that only the distribution of the use of commercially linked third parties differ with the use of sellers as a recommendation source. Guestbooks of sellers are used more often(M=4.22, SD=1.95) compared to commercially linked third parties which are used less (M=3.82, SD=2.15). Figure 5 shows the frequencies of the use of the three recommendation sources.

4.6.3 Purchase on webshops

(40)

[40]

well-established webshop’, ‘I never purchase at generic webshops’, and ‘I never purchase at niche webshops’. This was interesting to study, because there could have been many reasons why consumers prefer certain types of webshops. E.g. consumers can see it as a risky purchase when purchasing at a niche webshop which is unknown yet, or prefer a well-established webshop because of the ease of purchasing or lower prices. On the other hand consumers can also prefer generic webshops because of the personal treatment. To measure the significant difference, Chi-square statistic is used.

FIGURE 6 PURCHASE TYPE OF WEBSHOPS

It is shown that consumers more often purchase at well-established webshops (M= 5.95, SD=1.50), compared to generic webshops (M=4.90, SD=1.70) and niche webshops (M=4.61, SD=1.88). This can also be seen in figure 6, which shows the frequencies of purchase per type of webshop. The results show significant differences in the division of the categories. The division of the categories differ when the frequency of purchases at well-established webshops are compared with generic webshops (χ2=62.727, df=6, p<0.001), when purchases at niche webshops are compared with generic webshops (χ2=15.967, df=7, p<0.05), and the frequency of purchases also differ when purchases at niche webshops are compared with well-established webshops (χ2=63.530, df=6, p<0.001).

FIGURE 7 PURCHASE NEW WEBSHOPS

RATE THIS STORE!

RATE THIS STORE!

The impact of e-WOM on consumers’ online

retailer choice

RATE THIS STORE!

The impact of e-WOM on consumers’ online

retailer choice

Management Summary

Preface

Table of Contents

1. Introduction

1.1 Background problem

1.2 Problem statement

1.3 Research questions

1.4 Theoretical and social relevance

1.5 Structure of the thesis

2. Theory and hypotheses

2.1 Electronic Word-of-Mouth

2.2 Webshops

2.3 The role of trust

2.4 Conceptual model

3. Research design

3.1 Research method

3.2 Sampling design and procedures

3.3 Questionnaire design

3.4 Measurement

3.5 Plan of analysis

4. Results

4.1 Participants

4.2 Missing variables

4.3 Recoded variables

4.4 Scale reliability

4.5 Manipulation check

4.6 Descriptive results