Online reviews versus samples: the role of review valence, trailer and source credibility in the case of TV series

(1)

Online reviews versus samples: the role of review valence,

trailer and source credibility in the case of TV series

30

th

June, 2014

(2)

Master Thesis

Marketing Department University of Groningen

Online reviews versus samples: the role of review valence,

trailer and source credibility in the case of TV series

Student:

Andreana V. Gosteva

Josef Israelstraat 6, 9718GA, Groningen Student Number: 1753932

Email: a.v.gosteva@student.rug.nl

Supervisors: Dr. Jenny van Doorn (1st)

(3)

Preface

This thesis is written for completing my Master Degree in Marketing, at the University of Groningen. It marks a final chapter of a long journey but I hope it opens a way for new possibilities.

Along the way of my studies I have gained a lot of experiences but they would not have mattered without the wonderful people who will always remind me of the great chapter, called student life. I am very grateful to my thesis group members and Mrs. Van Doorn for all the help during the thesis writing. As Oscar Wilde has once said: “Success is a science; if you have the conditions, you get the result.” Therefore, I do hope that I and the rest of my thesis group members will have all the necessary conditions for the future to make the best of proving our skills and fulfill all of our ambitions in life.

Most importantly, however, I would like to thank my wonderful family and friends for the never-ending care and support. I do appreciate the effort of all my close people who immensely helped during the data collection process and contributed greatly to compiling my thesis.

(4)

4

Management Summary

As means of sharing information based on experience online reviews are a new way of distributing word-of-mouth through which consumers can give, consult or assign value to products or services (Shridhar and Srinivasan, 2012). Apart from reviews, the online environment also provides the option of sampling as a way of acquainting with a product. Sampling can assist in picturing the value of a product as well as in decreasing doubt levels among consumers. However, as customers face variability in reviews articulated by previous users’ or experts, the credibility of the review source has been challenged. Existing literature has mostly discussed the use of online consumer reviews in the pre-purchase stage of the buying; yet, less emphasis was placed on product sampling and the source of review (non-expert vs. expert) as factors influencing consumption.

The present study investigates the effects of review valence, sampling and source of credibility in the case of TV shows. Specifically, it examines whether offering an alternative to indirect product experiences (i.e. online reviews) such as sampling can help consumers improve their evaluations of a product. Additionally, as review valence can lead to higher uncertainty while sampling aims at mitigating uncertainty levels (Hu et al., 2009) the impact of providing both a sample and a review was researched. Furthermore, the effects on expressed interest in viewing a show depending on the source of the review (expert versus user) were measured. The model used is a 2(valence) x 2 (sample) x 2 (source of review) between-subjects design which has been tested by an online survey.

(5)

5

Among the key changes in modern consumer behavior has been the shift from a passive to an active and involved consumer. Nowadays, consumers are connected and communicate in various ways through social media channels like blogs, recommendation sites, online communities etc. (Hennig-Thurau et al., 2010). A distinctive perspective is now evolving, illustrating that customers are no longer exogenous to a firm but rather a part of its co-creation value, competitive strategy and innovation process. An essential concept in this context is customer engagement behavior (CEB), defined as the behavioral expression from a customer towards a brand or a firm which pushes beyond transactions (van Doorn et al., 2010). This way of interaction has transformed fundamentally the way consumers search and experience goods based on the ability to discover product quality before purchase.

1.1. Background Problem

Traditionally, the economics of information search models state that consumers seek information until the marginal cost of search equals its marginal benefit (Moorthy et al., 1997). The mass usage of the Internet, however, has given the opportunity for an easy, cheap and commonly used way to compare market goods and search for quality information in the form of electronic word-of-mouth (WOM) (Chen and Xie, 2009). WOM, as a channel for consumers to share information, is one of the most powerful communication means as it affects evaluations, attitudes, and, intentions to purchase products or services (Alreck and Settle, 1995). Online shoppers will not purchase a product if they are uncertain whether it suits their preferences or whether its quality is as good as marketed. Thus, referring to online product reviews has become a valuable resource for assessing product worth (Chevalier and Mayzlin, 2006).

(8)

8

Mayzlin, 2006). Thus, besides providing product information themselves, firms are identifying the benefits of online consumer reviews as a new marketing tool (Dellarocas, 2003).

Online consumer reviews have a very significant impact on sales since they are perceived as more quality representative (Cheong and Morrison, 2008). However, the sorting of consumer reviews is complicated, as they may give mixed reviews by liking some features of a product but criticizing others (Archak et al., 2011). Such variances in the nature of the online reviews (i,e. if they are positive or negative) can be best illustrated by the term valence (Liu, 2006). Scholars have found differential impacts of review valence suggesting that positive reviews boost sales but negative reviews harm sales. For example, Cheong and Morrison (2008) state that negative reviews can have damaging consequences for the brand equity and Ludwig et. al (2013) shows that positive consumer reviews will favorably influence product trustworthiness. Whereas some scholars have argued that the valence of customer reviews affects sales positively (e.g. Chevalier and Mayzlin, 2006; Li and Hitt, 2008), others have shown an insignificant relationship (e.g., Chen et al., 2011; Liu 2006). Nevertheless, the large volume and absence of order for qualitative information in customer reviews presents currently a dare for online vendors (Singh et al., 2011). Apart from consumer-generated reviews, a lot of consumers seek the recommendations of trusted professional experts (Cheong and Morrison, 2008; Schwartz et al., 2011). That is aligned with the fact that the information online is sometimes incomplete, biased or falsified (Chen and Xie, 2008; Yoo and Gretzel, 2009). In many industries, such as entertainment and financial services, therefore, professional critics and their expert opinions are extremely popular (Shaffer and Zettelmeyer, 2002). Compared to consumer reviews, their recommendations are professional and independent of the vendors. Known for their integrity, expert reviews have become a respected source of information for assessing product quality (Chen and Xie, 2005). Thus, the review source i.e. user or expert may influence the final judgment of the consumers (Grewal et al., 1994). Gathered experience with online reviews shows that consumers see experts as better informed and more reliable, hence, source credibility seems to influence consumption probability (Senecal and Nantel, 2004).

(9)

9

information from reviews, and have overlooked other information channels such as sampling (Chevalier and Goolsbee, 2003; Godes and Mayzlin, 2004). However, since samples are free of charge, the cost of trial is decreased (Heiman et al., 2001). In addition, as direct sources of information to the buyer, samples have a larger effect on sales than indirect practices such as advertising (McGuinness et al., 1992).

Previous studies indicate that presenting product samples is an effective marketing strategy to trigger current and future consumption. Sampling is efficient in presenting new products and it is an important tactic in the initial stages of a product’s life as it distributes product information through WOM (Jain et al., 1995). Besides, sampling has demonstrated increasing helpfulness for consumers in the long term and a tendency to buy in the short term. Some authors (e.g., Alba et al., 1997; Klein 1998; Lynch and Ariely, 2000) have advocated that when the Internet allows consumers to access the experiences of others and to collect information often hard to find in offline settings most goods can be associated with researchable attributes. Then, different types of information are related to different cognitive processes that influence the way information is assimilated, acquired, and processed (Johnson et al., 2003). Nevertheless, sampling remains rather poorly investigated in marketing and existing studies provide little guidance to practitioners about tactics that can successfully impact consumer choice.

1.2. Problem Statement

(10)

10

review) is interesting as it can result in different levels of mental understanding and product preferences.

Additionally, the role of the review source has also showed unclear results: some studies show that source expertise intensifies the message impact; others show that respondents trust more non-expert or user sources (Senecal and Nantel, 2004; Lee et al., 2007).For instance, Huang and Chen (2006) study how sales volume and customer reviews impact consumer online product choices, depending on the relative effectiveness of two recommendation sources (expert reviews vs. consumer reviews). The results showed that people refer more to the choices and evaluations of other users as cues for making own choices. On the other hand, Reinstein and Snyder (2005) argue that reviews are associated with uncertainty of quality, thus, the established reputation of experts tends to exert the most influence.Their study on box office revenues for movies briefly suggests that expert reviews can have a significant role for conveying information about goods of uncertain quality and increase consumer demand. Professional or expert judgments are presumed to demonstrate knowledge and understanding in order to determine what qualifies for excellence in a given field (Bourdieu, 1993). However, research mostly studied how online customer reviews affects the process of making a purchase decision and the need of more research on expert reviews can be identified (Chen and Xie, 2008; Dellarocas, 2003; Godes and Mayzlin, 2006).

(11)

11

reviews are written by expert or non-expert (i.e. user), therefore, is an interesting research direction.

Undeniably, in the context of WOM it can be clearly defined that online reviews overcome the shortcomings of vendor-centric marketing communication messages. Although online reviews generate important information insights for various goods and services, the type of information that consumers search for and the way they seek it to make choices, is different. Some goods and services are also hard to assess before purchase. Additionally, it has been established that mechanisms used by sellers to enable consumers see or experience product attributes before buying (e.g., consumer comments, expert references, multimedia demonstrations) increase the time spent on a web site and the probability of purchase is greater for experience than for search goods (Huang et al., 2009). Therefore, for the purpose of this research, experience goods are chosen as a main source of interest since assessing their product quality is naturally subjective, tied to uncertainty and they are difficult to evaluate (Daft and Lengel, 1984). Following this logic, the paper focuses on TV series as experience goods for which the quality is uncertain prior to consumption (Reinstein and Snyder, 2005). Trailer is taken as sample in this specific case, as it is designed with the purpose of screening TV shows and movies as an exhibition of content and quality. Overall, the research aims at studying the effects of review valence and sample presence or absence on consumption likelihood, and whether user or expert opinion can change this relationship. Therefore, the main research question could be formulated as:

“In what way does the effect of review valence and trailer presence differ for willingness to watch a TV show depending on whether the source of the review is an expert or a user?”

1.3. Research Questions

This study articulates several hypotheses and conducts an online-based research to investigate how consumers are influenced by online reviews depending on the review valence, trailer and the source of credibility. Besides the main research question, several sub-questions are suggested in the context of the paper:

 How does the presence of a trailer impact watching intentions for a TV show?

 Does review valence change the effect of trailer on willingness to watch a TV show?

(12)

12

 What is the joint impact of review valence, source of credibility and trailer on the willingness to watch a TV show?

1.4. Theoretical Relevance

Previous research (e.g., Alba et al. 1997; Klein 1998) has advocated that the Internet has altered the traditional relationship between buyers and sellers in an online environment. Specifically, it lowers the cost of collecting and distributing information (Hoffman and Novak 1996; Zettelmeyer et al.,2006) and stimulates new ways to learn about goods prior to purchase (Lynch and Ariely, 2000). As learning about goods becomes a vital part in an online environment, it is essential to investigate the possible interaction of sampling and review valence as they are both information means for consumers. Thus, the purpose of this paper is to make some basic contributions and validate how consumers evaluate product quality through online reviews as well as through own subjective deductions based on viewing a sample.

Furthermore, cognitive attention behavioral research illustrates that attention is selective, and individuals are inclined to focus more on certain types of information than to other types (Lowe and Steiner, 1968). Therefore, another objective is observing if a sample can be distinguished as a strong product quality signal that decreases consumers’ uncertainty about a product and results in higher interest among consumers. As consumers and experts’ product reviews differ in style, content, detail consideration etc. (Chevalier and Mayzlin, 2006) then the distribution and variability of product reviews can lead to different consumer reactions. From the perspective of the Signaling theory (Connelly et al., 2011) both consumer reviews and expert reviews indicate quality evaluation, but as the role of source of credibility is unclear, the study could provide an understanding of whether consumers relate more to user or expert reviews. Most importantly, the contributions of the study might be useful in providing insights and future research direction in explaining why certain combinations of user/expert positive or negative reviews have a less negative/positive effect on consumption probability than others and make an observation on combining samples and reviews.

1.5. Managerial Relevance

(13)

13

surveyed, the most time spent was contributed to digitally viewing TV shows and videos. Evidence shows that in a positive or negative review balance, people trust the majority’s judgment in shaping impressions (Purnawirawan et al., 2012). However, user reviews are related to subjective WOM while expert reviews are typically tied to technical seller-generated information (Chen and Xie, 2008) and knowing what influences consumer behavior in online trading is crucial. For managers, the results from the study regarding WOM can clarify whether they should actively seek various WOM messages produced by experts or/and users in order to reach larger audiences and get more people exposed to a certain product/service. In the case of TV shows networking it is crucial to achieve higher ratings as they lead to more benefits from advertising, thus, the more WOM accumulated, the more benefits generated. Besides, individuals with different levels of expertise look for different types of information – more experienced consumers desire specific attribute data, while less involved or experienced ones preferred the information to be easily visualized, interpreted and reproduced (Park and Kim, 2008). Then, in order to please the various consumers, offering a sample in addition to expert and/or user reviews can be the right marketing approach. Besides, providing a sample can greatly discount negative review impressions and improve consumer satisfaction due to the opportunity to form own evaluations. By understanding how sampling and online reviewing works best in online trading, managers can adjust their strategies and benefit from satisfying prior and after purchase expectations, improved positive attitudes and encourage customer loyalty. Although some consumers may consider neither product reviews nor samples as means to reduce product uncertainty, it can be assumed that when online sellers attach such information to their commodities, a large segment of potential customers can refer to online reviews and product sampling as a way to assess quality prior to purchase.

1.6. Structure of the thesis

(14)

14

2. Theoretical Framework

In order to research the effects of online reviews and sampling on consumption behavior, a good theoretical understanding of the variables involved is required. In this section, a quick theoretical foundation for the research will be provided, followed by the conceptual model and rationales for formulating the hypotheses. The chapter will also provide expectations about the outcomes from the suggested propositions.

2.1. Literature Review

2.1.1. Customer Engagement Behavior

Customer engagement has emerged as a fresh view in the field of customer management (Verhoef et al., 2010). It has been emphasized that the evolving concept of CEB is an essential paradigm in our highly networked society. Originating from the research of van Doorn et al. (2010), Verhoef et al. (2010) reflects that CEB is a behavioral expression towards a key entity (e.g. a brand or a firm), other than just an investigation of purchase behavior, caused by motivational drivers. The concept of CEB suggests that focusing on the behavioral sides of the relationship between the customer and the firm goes beyond transactions, consumption and repurchase intentions (van Doorn et al., 2010). CEB is built upon five main dimensions: valence (positive or negative), scope (temporal and geographic), form and modality, nature of impact and, lastly, customer goals. Additionally, van Doorn et al. (2010) founded a conceptual model proposing that CEB is tightly related to consumer characteristics, company dealings and the contextual environment. The established conceptual model shows that customer engagement antecedents involve consumer satisfaction and dedication driven by involvement and social interaction. Some consumer actions (e.g., WOM activity and online reviews) could be positive or negative thus reflecting the content of the customers.

Verhoef et al. (2010) and Bijmolt et al. (2010) recognize the significance of the impact of WOM and co-creation. For example, it has been indicated that buying and selling stocks as result of the valence of chatter can lead to gains for a company (Tirunillai and Tellis, 2012). Rather than forecasting future purchase levels, analytical models can line up to predict the amount of WOM recommendations. Assuming that WOM definitely affects revenues, companies may focus on targeting customers with a high inclination to WOM (Verhoef et al., 2010).

2.1.2. Word-of-Mouth

(15)

15

important to note that in this definition ‘informal’ and ‘among consumers’ states that WOM is grasped as communications by, and for consumers, excluding a company involvement in this communication route. WOM is established as an influential marketing tool since consumers prefer to follow WOM recommendations rather that mass media or other types of advertising (Engel et al,. 1969). Trusov et al.(2009) reason that WOM marketing is much more cost-effective than traditional marketing activities such as advertisements. They also prove empirically that there is a positive relationship between WOM communication strategies and firm performance by showing that referrals to WOM have a much stronger effect on securing new customers than traditional marketing tools.

An early study by Woodside and Delozier (1976) discusses that WOM can make a consumer inclined to make risky purchase decisions, such as, for example, buying a newly introduced product. They maintain that for this effect to occur, consumers should first be acquainted with the benefits of the new product by its attribute facts provided by the seller. However, the launch of the Internet-based shopping has transformed the role of WOM and the general buying behavior. Nowadays, consumers can share their opinions practically with every other interested Internet user in the world. This has significantly extended the range of WOM and the impact of online consumer reviews (Burton and Khammash, 2010).

WOM is reported to be more trustworthy, credible, and reliable when opposed to firm-based communications. Electronic WOM then provides more flexibility in the traditional role of opinion seeking and sharing. Previous research also showed that the effectiveness of online WOM relates to the strength of the bond between the recommendation source and the customer (Zhu and Zhang, 2010). The different forms of online WOM contribute various forms of value to the customer (Gruen, et al., 2006). Product reviews, as a form of electronic WOM, are of specific interest to this study. There is, however, a difference between electronic WOM and online reviews. Mainly, electronic WOM refers to a matter, originated from a company that is passed on by consumers, while an online review aims at consumer-generated content (Cheong and Morrison, 2008). For numerous product categories, such as books, games, beverages etc. online reviews are rather common. Online reviews normally discuss a product/service from a consumer’s viewpoint and in terms of a usage (Chen and Xie, 2008).

(16)

16

(e.g., providing digital samples and expert reviews) that can lessen those effects seems to be of great importance.

2.1.3. Online Reviews

Exposure to online reviews can lead to different behavioral responses among consumers. Most studies tested the effect of online reviews on purchasing. For instance, Chevalier and Mayzlin (2006) investigated online bookstores and found a positive link between online review volume, valence and purchasing behavior. They observed that although the websites provided more positive than negative reviews, the influence of negative reviews is generally greater than the influence of positive reviews. They also found that consumers choose to read the review text more willingly than referring to summary statistics. Another study of Dellarocas et al. (2007) explored the movie market and likewise found a positive relation between online reviews and purchasing actions. Zhou et al. (2008) linked different forms of review ratings on eBay and found those review valences are more effective than feedback score in influencing auction price. Clemons et al. (2006) encountered interesting results that for beer, a regularly purchased product, high-end reviews essentially carry more weight than low-end reviews. Ye et al. (2009) placed their research into the hotel business and the results showed that both the average online customer review rating and the variance of the ratings effect hotel bookings. Chen et al. (2012) claimed that the changes in the review valence, not the absolute valence, actually impact firms’ stock returns. However, when Neelamegham and Chintagunta (1999) empirically evaluated the relationship between WOM and weekly revenues they found no significant correlations.

(17)

17

positive information in the course of assessing a product (Ahluwalia, 2000; Maheswaran and Meyers-Levy, 1990). Subsequently, researcher and practical challenge can be focused on the question whether an impact by a negative review can be somehow tempered.

Table 1: Summary of Literature for Online Reviews

2.1.4. Source of Credibility for Online Reviews

Overall, there are three types of online reviews: consumers, companies and experts. Many online vendors use mainly rankings and product reviews by experts or consumers as a trusted tool for reducing perceived risk and increasing trust (Dellarocas et al., 2007). Therefore, in this sub-section, only consumer and expert reviews will be discussed from a credibility point of view.

Study Method Key Findings Basuroy, Chatterjee,

and Ravid ( 2003)

Time-series cross-section regression

Negative review volume influences revenue more than positive volume, but the influence of negative reviews decline over time.

Review volume has mixed effect on revenues. Senecal and Nantel

(2004)

Generalized estimating equations

Online product reviews 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

Negative online reviews have a negative effect and positive online reviews have a positive effect and on the customer’s buying intention.

Chevalier and Mayzlin (2006)

Differences-in-differences

Increase in review valence results in increase in relative sales.

The effects of negative (1-star) reviews are greater than positive (5-star) reviews

Clemons, Gao, and Hitt (2006)

Multiple egression Both review valence and variance are positively related to future sales.

High-end ratings are more significant than low-end ratings, since beer is a repeat purchase product.

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

The quality and quantity of online reviews has a positive effect on purchase intention. Involvement moderates these relationships.

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.

Chen, Wang, and Xie, 2011

Normative model development

Review valence and percentage of 5-star reviews does not have an impact on sales. Percentage of 1-star reviews has a negative impact on sales

Moe and Trusov, 2011

Multiple Regression Increases in review valence encourage the subsequent posting of negative

ratings, but discourage positive ratings. Social dynamics cause directly review valence effect for sales. Variance and volume have indirect impact on sales Tirunillai and Tellis,

2012

Classification Algorithms

(18)

18

Information credibility is a major concern when aiming at sharing information effectively, especially when it predicts consumer’s further actions (Grewal et al., 1994). Without information credibility, buyers are not keen to take action upon the information they have read (Cheung et al., 2009). Hence, a consumer will not be influenced by a product review, if they perceive the information as less reliable. Credible information engenders reliability and consumer’s trust (Chen et al., 2008). Habitually, it comprises the degree to which a person identifies information as believable, trustworthy, expert, and competent (Flanagin and Metzger, 2000). The two factors that matter the most for consumers are expertise and trustworthiness. Individuals are likely to accept as true information from a source associated with high expertise and reliability (Cheung et al., 2009).

Online reviews by consumers can be classified as user-generated content and their online has dramatically increased because many consumers share their experiences and rate the product/service online (Shridhar and Srinivasan, 2012). Consumer-generated reviews are mainly based on the personal involvement and are very dependent on individual preferences, characteristics and the situation in which the product was used (Chen and Xie, 2008). Unlike professional reviews, such reviews are often emotional, individually relevant and do not always include critical evaluations of practical aspects. For example, in the case of recommending movies, online consumer reviews do not contain evaluations about the technical or artistic features of the movie. Generally, the tone of a consumer-generated review is either positive or negative and from language perspective less formal (Chakravarty et al., 2010)

On the other hand, reviews written by experts (also known as third-party reviews) aim at guiding buyers towards more qualitative experiences. Expert reviews naturally have specialized knowledge about the evaluated product category and express an independent opinion about it (Chakravarty et al., 2010). Conceptually, professional review differs from advertising in two important ways: it is not necessary complimentary of product i.e. it can be unflattering and it is assumed to be independent, therefore, impartial and objective (Chen and Xie, 2008). The opinions in expert reviews are usually more balanced and formal than reviews provided by consumers. Expert reviews are a vital source of information for consumers precisely for products of uncertain quality (Chen et al., 2011; Reinstein and Snyder, 2005).

(19)

19

fundamental component of online information for many products (e.g., music, travel, restaurants). Hence, Chen et al. (2011) advises that when developing and introducing new products, the professional critical opinion should not be ignored as it is also highly valued by consumers.

2.1.5. Product Sampling in an Online Environment

Product sampling is an essential marketing strategy for experience goods like movies and music. However, after people have experienced partially the product in online environment, little research has tested if they have grasped its real value. Online product reviews turn out to be popular in the latest years as more consumers count on preceding users’ experience for guidance in purchasing. For some product like electronics, online consumer product reviews have larger impact on consumer behavior than any other media. However, although samples’ impact on consumers in a digital environment is generally under-researched, the small amount of existing literature demonstrates that product sampling is an effective marketing tactic that can be utilized to stimulate current and future consumption. Product sampling is a marketing instrument that has the potential to introduce a novel product in the initial stages of a product’s life cycle (Jain et al., 1995). By trial customers’ weak beliefs in products can be transformed into strong ones and result in commitment to buying. On occasion, sampling is a more effective marketing practice than some types of marketing promotions because the consumer's cost of trial is reduced (Ailloni and Cheros, 1984). The prevailing literature, however, provides insufficient findings how samples and online reviews can be related or under what conditions online retailers can offer one of these options or when to pair those (Hu et al., 2009). It remains unclear if sampling is more effective than reviews, especially in digital environment. Therefore, a research gap can be identified: the need for strategic approach for online sellers when to provide sampling and/or reviews for their goods and services.

In the context of this study, it is important to make clear that trailers will be used as samples, representing the qualitative portion of the researched TV series. As trailers are believed to be a unique form of marketing promotion and a type of narrative presentation, where marketing discourse as well as narrative liking are bundled (Kernan, 2004) they fit the rationale for sampling in this study.

2.2. Conceptual Model

(20)

20

Mayzlin 2006). Thus, product valence is selected as an independent variable for the conceptual model. The second predictor is product sampling as it has the power to stimulate consumption and it is considered an efficient marketing tool (Jain et al., 1995). Additionally, it has been established that expert recommendations are considered credible sources of information, especially for products of uncertain quality (Chen et al., 2011), thus, an expert source (versus non-expert) is assumed as a moderating variable. For the purpose of this study, willingness to watch was used as a dependent variable construct towards which the effects of the independent and moderating variables can be investigated.

The conceptual model (Figure 1) illustrates the hypotheses for the moderating effect for source of credibility on the relationship between review valence, trailer and consumption likelihood (H4a,b) based on the literature review. Furthermore, the model shows the hypotheses for the main effects of review valence and trailer on willingness to watch (H1, H2). The possible interaction effect between review valence and trailer on watching intentions is also tested (H3). The suggested model represents graphically the relationships of interest and it relates them to the hypotheses, described in the next section.

Figure 1: Conceptual Model

(21)

21 2.3. Hypotheses

2.3.1. Valence of Reviews

A reflective look in the literature review discloses that studies have indicated mixed results regarding the relationship between the valence of online reviews and consumers’ purchase and behavior intentions. Prior research shows that many authors (Table 1) showed that review valence does matter. Although there is an agreement about the distribution of valence in positive and negative assessments, a discrepancy about the level of the impact of positive and negative reviews still exists.

Positive information is often viewed as more ambiguous as having positive attributes does not automatically indicates that the target object exemplifies good quality or features (Lee and Youn, 2009). According to the Accessibility/diagnosticity theory (Feldman and Lynch, 1988), negative information is generally more analytical than positive one. Since the product rated to be positive can be of high, average and low quality, accessing quality in this case can be more uncertain. In contrast, negative information strongly advocates lower quality performance. Furthermore, this theory also underlines that the diagnosticity of information may be dependent on situational factors. When the consumer perceives reviewing as difficult or ambiguous, the diagnosticity of the existing information will escalate. In the meanwhile, negative WOM relates to potential harm, cost or risk, therefore, consumers would rather trust negative reviews in order to avoid potential purchase risk (Nga et al., 2013). Hence, the diagnosticity and impact of negative reviews seems much stronger in some situations, like when user expertise is low or the product brand is not well-known.

Another framework - the Prospect theory (a psychological tool stressing the negativity effect) can also be related to the matter. Overall, this theory states that people are risk averse in situations where they expect to gain something and in situations where they are likely to lose something (Kahneman and Tversky, 1984). When faced with purchase decision, individuals usually have a reward-seeking condition; they wish to gain something such as finding a good product that will bring value and fulfill their needs. Logically, if people read online reviews before buying goods, they might be more attentive to any negative information in the online reviews, and they will be focused on shunning from risk as much as possible.

(22)

22

for negative reviews, the opposite effect is expected. Following this theoretical reasoning, valence of reviews is used as the first independent variable in this research forming the first hypothesis as:

H1: Negative reviews (as opposed to positive ones) will decrease the intentions to watch a TV

show.

2.3.2. Impact of Sample on Watching Intentions

The Uncertainty Reduction Theory (Berger and Calabrese, 1975) provides a good theoretical base that can be used to study how individuals relate to available online information prior to purchase decisions. Basically, the concept is that through communication, people decrease uncertainty originating from situations in which experiences do not match expectations or when relationships change. The theory shows that when consumers have insufficient knowledge or cannot predict the results of consuming a product they tend to engage in uncertainty reduction efforts to alleviate the risks associated with ambiguity. Common ways to deal with uncertainty are via indirect information exchange (e.g., reading online reviews) or by seeking available samples. Thus, the consumers are able to perceive the essential quality of the product apart from the sale price. In cases when the sale prices are lower than the assumed intrinsic value, consumers are likely to purchase.

As noted previously, experience products are difficult to evaluate without actually viewing/experiencing the product (Eliashberg and Sawhney, 1994). Hence, the Signaling theory (Donath, 2007) can be useful for better evaluating samples. The theory reasons that there are definite signals available online about information sources – in fact, signals that are hard to falsify. Those signals are protected by rules of law, or they are costly to be gained or imitate, thus, they have reliable information quality and source expertise. As this research takes trailers as a sample representation of a TV show, it can be said that trailers as a type of advertising have their distinctive form of narrative story and exhibition, which can serve as signals of originality. Hence, they are a clear representation of the content and the associated quality attributed to TV series. The presence versus the absence of trailer could then help in investigating consumption likelihood and it is selected as the second independent variable in this paper. Therefore, having in mind those two theoretical references and in regard to investigating the effects of samples on consumption probability, the second hypothesis can be formulated:

H2: The presence of a trailer (as opposed to the absence of trailer) will increase the willingness

(23)

23

When an individual can detect that product performance is only slightly different from pre-usage expectations, the assessment should be in the latitude of acceptance and an assimilation effect should follow (Olshavsky and Miller, 1972). Thus, if a person is exposed to a positive review and a sample she/he likes they are likely to select the product. However, it is curious to observe what will happen if a negative review and a sample are both given as opinions for evaluation. Consistent with the Uncertainty Reduction Theory, negative reviews cause high levels of uncertainty and result in increases in information search behavior among customers. Besides, an increase in uncertainty yields a decrease in fondness of the product for the consumer. Thus, variances in product reviews generally raise the uncertainty of assuming a product’s quality. Consequently, review valence causes uncertainty and consumers might rely more on subjective and personalized options such as sampling. Since samples serve as a direct source of information to users, it can be expected that the effect of seeing them would increase the consumption likelihood as it will improve the generally low expectations or doubts associated with the content value. The reservations of a consumer towards the quality of a product can be converted into strong beliefs and liking when sampling is offered (Hu et al., 2009). Besides, according to the Information-processing theory, people’s evaluation process is seen as more cognitively driven and rational. This implies that when consumers try to decide upon obtaining a product, they will seek the ultimately satisfying attributes. In other words, a consumer will look for ways to benefits from the product (Bettman et al., 1998), hence, look for products with a certain set of attributes that deliver the benefits. Therefore, it can be assumed that if a consumer is faced with a negative review and a sample, and he/she finds the review to be much more negative than the sample itself, then it may be assumed that negative reviews will increase the willingness to see a show and decrease the negative impression of the review. This can be hypothesized as:

H3: A trailer has a significant positive effect on watching intentions when reviews are negative.

2.3.3. Impact of the Source of Review on Willingness to Watch

(24)

24

Consequently, both online reviews and sampling act as informative quality signals regarding the nature of the product and since expert reviews are considered impartial and objective (Chen and Xie, 2008) they can potentially impact watching intentions.

From a consumer’s perspective, there are two accessible signals about the core quality of a product which consumers utilize is a linear manner in order to make a purchase decision. Consistent with the analytical model proposed by Banker and Datar (1989) sensitivity and precision can be seen as signals of performance measurement. The relative meaning of each signal is proportional to the product of the sensitivity and the accuracy of the signal. Sensitivity depicts the degree to which the expectation of a signal fluctuates with an agent’s action while the precision of the signal shows absence of noise in the signal. Unlike online reviews, sampling is a signal with less noise for customers since online reviews can be interfered by sellers, authors, or producers. The Attribution theory implies that sampling is not always influential if consumers receive an external basis for their product usage behavior (Lawrence and Kamins, 1988). From a practical viewpoint, a sample provides only a “part” of the full product, therefore, it may be linked to a part product that expires after a certain time, or it could be a detachment of the value providing features of the product (Chellappa and Shivendu, 2005). Once presented with a framed message, Attribution theory suggests that consumers will make an attempt to evaluate whether the message contains an accurate representation or how much credibility is associated with the message source. The Attribution theory framework provides evidence that the influential impact of a message is normally reduced whenever consumers detect reporting or knowledge bias to the source (Eagley et al., 1978).

(25)

25

credibility effect is held to apply as well in the online environment (Lim et al., 2006). Therefore, positively or negatively outlined reviews presented by experts should impact product attitudes to the greatest extent. Hence, it is expected that when presented with a sample and an expert opinion, a consumer would rely more on the professional review as it is implying that an expert is knowledgeable of the whole product quality. Then, the impact of sampling on consumption probability would be decreased if the belief in the source expertise is high. In order to address the effects of credibility source on the relationship between samples, reviews and consumption, the following hypotheses are defined as:

H4a: The impact of a review on watching intentions is stronger when it is given by an expert.

H4b: An expert review (as opposed to user review) decreases the effect of sampling on

consumption for a TV show.

3. Methodology

This chapter illustrates and outlines in what way the research is conducted. More specifically, the type of research is defined and presented, as well as the data collection method and types of analyses which were employed. Furthermore, the chapter includes the characteristics of the respondents, the data collection, the overview of the measurement scales, the manipulation checks and the reliability of scales.

3.1. Research Design

(26)

26

Carman (1990) found that one of the key influences on consumer expectations is previous knowledge or involvement with a product or service. Thus, in order to avoid bias tied to previous TV show buzz, the new and relatively less-known show “Go On” (starring Matthew Perry) was selected. Generally, the intention of the survey was to create a realistic atmosphere for the participants in order to predispose them to give their honest opinions. Therefore, the layout of the popular and authoritative source for movies and TV shows IMDb web site (www.imdb.com) was used and modified for the purposes of the thesis research. The captions and the text of the reviews were the same, except that they were manipulated in terms of negative or positive content. The original trailer (89 seconds in length) of the show was also presented. Besides, the source of the review (use/expert) was also made visually distinguishable. In order to stick to the original layout of the IMDb site, evaluation stars were also assigned to the TV show. The types of positive/negative expert/user reviews layouts can be found attached in Appendix 1.

As the experimental design was selected to be between-subjects, each participant is exposed to only one of the eight possible scenario conditions to also ensure that any possible carry-over effects are avoided (Table 2). Each scenario differed in review valence, trailer presence and source of review; however, the last part of each condition consisted of general information questions like gender, education, nationality etc. Besides, several general questions concerning frequency of reading online reviews and importance of trailer were specifically created for the survey.

Scenario # Participants Review Valence Trailer Source Of Credibility

1 31 Positive Yes User

2 33 Negative Yes User

3 25 Negative No User

4 27 Positive No User

5 27 Positive Yes Expert

6 26 Negative Yes Expert

7 35 Positive No Expert

8 37 Negative No Expert

Table 2: Experimental Conditions per Scenario

3.2. Data Collection

(27)

27

The total number of participants who completed the survey is 241 and no respondents were excluded due to missing data. The general descriptives of the sample will be discussed in the next chapter.

3.3. Variables, Measurements and Manipulations

Apart from the general socio-demographics questions, all measurement scales used in the current study were based on 7-point Likert scales (1=strongly agree; 7=strongly disagree). The control variable gender is measured on a dichotomous scale; age, education, occupation and marital status are measured on categorical scales. The scales’ validity and reliability are discussed in the next chapter.

3.3.1. Review Valence

As already mentioned, review valence is used as an independent variable in this research. There are two conditions for the valence of the review – positive and negative. Both the content of the reviews and the rating of the TV show are manipulated. For the positive review, 9 (out of 10) stars are assigned; for the negative review 3.2 stars are given. The reviews were written as informational statements on plot, cast and genre of the TV show aiming to make them more informative and persuasive.

Several questions were implemented in the survey in order to check the levels of the respondents’ perceptions of review valences. To test the manipulation for review valence, respondents were asked to evaluate the reviews as positive/negative; favorable/unfavorable and good/bad and on a 7-point Likert scale as previously used by Ahluwalia et al. (2000). Only the first two questions are used to check manipulations. One- and Two-way ANOVA were used to validate if there were significant differences in the means for the positively and negatively valenced reviews. The results of the manipulation checks are included in the next chapter.

3.3.2. Source of Review

(28)

28

7-point Likert scale is used as recommended by Senecal and Nantel (2004) for testing influences of online product recommendations on consumers. However, for the purpose of the study, only the first item was used for manipulation checks. The results for this manipulation are also included in the next chapter.

3.3.3. Willingness to Watch

The scale for measuring the dependent variable willingness to watch is composed of 7-point Likert scale with 4 items, formed as statements that are employed to measure the likelihood of someone purchasing a product. While termed purchase intention by Berens et al. (2005), it is regarded here more as a measure of attitude towards the act of consuming (i.e. intentions towards watching) attributable to its hypothetical phrasing and the third item which involves recommending the product to others. The statements used are: “If I was planning to watch a TV series of this type, I would choose this one.”/ “I would watch the TV show.”/ “If a friend was looking for a show of this type, I would advise him/her to watch the TV show.”

3.4. Plan of Analysis

The next chapter will be devoted to testing the hypotheses in order to properly answer the research questions. Initially, the sample descriptives will be briefly presented. Then, the reliability of the multiple-item scales will be presented by Cronbach’s alpha results. A linear multiple regression analysis is selected to estimate the relationship between the two independent variables, the proposed moderator and the dependent variable. Finally, the hypotheses will be tested according to the outcomes of the regression.

4. Results

4.1. Sample

(29)

29 % N Gender Male 38.2 92 Female 61.8 149 Age < 18 4.1 10 19-25 42.7 103 26 - 30 41.5 100 31 - 35 5.8 14 36 - 40 2.1 5 > 40 3.7 9

Education High School 16.6 40

Bachelor 34.9 84 Master 39 94 PhD 1.2 3 Diploma/Professional Degree 5.8 14 Other 2.5 6 Occupation Student 33.2 80 Full-time employed 45.6 110 Part-time employed 6.6 16 Self-employed 6.2 15 Unemployed 6.6 16 Other 1.7 4

Marital Status Single 45.6 110

In a relationship 40.7 98

Married 6.2 15

Married with children 7.5 18

Table 3: Description of Demographics

In terms of the frequency of reading online reviews before selecting which TV show to watch (Table 4), the statistical outcomes showed that 30.3% read reviews occasionally, 22.4% rarely and 18.7% stated they read reviews most of the time. Most of the participants (36.9%) also stated that they spend on average less than 10 minutes or 10-15 minutes (25.7%) on consulting reviews before making a choice in TV shows. As for trailer, 32.4% of the respondents have stated that it is very important followed by 26.6% who considered it somewhat important. The average time of watching the trailer was 63.48 seconds (out of 89 seconds full-length video).

Watching Frequency Consulting Online

Reviews

Review Reading Time

% Trailer Importance %

% Not reading at all 16.6 Not at all important 7.1

Never 2.9 20.3 < 10 min 36.9 Of low importance 12

Rarely 14.5 22.4 10 – 15 min 25.7 Slightly Important 11.2

Sometimes 29 30.3 15 – 20 min 11.6 Neither important nor unimportant

5

Most of the time

42.3 18.7 20 - 30 5.8 Somewhat important 26.6

Always 11.2 8.3 30 - 40 2.9 Very important 32.4

>1 hour 0.4 Extremely important 5.8

Table 4: General Descriptives

4.2. Correlations and Reliability of Scales

(30)

30

correlation analysis is used for justifying the use of new sum variables that portrays the whole construct up in one score. The results showed that the scale items are within the desired range of 1≤ r ≥1; the outcomes are attached in Appendix 3.

Reliability is key issue when variables from summated scales are applied as predictor components in conceptual models. Hence, it is crucial to know whether the same set of items would elicit the same reactions if the same questions are reorganized and re-distributed. Cronbach's alpha testing, therefore, is used to determine the internal consistency or average correlations of items in the survey. The testing of internal consistency showed results on the high side of reliability (above 0.7). Finally, the items of the scales were summated for further analyses. The tests of the scales’ reliability are summarized in the Table 5.

Construct M SD α Scale Question Measurement

Review Valence

Ahluwalia et al. (2000)

3.89 2.24 0.957 I think the review was: - positive………..negative - favorable……..unfavorable -good…………..bad 7-point Likert Scale(1=strongly disagree; 7=strongly agree) Source of Credibility

Senecal and Nantel (2004)

3.63 1.63 0.904 In my opinion, the source of review is: - an expert…………not an expert - experienced……..unexperienced - qualified………unqualified - skilled………unskilled - knowledgeable……..unknowledgeable 7-point Likert Scale(1=strongly disagree; 7=strongly agree) Willingness to watch Berens et al. (2005)

4.13 1.60 0.915 1. If I was planning to watch a TV series of this type, I would choose this one.

2. I would watch the TV show.

3. If a friend were looking for a show of this type, I would advise him/her to watch the TV show.

7-point Likert Scale(1=strongly disagree;

7=strongly agree)

Table 5: Internal Consistency of Questions

4.3. Manipulation Checks

(31)

31

(Mnegative=6.27; SD=1.05 compared to Mpositive=1.71; SD=0.89). In addition, when asked the question “I think the review was… (favorable/unfavorable)” a significant difference between the negative and the positive review conditions was also indicated (F (1,239)= 974, p<0.05). The means for negative reviews were again higher (Mnegative=5.98; SD=1.19 compared to Mpositive=1.83; SD=0.84). Therefore, the manipulation checks for review valence were considered successful.

The question “In my opinion, the source of the review is….(expert/not an expert)” was selected for testing the manipulation of expert and user as source of credibility and it showed significance, F(1,239)=344, p<0.05. The means for expert reviews are lower than those for user (Mexpert=1.96, SD=1.27; Muser=5.41, SD=1.61) and the manipulations were confirmed as valid. After performing one-way ANOVA, the two-way ANOVA was used for further testing since there is more than one independent variable and it was necessary to check if the other variables affect the manipulations. The manipulation for expert as a source of credibility for the question “In my opinion, the source of the review is….(expert/not an expert)” was validated as successful as it was significant, F (1,241)=380.549, p=0.000 and resulted in higher means for user, than for expert Mexpert=1.892 and Muser=5.443. The two-way ANOVA testing for manipulation of the review valence for the question “I think the review was: ….(positive/negative)”) was also successful and significant, F(1,241) =1315.331, p=0.000 with higher mean for negative review Mnegative=6.271, Mpositive=1.692. The second manipulation for the question “I think the review

was… (favorable/unfavorable)” was also valid and significant, F(1,241)=972.437,p=0.000 and had a higher mean in the case of negative review Mnegative=5.978, Mpositive=1.818.

4.5. Regression Analysis

For testing the hypothesis, regression analysis was preferred over ANOVA as it is seen in a way as a more flexible exploration technique. The rationale behind that decision is because in ANOVA one can only use variables with a limited number of categories as independent variables. Regression, in contrast, can comprise all types of variables (with the help of dummy variables).

(32)

32

negative review =1 (positive=0), trailer presence = 1(trailer absence =0) and expert =1(user=0). Additionally, the demographic variables age, gender, and education and are taken into account as control variables, and dummy coded in order to be included in the regression as well. The regression equation developed is:

Yi = α+ β1Z1i + β2X1i+ β3X2i + β4X3i + β5X4i + β6X5i +β7K1i + β8K2i+β9K3i+ β10K4i +β11K5i +β12V1i + β13S1i + β14T1i + εi α = Intercept Yi = Willingness to watch Z1i= Gender where: (Z1=male=1; female=0) X1i= Age (<18/19-25/26-30/31-35/36-40/>40) where: X1= age group <18; other=0

X2= age group 19-25; other =0 X3=age group 26-30; other =0 X4=age group 31-35; other = 0

X5 = age group 36-40; other =0 [age group > 40 is used as the reference group] K1i= Education (High School/Bachelor/Master/PhD/Diploma/Other educ.) where:

K1=High School, other = 0 K2=Bachelor; other = 0 K3=Master, other=0 K4=PhD; other = 0

K5=Diploma; other =0[category “Other” is the reference group] V1i= Review valence where:

(V1=negative valence, positive=0) S1i=Source of reviews where:

(S1=expert; user=0) T1i=Trailer where: T1=trailer; no trailer=0 εi= Error term i = Respondent

4.3.1. Regression Results: Main Effects

(33)

33

The main effects of the independent variables trailer review, review valence and expert and the interactions between them were tested and the full fit of the model was significant, F(10, 230)=15.815, p=0.000. The R2 value explained 40.7% of variance, however, the Adjusted R2 (as more relevant as an analytical tool in multiple regression) accounted for 38.2% of model prediction. The outcomes for the main effect illustrated that review valence (dummy for negative review=1 was used) is significant, F (10,230)= - 6.257, p< .05(p=.000) and the B is with negative direction (b= -2.205). Therefore, negative reviews as opposed to positive ones, decrease watching intentions, as predicted in H1. (Fig. 2)

Figure 2: Main Effect of testing Review Valence

Trailer as predictor of willingness to watch is significant, F (10,230)=3.263, p< .05(p= .001) and when present, it increases watching intentions which was hypothesized in H2 (Fig. 3). Expert had no significant effect as a predictor of watching intentions in the main effects, F (10,230) = - .644, p> .05 (p=.520). All the controlled variables gender (F (10,230) = - .257, p>0.05 (p=.797)), age ((F (10,230) =.511, p> .05(p=.610)) and education ((F (10,230) = - .641, p> .05(p=.522)) were insignificant (Table 6).

4.9694 3.303 0 1 2 3 4 5 6

Positive Review Negative Review

(34)

34

Figure 3: Main Effect of testing Trailer

4.3.2. Regression: Interaction Effects

The interaction between negative review and trailer was significant, F (10, 230)=4.541, p<.05(p=.000) and resulted in increased watching intentions; in other words when trailer is present, that negatively valenced reviews are offset by the effect of the sample, which is in line with H3. (Fig.4)

Figure 4: Interaction Effects of Trailer and Negative Reviews

In order to confirm if expert has a moderation effect on the relationship between each of the independent variables and the dependent one, the nature of this relationship must show changes as the values of the moderating variable change. The interaction between a negative review and expert resulted in a negative direction b= - .068, and it was not significant, F (10, 230)= - 0.148,

(35)

35

p> .05 (p= .882). That fails to support H4B, meaning that the impact of an expert opinion does not have enough statistical power to explain change or strength of the impact of review valence on watching intentions.

The interaction between trailer and expert was significant, F (10, 230)=4.05, p< .05(p= .000), nonetheless the presence of trailer does not reduce the willingness to watch a show but actually increases it. That means the relationship between trailer and willingness to watch is strengthened by the presence of an expert recommendation and it can be concluded that an expert review does not decrease the effect of sampling which fails to support H4a (Fig.5).

Figure 5: Interaction Effects of Expert and Trailer

(36)

36

Main Effects Interaction Effects

B Std. Error t p B Std. Error t p (Constant) 4.772 .457 10.450 .000*** 5.246 11.215 .000*** Gender - .168 .181 - .926 .355 - 0.044 - 0.013 -0.257 0.797 Age .143 .094 1.522 .129 0.046 0.028 0.511 0.610 Education -.099 .083 - 1.196 .233 - 0.05 - 0.035 -0.641 0.522 Trailer .441 .177 2.498 .013** 1.101 0.344 3.263 .001*** Expert .196 .178 1.102 .272 - 0.212 - 0.066 -0.644 0.520 Negative review -1.632 .176 -9.248 .000*** - 2.205 -0.689 -6.257 .000*** Negative Review × Expert -0.068 -0.019 -0.148 0.882 Trailer × Expert 1.909 0.494 4.05 .000*** Negative Review × Trailer 2.161 0.58 4.541 .000*** Negative Review × Trailer × Expert -2.035 -0.394 -3.096 .002*** R2 0.304 0.407 Adjusted R2 0.286 0.382 df 6(234) 10 (230) F 17.032 15.815 Model Sig. 0.000*** 0.000***

Note: DV = willingness to watch; *Coefficient is significant at: p< .10; **Coefficient is significant at: p< .05;

***Coefficient is significant at: p< .01

Table 6: Regression Model

Figure 6: Three-way Interaction Effects

5.7778 4.0645 4.9619 _5.2099 3.6282 4.0303 2.6486 2.9733 0 1 2 3 4 5 6 7

Expert User Expert User

(37)

37

4.3.3. Hypotheses Testing: Overview

This section will simply summarize the test results so far and evaluate the hypotheses support using the data output from the regression analysis and the presented corresponding graphical representations from the chapter. The results will specify whether the hypotheses are supported and provide a brief description in a tabular form. This has been decided for the sake of convenience and overview of the hypotheses and the observation results is provided in Table 7.

Hypothesis Supported Results

H1: Negative reviews (as opposed to positive ones) will decrease the intentions to watch a TV show.

Yes

The effect of review valence is significant, F (10,230)= - 6.257, p< .05(p=.000). Negative reviews decrease watching intentions, positive reviews increase watching intentions. This is illustrated in Fig 2.

H2: The presence of a trailer (as opposed to the absence of trailer) will increase the willingness to watch a TV show.

Yes

Trailer as predictor of willingness to watch is significant, F (10,230)=3.263, p< .05(p= .001) and it increases watching intentions when present.

H3: A trailer has a significant positive effect on watching intentions when reviews are negative.

Yes

The interaction between negative reviews and trailer was significant, F (10, 230)=4.541, p<.05(p=.000) and it resulted in increased watching intentions.

H4a: The impact of a review on watching intentions is stronger when it is given by an expert.

No

An expert review did not moderate the impact of review valence on watching intentions as it had no significant effect at all, F (10, 230)= - 0.148, p>0.05 (p=0.882).

H4b: An expert review (as opposed to user review) decreases the effect of sampling on consumption for a TV show.

No

The interaction between trailer and expert review was significant, F (10, 230)=4.05, p<0.05(p=0.000), but it increased the effect of trailer for watching intentions which only strengthened the relationship.

Table 7: Hypotheses Summarized

Online reviews versus samples: the role of review valence, trailer and source credibility in the case of TV series

Online reviews versus samples: the role of review valence,

trailer and source credibility in the case of TV series

30

June, 2014

Master Thesis

Marketing Department University of Groningen

Online reviews versus samples: the role of review valence,

trailer and source credibility in the case of TV series

Preface

Management Summary

Contents

2. Theoretical Framework

3. Methodology

4. Results

5. Conclusions