How do sender expertise and argument quality influence the
credibility of online consumer reviews?
-The moderating role of involvement and susceptibility to social influence in the golf context-
How do sender expertise and argument quality influence the
credibility of online consumer reviews?
-The moderating role of involvement and susceptibility to social influence in the golf context-
Alessandro Maria Balice University of Groningen Faculty of Economics and Business
Department of Marketing MSc Marketing Management Master Thesis January 12, 2015 Damsterdiep 196 9713 EN Groningen Tel.: 06 33337856 E-mail: alessandromaria.balice@gmail.com Student number: s2620405
Management
summary
The aim of this empirical research is to analyze the relationship between the sender’s expertise with the product and the quality of the arguments presented in an online consumer review (OCR) on the credibility of the review itself. Two more variables are used as moderators: receiver susceptibility to social influence for sender expertise and product involvement for argument quality.
The effects of the variables are tested with a 2 (sender expertise: expert vs. non expert) x 2 (argument quality: strong vs. weak) between-participants experimental design. 151 respondents filled in an online survey with five questions, based on different items, to measure the credibility of an OCR. Then, other four questions were asked, concerning background information: age, gender, country and use of OCRs.
Results showed that argument quality and sender expertise positively affect the OCR credibility. Furthermore sender expertise resulted to positively impact the effect of argument quality on OCR credibility. However, no empirical evidence was found for the interaction effects of involvement on argument quality and receiver susceptibility on sender expertise.
Acknowledgment
This master thesis is one of my last steps before ending my university career and my life as a student. It has been a long and formative process, where I have learned a lot about myself and my limits, about my ways of facing problems and my way to solve them. It has been a process surrounded by thousand difficulties that made me stronger and more conscious about my capabilities. This process began in Italy, with my studies in Economics and Business Administration in Rome, and it ended in a new city in a foreign country, Groningen.
I will always consider my decision to finish my studies in another country as one of the best ones I have made so far. With this new experience I met extraordinary people who changed my life and who will always be part of it. I discovered a different teaching system that helped me to analyze better the concepts and focus more on the main aspect of the things.
My first big “thank you” goes to my family, that gave me this unique opportunity for which I feel blessed. With their help and with the support they gave me with all my friends back in my home town, I have always felt as I never left.
People that I would have considered as strangers or normal classmates one year ago are now a constant part of my life and finally became friends that I will never forget. I am truly thankful for all those here that supported me and pushed me whenever I needed it. They helped me to make things easier and to face exams with a smile, instead of a worry.
However, the conclusion of this process would have not been possible without my first supervisor, Liane Voerman, always available for any help or consult for my thesis, always providing at the same time constructive and instructive feedbacks. The reason why I am so happy about my topic and about my paper is that, with her enthusiasm and passion, she transmitted assurance and made me deal with problems in the most positive way. I consider passion and positivism the core of a job and if you can transmit these values to people around you, I think you have reached an important goal.
Together with hers, my last, but no less important thanks go to my fellow and friends master thesis students who, particularly last days, helped me a lot with good feedbacks and advices. They all deserve a bright future and career.
1. Introduction 7
1.1 Communication and word of mouth 7
1.2 Electronic word of mouth 8
1.3 Online consumer reviews and their credibility 9
1.4 Factors influencing credibility 10
1.5 Problem statement and research question 11
1.6 Academic and managerial relevancy 12
1.7 Structure of the thesis 14
2. Theoretical framework 15
2.1 Independent variable: sender expertise 15
2.2 Independent variable: argument quality 16
2.3 Moderators 17
2.3.1 Involvement 17
2.3.2 Susceptibility 18
2.5 Conceptual model 20
3. Methodology research 21
3.1 Research design and participants 21
3.2 Sample and procedure 22
3.3 Operationalization and reliability of scales 25
3.3.1 Operationalization 25
3.3.2 Reliability of scales 27
3.4 Manipulation of independent variables 28
3.5 Plan of analysis 29
4.1 Descriptives of the sample 30
4.3 Model selection 32
4.4 Hypotheses validation 38
5. Conclusion 39
5.1 Discussion 39
5.2 Limitations and further research 40
References 41
1. Introduction
1.1 Communication and word of mouth
Since the beginning of its development, Internet has literally changed the way both consumers and companies see and perceive products and goods, by enabling users to interact and share opinions about anything, working as a fundamental tool for interpersonal communication (Riegner, 2007). Many companies can count on the value of the World Wide Web to exploit their new services and make them more noticeable and attractive for new potential customers (Cronin, 1997).
Buying behavior is changing and people use online channels to evaluate better products and share opinions about their needs. They are now able to get all the advantages of the web by placing orders for both primary and experience goods, for personal consumption or as gifts for friends (Lohse et al., 2000). Through the Net, it is possible to gather information about what one needs and wants and have a clear picture of all the features of the searched goods, that are more difficult to gather in the offline world. The way people communicate is evolving year by year and online services, such as emails, forums and blogs are part of this process (Katona and Sarvary, 2014).
Relevant for this thesis, is that consumers are more willing to rate products, leaving comments or writing reviews about anything through a digitalized platform that is becoming a more and more popular (Robson et al., 2013).
In fact, nowadays consumers share opinions, ideas and information and communicate online talking about their experiences about many topics: movies, vacations, music, books (Berger, 2014).
Traditional WOM is more likely to be associated with family members’ and friends’ opinions about product or services (or people one may know) and the main feature is its interpersonal influence, that enhances their credibility (Berger, 2012).
1.2 Electronic word of mouth
Internet has changed the way people communicate and acquire information and consumers are more willing to share their ideas on social medias to find a “human touch” to believe more in what they are seeing and to have more direct feedbacks from people (Katona and Sarvary, 2014). Online, people can for example better examine a product, compare prices and look for reviews or specific characteristics of products; these characteristics might be more difficult or inconvenient to find in the offline world and usually require more time, such as going to the store, finding long queues or unprepared employees that sometimes can’t help (Rigby, 2011).
According to Senecal and Nantel (2004), with the introduction of the many new ways of sharing ideas online, the interpersonal influence has evolved into a sort of impersonal online communication. This way people communicate through platforms that in some case can be more influential than traditional word of mouth. The main reason is that written communication differs from an oral and face-to-face communication, giving a person more time to think about what he or she is reading and more time to evaluate the real meaning of the sentences. This process leads to a higher perceived credibility of the source and it enhances the consumer’s willingness to rely on online sources to gather the information needed (Berger, 2014).
So, the term WOM has evolved into a new and more innovative concept called electronic word of mouth (eWOM). EWOM communication refers exclusively to the online world, where consumers can share opinions and experiences with many other peers (Hennig-Thurau et al., 2004). The anonymity, and together with it the role of the sender, plays an important role in this type of communication (Sridhar and Srinivasan, 2012).
1.3 Online consumer reviews and their credibility
Online consumer reviews (OCRs) are a particular kind of eWOM: they are defined by Mudamby and Schuff (2010, p.186) as “peer-generated product evaluation[s] posted on company or third-party web sites” and by Hennig_Thurau et al. (2004) as any positive or negative statement about products, services and companies, written by consumers who want to inform and share opinions about their experiences to other readers (Park and Park, 2008).
Individuals search online to learn about new products and evaluate the different alternatives that websites provide before making decisions concerning the purchase of a good and they face several consumer reviews that cannot be ignored (Floyd et al., 2014). Grant et al. (2007), to underline the factors that affect online search behavior and understand how consumers evaluate a review, distinguish between information source utility, personal factors and product factors. Consumers pay a lot of attention to the reviews they find on the online platforms because they want to make the right choices and often use them as a point of comparison, becoming more and more important in the decision processes (Reichelt et al., 2014).
Credibility refers to the characteristics that make people believe and trust something or someone (Wathen and Burkell, 2002) and it is likely that a credible review is also perceived as a believable one (Fogg et al., 2001).
Everybody perceives and evaluates the content of a review in different ways. A review can be perceived as more or less credible by individuals for different reasons: some evaluate messages on their valence and level of sidedness, noticing positive, negative or neutral stimuli to that, depending on the positive or negative arguments that support the review; on the contrary, others may focus more on the quality of the arguments presented by the sender (Cheung et al., 2009). Relevant for this thesis is to understand how consumers perceive these arguments as credible.
Cheung et al. (2012) explain how message credibility is one of the main concepts in communications’ research and associate the term credibility to the degree to which people consider something as believable. From here the explanation mentioned above about the belief that a credible review is often referred to a believable review (Fogg et al., 2001). Furthermore, it is stated that a credible review is also considered as factual and a more accurate review is perceived as more credible (Cheung et al., 2012).
OCRs credibility can be influenced and determined by different factors. Those relevant for this study will be introduced in the next paragraph.
1.4 Factors influencing credibility
Antecedents to credibility have been studied already. Cheung et al. (2012) demonstrated the connection between credibility and review acceptance, defining key antecedents to review credibility such as argument strength, source credibility, confirmation of prior beliefs, recommendation rating and consistency. Moreover, many other characteristics appear clearly visible to the reader with the first impact: the structure of the text, the length of the text and the number of comments and star rating of the item. Paying more attention to the text, other elements become more evident, such as the argumentation used to justify a positive or negative comment, the reasoning used by the sender to explain concepts and the quality of the text and expressions used. Argumentation is a concept that can be linked to both quality and quantity aspects of a written communication and can be seen as a really persuasive and influential factor in the eyes of the reader (Lee et al., 2008). Only the quality aspect will be relevant though for this thesis. The higher the complexity of the concepts mentioned and combined together is, the higher the perceived argument quality provided by the sender is. Thus, argument quality is considered as one of the main factors that can increase the credibility of a review, raising the likelihood of a positive language expectancy variation (Jensen et al., 2013).
Then, another important factor that can impact OCRs credibility is the sender expertise with a product. Hovland, Janis, and Kelley (1953) developed a model to measure credibility, that included trustworthiness and expertise. The authors stated that a source that shows a higher level of these two components should have a greater impact and persuasion and be perceived as more credible.
Consumers are driven by conformity and social acceptance when making any kind of decision and they often behave in order to be accepted or liked by peers, after observing what others are doing. This theory has been also demonstrated with a conformity experiment (Asch, 1955).
The degree of the influence by other individuals depends on several factors such as age, gender (Hsieh et al., 2006), emotions, motivation (Arnold and Reynolds, 2003), susceptibility to social influence (Aral and Walker, 2012), involvement, commitment and familiarity (Gill et al., 1988). Through these factors people communicate and share opinions about their own experiences, creating a sort of interpersonal online communication through online consumer reviews (Berger, 2014).
When individuals are more susceptible to social influence, the role of the expert is very important and can positively or negatively affect the credibility of a review (Reichelt et al., 2013). It will be newsworthy then looking at the positive effect that a review written by an expert has on the OCR credibility when individuals are more susceptible to social influence.
Involvement is an important factor that can moderate the effect of the quality of the arguments on the credibility of the review. Previous research has proved indeed how these two concepts are strictly related and it will be interesting for this study to analyze the effect that argument quality has on credibility, when involvement is present.
1.5 Problem statement and research question
The dependent variable will be the OCR credibility that, as anticipated before, can be influenced by a multitude of factors. For this thesis two factors as independent variables will be chosen and expected to have an impact on credibility: argument quality and sender expertise. The reason for choosing these two concepts is the lack of previous research on the influence on review credibility of these variables combined together.
Given these variables, the following research question will be finally developed:
“How do sender expertise and argument quality influence the credibility of online consumer reviews, given the moderating role of involvement and susceptibility to social influence?”
1.6 Academic and managerial relevancy
The combination of the concepts presented in this paper makes this thesis unique. Although sender expertise and argument quality have been already used to measure credibility, these two concepts have never been used together as independent variables. Most researches focused on receiver expertise, rather than sender expertise. Moreover, involvement and susceptibility to social influence have been mostly utilized together with credibility as independent variables, while in this case their function is to moderate the effect of other two independent variables. Another difference that characterizes this thesis is that generally research focuses on the effects of both normative and informative social influence; although this difference has never been mentioned yet, the second chapter will explain the reasons why only informative social influence will be taken into account.
Nowadays companies have to consider the digital world as part of their future success and a better understanding of how to do that can be provided by focusing on new ways to attract consumers, spreading knowledge about products through electronic word-of-mouth and keeping a close contact with customers. Companies could indeed exploit online review to enhance the credibility of their offerings and together with it the credibility of the brand itself. Understand then how to make these reviews more credible and “powerful” for those who will read them and understand what are the primary characteristics that a customer is normally looking for should be a firm’s primary objective.
accessory that cannot be found in his golf club or his own country, this requested item can be easily ordered through online channels and delivered in few days.
Figure 1.1 shows a representation of a typical website specialized in golf reviews, that will be used as the main source for this study.
1.7 Structure of the thesis
2. Theoretical framework
This chapter will provide more insights and relevant literature about the explanatory variables and their impact on review credibility as dependent variable. The variables will be analyzed to build then the conceptual model presented in the last section of the chapter. The first independent variable will be sender expertise, more specifically the experience the sender has acquired over time about a specific product. The second independent variable concerns the argument quality, which reflects one of the main characteristics of the message written by the sender. The concepts of involvement and susceptibility to informational social influence as moderators will be introduced and explained. Each explanation of these variables will be followed by subsequent hypotheses. Finally, the conceptual model will constitute the last part of this chapter, representing graphically the relationship between the variables.
2.1 Independent variable: sender expertise
So far, the role of the expert as a sender has been considered a lot in previous research. When referring to the influence a single person could have on others via WOM communication, many authors found that experts are more influential than non-experts, and are sometimes seen more as negative opinion leaders (Bansal and Voyer, 2000; Leonard-Barton, 1985).
In fact, an expert is considered as more knowledgeable about certain products or services than a non-expert and, thus, ordinary people would consider their opinions as more credible (Bone, 1995). Many authors found consistency with previous research for this notion and concluded affirming that expertise is one of the main bases for credibility (Radighieri and Mulder, 2013).
Often, OCRs written by experts lead to a higher number of product sales. Floyd et al. (2014) compared a review written by an expert with a review written by a non-expert. The results supported the general theory that says that people with a higher expertise are perceived as more credible than non-experts and have a more persuasive and stronger impact on others. Thus, they are expected to have a positive effect on credibility.
Given the reasoning above, the following hypothesis will be developed:
2.2 Independent variable: argument quality
Previous research has shown how authors have tried to focus on both quantitative and qualitative aspect of a message. Lu et al. (2014), for instance, have elaborated a model with relative formulas to measure the comprehensiveness of the review, the readability and the objectivity of it under the quantitative aspect. For comprehensiveness the authors refer to the number of sentences in the text of a review and they state that long texts, containing more information, are expected to be more comprehensive. To assess the readability of a review a particular index able to measure the writing style was used. Lastly, objectivity was measured dividing the total number of objective sentences by the total number of sentences presented in a review.
Yet, the authors did not consider the quality of the arguments and focused only on quantity. The quality of the message has always been considered as one of the major influencing criterion for communication and persuasion (Miller and Levine, 1996; Slater and Rouner, 1996).
Petty and Cacioppo (1981) refer to argument quality as a subjective perception of the arguments presented in a message, that can be seen as strong or weak according to the reader’s evaluation. The results of their study showed that a message that contains stronger and more convincing arguments is expected to provide more positive responses.
A good argument quality is also enhanced by the information presented in the review. More informative texts enhance the credibility of web information: in the context of online consumer reviews, strong, convincing and good arguments, together with relevant information provided, will let the reader perceive the review as more credible (Sussman and Siegal, 2003; Cheung et al., 2009). Furthermore, senders can show both positive and negative aspects of a specific product or service in a review: the fact that a sender can focus on both aspects, making use of good and strong arguments to support his opinion, will result in less review derogation and thus strengthening the review credibility (Jensen et al., 2013).
The previous arguments lead then to the following hypothesis:
To conclude, it might be interesting to see if there is also an interaction effect between sender expertise and argument quality. The perceived credibility of the quality of the arguments presented to the reader can be influenced indeed by the degree of expertise with a product of the sender. A review with a weak argument quality can be perceived for instance as more credible when written by an expert.
For this reason, the hypothesis below will be tested:
H3: The positive effect of argument quality on review credibility is higher when the review is written by an expert.
2.3 Moderators
Two moderators were chosen for this thesis. Product involvement was chosen to moderate the effect of argument quality on an OCR credibility. The Elaboration Likelihood Model will explain better the link between involvement and argument quality. The second moderator, finally, concerns the receiver’s susceptibility to social influence and it was chosen to measure the effect of the sender expertise with a product on review credibility when receivers are more susceptible.
2.3.1 Involvement
Involvement refers to the personal relevance of a certain issue, the motivation and the importance an individual gives to a certain objective, in this case the topic of the review (Sussman and Siegel, 2003; Cheung et al., 2012). The level of involvement is based on the importance attributed to a product or service by a consumer (Laurent and Kapferer, 1985), based on individual needs, interests and values and it may vary depending on consumers’ characteristics and decisions that have to be taken. Individuals who are highly involved are more motivated to understand a message and base their evaluation on the quality of the arguments provided (Sussman and Siegel, 2003).
A person that presents more ability to process the information is more likely to be involved in the elaboration process when the quality of the arguments is good and, in fact, when readers are more interested and motivated to understand a message in a review, the central cue will have the greatest influence on their judgements (Cheung et al, 2012).
On the other hand those whose abilities are limited and that are less involved, focus more on non-content cues and follow the peripheral route. When the reader is not motivated or unable to pay more attention to the information presented, the peripheral cue has a more important role in influencing judgement; uninvolved individuals are not likely to elaborate in detail a message (Cheung et al., 2009). The results of the ELM experiment demonstrated that for high involvement consumers argument quality was fundamental for an attitude change and for a higher perceived credibility. High involvement consumers consider reviews with a strong argument quality and with a logical reasoning behind as more credible and reliable (Park et al., 2007).
This, translated in the OCRs case, means that the linear relationship between credibility and quality of the arguments presented in a review can be influenced by different levels of involvement.
Low involvement consumers are expected then to simply accept the argumentation presented in the review and believe it to be more credible, while high involvement ones are expected to seek and analyze as much as they can from the message in the review.
Finally, given these explanations, the arguments lead to the following hypothesis:
H4: The positive effect of argument quality on review credibility is higher when product involvement is high.
2.3.2 Susceptibility
It is important to analyze the concept of reference groups to understand why people behave in a certain way and, according to the theory of social comparison by Festinger in 1954, individuals tend to make comparison with others on different attributes to enhance their ideas and judgements. To make this comparisons, individuals use reference groups, comparing themselves with others who posses the same amount of knowledge about the attributes searched (Moschis, 1976).
Information can be gathered in two different ways: observing what most people are doing, or requesting the information from others (Mourali and Laroche, 2005). Informational influence can affect individuals’ decision processes concerning evaluations about products and brands (Pincus and Waters, 1977).
This kind of social influence is even stronger for consumers who are highly susceptible to the information they find on the web and thus, this is expected to increase the usage of online consumer reviews by a consumer and consequently their credibility (Bearden et al., 1989).
The reason for focusing on the informational social influence for this study is that it appears difficult to visualize the behavior of individuals in virtual communities and in OCRs (de Valck et al., 2009). Informational social influence will provide a better perspective to explain the impact of susceptibility. OCRs are present indeed in a community context where a lot of information can be provided (Zhang et al., 2010).
People normally tend to be more influenced by friends or well known faces than strangers, seeing them as more credible and reliable (Fennis, 2010). When reading an OCR, a consumer is indeed more likely to believe in the ones written by someone they directly or indirectly know rather than anonymous ones. Often people can recognize an expert when reading an OCR by only reading his name and then associate his name to a more credible review. For instance, when reading an OCR about golf, seeing that the review has been written by Tiger Woods can be relevant for the reader. The identity of the sender and his expertise with the product are expected then to be very important for highly susceptible individuals.
The subsequent hypothesis is then formed:
2.5 Conceptual model
Based on the theoretical framework, a conceptual model is provided below with figure 2.1. Here the two explanatory variables are represented: sender expertise and argument quality. These two variables are expected to have an impact on the OCR credibility, as the model graphically represents with the hypotheses explained previously (respectively through hypothesis 1 for sender expertise and hypothesis 2 for argument quality). Furthermore, the two moderators are shown: receiver’s product involvement for argument quality with hypothesis 4 and receiver’s susceptibility to informational social influence for sender expertise with hypothesis 5.
3. Methodology research
In this chapter the experimental setup of the empirical research for the OCRs in the golf industry will be discussed. The operationalization of the concepts introduced so far will be followed by the manipulation check of the independent variables. In the first part of the chapter a description of the research design, with number of participants and an overview of the conditions will be provided. Two examples representing two different conditions will be displayed with a screenshot that shows an OCR about golf. After that, the sampling frame and the procedure will be explained and the operationalization of the independent variables in relation with the dependent variable and the moderators will be conducted. To conclude, the last section will cover a plan for the data analysis.
3.1 Research design and participants
In order to get more insights about the variables that have an impact on review credibility, an experimental design with sender expertise (expert / non expert) and review argument quality (strong / weak) as independent variables will be conducted. The moderating variables that are expected to influence the effect of the independent variables on OCR credibility are respectively receiver’s susceptibility to informational social influence for sender expertise and receiver’s involvement for argument quality.
An experimental research design concerns the determination of a causal relation between a defined number of variables, where one (or more) are manipulated and other control variables are added (Malhotra, 2007). The experiment will be done using a 2x2 between participants design, with four conditions. The reason for not choosing a within participants design are sequence effects (such as confusion for the different conditions), order effects (interpretation of condition 2 in light of condition 1 and so on) and practice effects (concerning performance measures). Furthermore, a within participant experimental design would let the respondents spend more time than expected on the survey and then furnish responses that are not reliable for lack of concentration and loss of patience (Whitley and Kite, 2012).
Table 3.1 Overview of conditions.
Each condition requires at least 30 participants to be considered as reliable, so in total 120 respondents are needed.
3.2 Sample and procedure
The Italian and Dutch population of people who are interested in or practice golf will constitute the largest part of the sample, due to limited time and sources to gather the data. According to the specific case of OCRs, respondents that are more interested in online purchasing and comparing reviews and products will be selected.
The sample frame will be then constituted by golf player or people who have previous experience with golf and are more willing to look for products in the online world. Because of the large use of technology by young people, the average age of the sample is supposed to be relatively young. Furthermore, although the number of women golf players is increasing, this sport is still mainly practiced by men, and for this reason a larger number of male respondents is expected.
Data will be gathered through an online survey. The digitalization will help both the author and the respondents to provide a more appropriate study and a better organization of the online data, time savings for the fulfillment of the questionnaire in terms of time spent to look for the ‘right’ answers and less time pressure for finishing it (the respondent will be indeed free to close and restore a session whenever it will be more advantageous for him).
A simple welcome page will be followed by a short introduction about the topic, the organization of the survey and types of questions the respondent will see. After the welcome page, a graphic representation of an online consumer review on golf equipment (specifically a golf driver) will appear. The review itself is taken from the website www.golfonline.co.uk, which is specialized in golf products and where many consumer reviews are available. Yet, the text of the review will be self-made and other elements such as price, sponsors and similar items will be deleted in order to avoid confusion. Respondents might identify the degree of experience of the sender only looking at
Conditions
Expert sender
Non-expert sender
Strong argument quality Condition 1 Condition 3
Figure 3.1 and figure 3.2 provide a graphic representation of two of the four conditions presented in the survey, respectively Expert - Strong argument quality and Non-expert - Weak argument quality. The other two conditions are displayed in Appendix 1, with figure 1 and figure 2.
Figure 3.1 Expert sender - Strong argument quality condition. (Source: adapted by www.golfonline.co.uk)
Figure 3.2 Non-expert sender - Weak argument quality condition. (Source: adapted by www.golfonline.co.uk)
After the picture of the review, nine questions will be asked to the participants. An overview of all the questions is presented in appendix 4. The first set of questions will be about the credibility of the review itself, followed by questions for the manipulation check, concerning the level of expertise of the sender and the quality of the arguments given. The questions to test the manipulation are done in order to understand if people actually perceive the different degree of expertise and argument quality in the conditions they are disposed to. The third set of questions dedicated to the experiment will be focused on involvement and degree of susceptibility to informational social influence. Finally, respondents will be asked to complete the survey filling in
reviews. The survey will end with a thankful page for the participation. It will be in English and distributed online with Qualtrics.com. Participants were principally collected by email and electronic-word-of-mouth through social networks and results will be anonymous.
3.3 Operationalization and reliability of scales
To test the hypotheses, the concepts mentioned before should be operationalized under different dimensions and scales used previously. Table 3.2 provides an overview of all the concepts and relative items, to have a clear picture of the scales used for the concepts presented.
3.3.1 Operationalization
Review credibility has already been operationalized by Cheung et al. (2012). The authors measured it with four different items on a 5-point Likert scale (“strongly disagree - strongly agree”): review believability, factuality, accuracy and credibility.
The manipulation check for the expertise was tested with a scale provided by Susan and Siegal (2003), that used two items on a 7-point Likert scale (respectively “novice - expert” and “not at all - to a great extent”). The scale has been readapted because the authors with their research wanted to measure the effect of argument quality on the perceived message usefulness, using recipient’s expertise as moderator; in this case questions will be asked and focused on the sender’s expertise.
Argument quality has been investigated before by several authors and, to measure it in this empirical research and for the manipulation check, the scale used by Cheung et al. (2012) has been chosen. The authors adapted the scale from Zhang (1996) and measured argument quality on a 5-point Likert scale (strongly disagree - strongly agree) with five items, in order to define the arguments as convincing, strong, persuasive, good, together with the informativeness of the review. The fourth concept regards receiver’s product involvement as moderator and will be measured on a a 5-point Likert scale (strongly disagree - strongly agree) with three items, using the scale provided by Cheung et al. (2012).
Concept Item Scale Cronbach’s alpha Dependent variable Review credibility Source: Cheung et al. (2012)
• I think the review is believable
• I think the review is factual
• I think the review is accurate
• I think the review is credible
5-point Likert: Strongly disagree - Strongly agree 0,860 Independent variables Sender expertise Source:
Adapted from Susan and Siegal (2003)
• How informed is the writer on the subject matter of this issue?
• To what extent is the writer an expert on the topic of this review?
7-point Likert:
Novice - Expert
Not at all - To a great extent
0,944
Argument quality Source: Cheung et al. (2012)
• The arguments in the review are convincing
• The arguments in the review are strong
• The arguments in the review are good
• The review is informative
5-point Likert: Strongly disagree - Strongly agree 0,892 Moderators Receiver involvement Source: Cheung et al. (2012)
• I was very involved in the topic of this review
• It was important for me to get information from this review
• I am interested in the topic of this review
5-point Likert:
Strongly disagree - Strongly agree
Table 3.2 Operationalization table.
3.3.2 Reliability of scales
To access the reliability of all concepts, unidimensionality was tested with a factor analysis with a Varimax rotation method, after which internal consistency was measured.
A factor analysis is appropriate if the value of Kaiser-Meyer-Olkin (KMO) is higher than 0,50 and if Bartlett’s sphericity test results significant. Communalities of items should be higher than 0,50 and factor loadings that show values higher than 0,50 are considered as acceptable (Malhotra, 2009).
Factor analysis showed unidimensionality for all scales (see Appendix 2, table 1 for output). A second factor analysis for the individual factors was run afterwards, followed by a reliability analysis. The Cronbach’s alphas were high for all (ranging from 0,824 to 0,944), although for receiver’s susceptibility to informational social influence, deleting the first item (“To make sure I buy the right product or brand I often observe what others are buying”), resulted in an increase of it (from 0,824 to 0,852). As a consequence, this item was left out for the next steps. The results are presented in Appendix 2, table 2.
After measuring internal consistency with Cronbach’s alpha values, a further step was made to compute variables, by calculating the average of the items within the variable to have one single construct, and then proceed with the manipulation check.
Receiver susceptibility to informational social
influence Source: Bearden et al. (1989)
• To make sure I buy the right product or brand I often observe what others are buying
• If I have little experience with a product I often ask my friends about a product
• I often consult other people to help choose the latest
alternative available from a product class
• I frequently gather information from friends or family about a product before I buy
3.4 Manipulation of independent variables
In order to assess the validity of the model a manipulation check for the independent variables was conducted. Table 3.3 shows an overview of the results given after manipulating both sender expertise and argument quality.
Table 3.3 Overview of manipulation test.
Sender expertise was measured on a 7-point Likert scale, with a neutral value of 4. P-value of expertise is 0,000, with a t-value of 8,653. The means show indeed that when respondents were exposed to an expert condition, they perceived the level of expertise as higher, with a mean value of 5,6456 and a standard deviation of 1,4723. When they were exposed to a non-expert condition the mean resulted to be under the neutral level (3,6000) with a standard deviation of 1,4029, resulting then in a lower perceived level of expertise.
Argument quality was measured on a 5-point Likert scale, with a neutral level of 3. Table 3.3 shows the significance level of argument quality of 0,001 and with a t-value of 3,405. In this case the mean when participants were exposed to a strong quality of the arguments resulted 3,7380 with a standard deviation of 0,0813, while when exposed to a weak argument quality the mean is 3,3026, with a standard deviation of 0,0970. This shows that participants in a strong argument quality condition perceived the quality of the arguments as higher than those who were exposed to a weak argument quality condition, although the difference is not as much larger as in the expertise case.
Manipulation check Mean St. Dev. t-value p-value
3.5 Plan of analysis
After the factor and reliability analyses a descriptive of the sample and conditions will be provided, in order to have an overview of the gender, age, country and use of OCRs between the participants.
After comparing the means and evaluate the differences between the four groups, a multiple regression will be done with OCR credibility as dependent variables, sender expertise and review argument quality as independent variables and with respectively susceptibility to interpersonal social influence and product involvement as moderators. Five models in total will be compared. The first model will have only review argument quality and sender expertise as independent variables. An interaction effect, computing argument quality and sender expertise, will be added to the second model. Then, involvement as moderator and the interaction between involvement and argument quality will be added for the third model. The fourth model will include susceptibility as moderator and the interaction between susceptibility and sender expertise. Finally, for the last model all the variables mentioned before will be combined together. This model will be composed then by argument quality, sender expertise, interaction between argument quality and sender expertise, involvement, interaction between involvement and argument quality, susceptibility and interaction between susceptibility and sender expertise.
In order to interpret and estimate correctly the models it is important to do a check for multicollinearity. In case of moderate or high multicollinearity, the regression model(s) that present(s) it will be interpreted with caution and then multicollinearity will be solved with a median split on moderators, followed by a two-way ANOVA. After the check for multicollinearity the interpretation of the results will be possible and the estimation of all models will be done. The interpretation of the tables will be followed then by the assessment of the best model, done by comparing the adjusted R2 of the models. To conclude, the hypotheses validation will be done and
4. Results
In this chapter, the results of the empirical research will be presented. Descriptive statistics will open the chapter, then a comparison between the means of the four conditions will be provided, to measure possible significant variations between the groups. Then, five models will be compared with a multiple regression analysis, followed by an ANOVA to solve multicollinearity problems. Finally, an overview of the hypotheses with their results will be given.
4.1 Descriptives of the sample
Although 238 people started the survey, only 151 completed it. The results in table 4.1 show a higher and unexpected percentage of a female population 53% and a male percentage of 47%, with respectively 80 female respondents and 71 male respondents. As expected, the average age of the sample was young, ranging from 14 and 60. The survey was completed mainly by Italian (58%) and Dutch (28%) respondents (Appendix 3, table 3 for output).
The results given by the use of online consumer reviews show that the mean provided by the samples was 4,55, with a minimum level of 1, a maximum of 7 and a neutral of 4,. It shows that the respondents who took the survey do make use of online consumer reviews.
Table 4.1 Descriptives of the sample.
Table 4.2 provides an overview of the descriptives per condition. In the table are visible the differences between the four conditions, according to gender and age, with the respective means. All groups are quite homogeneous and there is no significant variation between the groups.
The first group has exactly the same number of male and female participants, while the larger difference about the age is present in the third group, with 16 male respondents and 21 female
Variable Results
Gender (n=151) Male: 71 (47%)
Female: 80 (53%)
Age (n=151) Mean: 24
Range: 14 - 60
Use of online consumer reviews (n=147) Mean: 4,55
Table 4.2 Descriptives per condition.
Table 4.3 shows the ANOVA table where the different means of the OCRs use are displayed. Even in this case, means are quite close to each other. The average mean is indeed 4,55, with the highest value of 4,71 for group 2 and the lowest one for group 1 with a value of 4,40. The p-value is 0,871, with a F-value of 0,236. No significant variations then is present when comparing the four groups.
Group Gender N Age mean Std.
Deviation t-test P-value 1 Expert sender - Weak argument quality Male Female 21 21 24,10 24,90 2,448 7,133 -0,493 -0,493 0,625 0,626 2 Non expert sender- Strong argument quality Male Female 16 19 23,94 25,16 1,482 7,411 -0,646 -0,701 0,522 0,491 3 Expert sender - Strong argument quality Male Female 16 21 22,88 25,43 1,996 7,352 -1,348 -1,520 0,186 0,142 4 Non-expert sender - Weak argument quality Male Female 18 19 25,67 23,79 9,204 1,803 0,871 0,849 0,389 0,407
Group Use of ocr mean Std. Deviation F-stat p-value
4.3 Model selection
Variance Inflation Factors (VIF) indicate the presence of multicollinearity. Values between 4 and 10 indicate moderate multicollinearity and values higher than 10 indicate strong multicollinearity (Myers, 1990). The results showed moderate multicollinearity for model 3 and high multicollinearity for model 4 and 5 (Appendix 3, table 4 for output).
The models in table 4.4 will be then interpreted with caution and then a solution to solve multicollinearity will be provided.
n.s.= not significant, *p < .10, **p < .05, ***p < .01 Table 4.4 Regression models.
The first base model contains the main effect of the independent variables on the dependent variable and it is overall significant, with p=0,001 and F=7,801. Both coefficients are highly significant and this means that both sender expertise and argument quality have a positive significant effect on OCR credibility.
For the second model the interaction effect between sender expertise and argument quality is added. This model is overall significant, with p-value=0,000 and F-value=6,669. The interaction between sender expertise and argument quality results to be the only significant result, so there is apparently no significant effect of sender expertise or argument quality on OCR credibility.
The main effects of the first moderator (involvement) and the interaction effect with argument quality and receiver involvement are added to the third model. For the fourth model, susceptibility with main effect and interaction effect with sender expertise are added to the variables presented in the second model. Finally, the fifth model contains all the main effects and interaction effects presented in the previous models.
However, model 3, 4 and 5 present moderate and high multicollinearity and thus it results to be difficult to interpret the results and estimate the best model with the highest R2 adjusted (that
seems to be the third one, with a value of 0,195).
n.s.= not significant, *p < .10, **p < .05, ***p < .01 Table 4.5 ANOVA table with corrected models.
The first model is overall significant, with p=0,001 and F=7,801. Both variables, sender expertise and argument quality are highly significant.
The second model is overall significant, with p=0,000 and F=6,669. Even in this case sender expertise and argument quality, together with their interaction added to the model, are highly
Model 1 2 3 4 5
Involvement and the interaction between involvement and argument quality are added for the third model. The model is overall significant, with a p=0,000 and F=6,216. In this case, the interaction effect between involvement and argument quality is not significant, while all other variables remain significant.
In the fourth model, receiver susceptibility and its interaction with sender expertise are added to the variable of the second model. The model is overall significant with p=0,001 and F=4,681. As for model three, the interaction effect of the moderator added is not significant.
The fifth and last model presents all the variables of the previous models combined together and results overall significant, with p=0,000 and F=4,649. Sender expertise, argument quality, their interaction and involvement result significant, while the interaction between involvement and argument quality, susceptibility and interaction between susceptibility and sender expertise are all not significant.
To estimate the best model and test then the hypotheses a model comparison has to be done and the model with the highest R2 adjusted will be interpreted. As the multiple regression run before
showed, the best model in this case is the third model, that has a R2=180 and R2 adjusted=151,
The interaction effect between sender expertise and argument quality is significant, with a p=0,051 and F=3,885. The positive direction is showed in figure 4.3. The plot shows that the credibility of the argument presented in a review is higher when the sender is an expert.
Involvement results highly significant, with a p=0,003 and F=9,134, although the interaction between involvement and argument quality is not significant. The direction of the interaction is shown in figure 4.4.
Finally, receiver susceptibility is not present in the third model. Adding susceptibility and its interaction with sender expertise would not increase the R2 and in both model 4 and model 5 the
interaction effect results not significant. The direction of the interaction between receiver susceptibility and sender expertise are presented in figure 4.5.
4.4 Hypotheses validation
The results of the hypotheses made in chapter 2 will be summarized in table 4.6.
Table 4.6 Hypotheses validation table.
Hypothesis Result
H1: The level of expertise with the product of the sender has a positive effect on review credibility.
Accepted
H2: The argument quality of an OCR has a positive effect on review credibility.
Accepted
H3: The positive effect of argument quality on review credibility is higher when the review is written by an expert.
Accepted
H4: The positive effect of argument quality on review credibility is higher when product involvement is high.
Rejected (No significant effect)
H5: The positive effect of the sender’s expertise with a product on review credibility increases when receivers are more susceptible to informational social influence.
5. Conclusion
This final chapter will provide a general discussion of the results obtained in chapter 4, together with an explanation of the hypotheses. The second and final part will present then limitations of this research and suggestions for further research.
5.1 Discussion
In this study, the effect of expertise and quality of the arguments on the credibility of online consumer reviews was examined. The moderation effect was given by two other variables: susceptibility to informational social influence for sender expertise and receiver involvement for argument quality.
Consistent with current literature, argument quality has a significant positive effect on the review credibility (Cheung et al., 2012). All the models show indeed that argument quality is highly significant in all cases.
Sender expertise also results to have a significant positive effect on credibility for all models, as previous research has shown before. Dholakia and Sternthal (1977) focus indeed on the role of the expert and concluded stating that, since experts are considered as more credible than non-experts, the information they provide is perceived as more credible. In this study the interaction effect between argument quality and sender expertise has been considered and the results show a significant and higher positive effect of argument quality on review credibility when the review is written by an expert. This also means that a review that presents a weak argument quality is perceived as more credible when written by an expert, than when it’s written by a non-expert.
Receiver product involvement was expected to have a positive impact on the effect of argument quality on review credibility. Although the main effect of involvement on OCR credibility is highly significant, the results show no significant effect for the interaction between involvement and argument quality on credibility. So the positive effect of argument quality on review credibility is not higher when receiver product involvement is high.
review written by an expert doesn’t increase when the receiver is more susceptible to informational social influence.
5.2 Limitations and further research
Although the study provided interesting results, it also has several limitations. For instance, concerning the sample, a relatively small number of participants was considered and only the Italian and Dutch population constituted the largest part of it. It might be interesting for further research to analyze and include the differences between countries concerning OCRs use for example and have a wider range to see the main variations. Another implication is that golf is practiced more in certain countries than others, because of social reasons or lack of economic resources, and thus, this will probably lead to different results. Less developed countries are not willing to make a large use of OCRs or play golf and then, probably, a more interesting study would concern the differences between continents, such as Europe vs. USA, where golf is mostly practiced. With a more extensive research and no time or source limitations this could be possible for further researches.
Moreover, although young consumers are the ones who are more willing to make use of online consumer reviews because of the new internet era, the average age was relatively young. For the variables used in this study indeed, the effect of factors such as involvement or susceptibility can change according to many reasons: previous experiences, educational background, social status and income (Gill et al., 1988; Hsieh et al., 2006). Including an older population sample could have provided different results.
Furthermore, this study was only focused on people who belong to the golf industry and, thus, a relatively restricted sample. Golf requires concentration and determination and the level of involvement with a product might be perceived differently for a golf player than for someone else, as well as the degree of susceptibility.
References
Aral, Sinan and Walker, Dylan (2012), “Identifying Influential and Susceptible Members of Social Networks”, Science, 337 (July), 337-341.
Arnold, Mark J. and Reynolds, Kristy E. Reynolds (2003), “Hedonic shopping motivations”,
Journal of Retailing, 79 (2), 77-95.
Asch, Solomon E. (1955), “Opinions and social pressure”, Scientific American, 193 (5), 31-35.
Bansal, Havir S. and Voyer, Peter A. (2000), “Word-of-mouth processes within a services purchase decision context”, Journal of Service Research, 3 (2), 166–177.
Bearden, William O., Netemeyer, Richard G. and Teel, Jesse E. (1989), “Measurement of consumer susceptibility to interpersonal influence”, Journal of Consumer Research, 15 (March), 473-81.
Berger, Jonah (2014), “Word of mouth and interpersonal communication: A review and directions for future research”, Journal of Consumer Psychology, 24 (4), 586.
Bone, Paula F. (1995), “Word-of-mouth effects on short-term and long-term product judgments”, Journal of Business Research, 32 (3), 213–223.
Cheung, Cindy M. Y., Sia, Choon-Ling and Kuan, Kevin K. Y. (2012), “Is This Review Believable? A Study of Factors Affecting the Credibility of Online Consumer Reviews from an ELM Perspective”, Journal of the Association for Information Systems, 13 (8), 618-635.
Cheung, Cindy M. Y., Luo, C., Sia, Choon-Ling and Chen, H. P. (2009), “Credibility of electronic word-of-mouth: Informational and normative determinants of online consumer recommendations”. International Journal of Electronic Commerce, 13(4), 9-38.
Cronin, Mary J. (1997), “Global Advantage on the Internet: From Corporate Connectivity to International Competitiveness”, 1st, John Wiley & Sons, (New York, NY, USA).
Deutsch, Morton and Gerard, Harold B. (1955), “A study of normative and informational influence upon individual judgment”, Journal of Abnormal and Social Psychology, 51 (November), 629-36.
de Valck, Kristine, van Bruggen, Gerrit H. and Wierenga, Berend (2009), "Virtual Communities: A Marketing Perspective”, Decision Support Systems, 47 (3), 185-203.
Dholakia, Uptal M., Bagozzi, Richard P. and Pearo, Lisa Klein (2004), “A social influence model of consumer participation in network-and-small-group-based virtual communities”,
International Journal of Research in Marketing, 21 (3), 241-263.
Fennis, Bob M., and Stroebe, Wolfgang (2010), “The psychology of advertising”,
Psychology Press. Hove.
Floyd, Kristopher, Freling, Ryan, Alhoqail, Saad, Cho, Hyun-Young and Freling, Traci (2014), “How Online Product Reviews Affect Retail Sales: A Meta-analysis”, Journal of Retailing, 90 (June) (2), 217-232.
Fogg, B. J., Marshall, Jonathan L., Laraki, Othman, Osipovich, Alex, Varma, Chris and Fang, Nicholas (2001), “What makes Websites credible? A report on a large quantitative study”. Conference on Human Factors in Computing Systems. New York: Association for Computing Machinery, 61-68.
Grant, Robert, Clarke, Rodney J., and Kyriazis, Elias (2007), “A review of factors affecting online consumer search behavior from an information value perspective”, Journal of Marketing
Management, 23 (5-6), 519-533.
Guo Guoqing, Yang Xuecheng (2006), “Word-of-mouth marketing and its application in the era of Internet”, Finance and Trade Economic, 9, 58-61.
Hennig-Thurau, Thorsten, Gwinner, Kevin P., Walsh, Gianfranco and Gremler, Dwayne D. (2004), “Electronic Word-of-Mouth via Consumer-Opinion Platforms: What Motivates Consumers to Articulate Themselves on the Internet?”, Journal of Interactive Marketing, 8 (1), 38-52.
Hovland, Carl I., Janis, Irvig and Kelley, Harold H. (1953), “Communication and Persuasion”. Psychological Studies in Opinion Change, New haven: Yale university Press.
Hsieh,Yi-Ching, Hung-Chang Chiu, Chia-Chi Lin (2006), “Family communication and parental influence on children's brand attitudes”, Journal of Business Research, 59 (September), 1079–1086.
Jensen, Matthew J., Averbeck, Joshua M., Zhang, Zhu and Wright, Kevin B. (2013), “Credibility of Anonymous Online Product Reviews: A Language Expectancy Perspective”, Journal
of Management Information Systems, 30 (1), 293-323.
Jolson, Marvin A., Bushman, Anthony F. (1978), “Third-party consumer information systems: the case of the food critic”, Journal of Retailing, 54 (4), 63–79.
Katona, Zsolt and Sarvary, Miklos, (2014), “B2B Social Media--"It's Communication, not Marketing””, California Management Review, (Spring), 56 (3), 142-156.
Lee, Jumin, Park, Do-Hyung and Han, Ingoo (2008) “The effect of negative online consumer reviews on product attitude: An information processing view”, Electronic Commerce
Research and Applications, 7 (June), 341–352.
Lee Hee, Law, Rob and Murphy, Jamie (2011) “Helpful reviewers in TripAdvisor, an online travel community”, Journal of Travel & Tourism Marketing, 28(7), 675–688.
Leonard-Barton, Dorothy (1985) “Experts as negative opinion leaders in the diffusion of a technological innovation”, Journal of Consumer Research, 11 (4), 914–926.
Lim, Boon C. and Chung, Cindy M. Y. (2011), “The impact of word-of-mouth communication on attribute evaluation”, Journal of Business Research, 64, 18–23.
Lohse, Gerald, L., Bellman, Steven and Johnson, Eric J. (2000) “Consumer Buying Behavior on the Internet: Findings from Panel Data”, Journal of Interactive Marketing, 14, (1), 15-29.
Lu, Yingda, Jerath, Kinshuk and Singh, Param Vir, (2013), “The Emergence of Opinion Leaders in a Networked Online Community: A Dyadic Model with Time Dynamics and a Heuristic for Fast Estimation”, Management Science, 59 (8), 1783-1799.
Malhotra, Naresh K. (2007), “Marketing research: an applied orientation”. New Jersey, NJ: Pearson Education.
Miller, David M. and Levine, T. R. (1996). “Persuasion. In M. B. Salwen & D. N. Stack (Eds.), “An integrated approach to communication theory and research””. Mahwah, New Jersey: Lawrence Erlbaurn Associates, 261-276.
Moschis, George P. (1976), "Social Comparison and InformalGroup Influence”, Journal of
Mourali, Mehdi, Laroche, Michel and Pons, Frank (2005), "Individualistic orientation and consumer susceptibility to interpersonal influence", Journal of Services Marketing, 19 (3), 164-173
Mudambi, Susan M., and Schuff, David (2010), “What makes a helpful online review? A study of customer reviews on amazon.com”. MIS Quarterly, 34 (March),185–200.
Myers, Raymond H. (1990), “Classical and Modern regression with Applications”, 2nd Edition, Boston, MA: Duxbury.
Park, Whan C. and Lessig, Parker V. (1977), “Students and housewives: differences in susceptibility to reference group influence”, Journal of Consumer Research, 4 (September), 102-10.
Park, Do-Hyung, Lee, Jumin and Han, Ingoo (2007), “The effect of on-line consumer reviews on consumer purchasing intention: the moderating role of involvement”, International
Journal of Electronic Commerce, 11 (4), 125–148.
Park, Do-Hyung and Park, Se-Bum (2008), “The Multiple Source Effect of Online Consumer Reviews on Brand Evaluations: Test of the Risk Diversification Hypothesis”. Advances in Consumer Research, 35, 744-745.
Petty, Richard E. and Cacioppo, John T. (1981), “Attitudes and Persuasion: Classic and Contemporary Approach”. Dubuque, IA: William C. Brown.
Petty, Richard E. and Cacioppo, John T. (1986), “The Elaboration Likelihood Model of Persuasion”, Advances in experimental social psychology, (19), 673-675.
Pincus, Steven and Waters, L. K. (1977), "Informational Social Influence and Product Quality Judgments," Journal of Applied Psychology. 62 (5), 615-619.
Reichelt, Jonas, Sievert, Jens and Jacob, Frank (2014) “How credibility affects eWOM reading: The influences of expertise, trustworthiness, and similarity on utilitarian and social functions”, Journal of Marketing Communications, 20 (1-2), 65-81.
Riegner, Cate (2007), “Word of mouth on the web: the impact of web 2.0 on consumer purchase decisions”, Journal of Advertising Research, 47(4), 436-447.
Rigby, Darrell (2011), “The Future of Shopping”, Harvard Business Review, (December), 65-76.
Robson, Karen, Farshid, Mana, Bredican, John and Humphrey, Stephen (2013), “Making sense of online consumer reviews: a methodology”, International Journal of Market Research, 55 (4).
Senecal, Sylvain and Nantel,, Jacques (2004), “The influence of online product recommendations on consumers’ online choices”, Journal of Retailing, 80 (2), 159-169.
Slater, Michael D. and Rouner, Donna (1996), “How message evaluation and source attributes may influence credibility assessment and belief change”, Journalism and Mass
Communication Quarterly, 73(4), 974–992.
Sridhar, Shrihari and Srinivasan, Raji (2012), “Social Influence Effects in Online Product Ratings”. Journal of Marketing, 76 (5), 70-88.
Sussman, Stephanie W. and Siegal, Wendy S. (2003), “Informational Influence in Organizations: An Integrated Approach to Knowledge Adoption”, Information Systems Research, 14 (1), 49-65.
Tsang, Alex S. L. and Prendergast, Gerard (2009), “Is a ‘star’ worth a thousand words?”,
European Journal of Marketing, 43, 1269–1280.
Tseng, Shawn, Fogg, B.J. (1999), “Credibility and computing technology”, Communications
Wathen, Nadine C. and Burkell, Jacquelyn (2002), “Believe it or not: Factors influencing credibility on the Web”, Journal of the American Society for Information Science and Technology, 53(2), 133-144.
Whitley, Bernard E. Jr and Kite, Mary E. (2012), “Principles of Research in Behavioral Science: Third Edition”. New York, Taylor & Francis.
Zhang, Yong (1996), “Responses to humorous advertising: The moderating effect of need for cognition”, Journal of Advertising, 25 (1), 15-31.
Zhang, Jason Q., Craciun, Georgiana and Shin, Dongwoo (2010), “When does electronic word-of-mouth matter? A study of consumer product reviews”, Journal of Business Research, 63 (12), 1336-1341.
Zhang, Kem Z.K., Lee, Matthew K.O. and Zhao, Sesia J. (2010), “Understanding the informational social influence of online review platforms”. ICIS 2010 Proceedings. Paper 71.
Appendix
Appendix 1. Conditions displayed to participants (not present in the text)
Appendix 2. Factor and reliability analyses
Component
Item 1 2 3 4
I think the review is factual
0,758 0,038 0,166 0,123
I think the review is accurate
0,730 0,091 0,052 0,283
I think the review is credible
0,833 0,142 0,219 0,027
How informed is the writer on the subject matter of this issue?
0,325 0,017 0,007 0,902
To what extent is the writer an expert on the topic of this review?
0,329 -0,019 0,082 0,883
The arguments in the review are convincing
0,770 0,117 0,130 0,287
The arguments in the review are strong
0,702 0,042 0,105 0,366
The arguments in the review are persuasive
0,638 0,035 0,093 0,411
The arguments in the review are good
0,797 0,062 0,139 0,256
The review is informative
0,776 0,125 -0,006 0,150
I was very involved in the topic of this review
0,242 0,086 0,879 0,022
It was important for me to get
information from this review
0,192 0,097 0,908 0,046
I am interested in the topic of this review
0,073 0,122 0,901 0,057
To make sure I buy the right product or brand I often observe what
