Electronic WOM
How do online reviews influence consumers?
Hilde Kiers
Electronic WOM
How do online reviews influence consumers?
Author Hilde Kiers De Venne 28 9411 BW Beilen 06-24850382 h.r.kiers@student.rug.nl 1830392 University of Groningen
Faculty of Economics and Businesses MScBA Marketing Management Master Thesis
November 2010
PREFACE
During the six months I wrote this thesis I realized that my career as a student will terminate soon. The aim was to finish the marketing study and at the moment I am writing the last sentences of this thesis I can say that I reached this aim beyond my expectations with full dedication, joy, and with feelings of proud in particular. I had the opportunity to deepen myself into a topic which is of personal interest and plays a major role in business nowadays.
Therefore I have chosen this topic since I expect to participate with it during my career and I acknowledge the importance of it within the current online environment. Besides the fact that the topic fascinates me, I have broadened my view and knowledge in this specific field. Additionally, although this study is based on theory only, the relation with the practical business has been very clear and understandable.
Many thanks is devoted to dr. W. Jager who provided me with great support and, besides his enthusiasm for this topic, gave me helpful feedback based on his expertise knowledge. Furthermore, I would like to thank my friend Gijs Leffers for his listening ear and patience at several frustrated moments which were not that stimulating but which belongs to the process of writing a thesis.
SUMMARY
Nowadays online Word-of-Mouth communication is becoming more and more important. This research investigates this importance specifically for online consumer reviews, which are written by consumers after experiencing a product or service. Earlier research explored that this form of WOM might affect the purchase behavior of consumers. However not only products have been reviewed online, this purchase type played a major role in earlier studies. This study therefore focuses on service reviews and compared differences with product reviews. More specifically, we have focused on search goods, experience goods, and credence goods, from which the latter two are services.
This research on online consumer reviews for products and services has been limited to two important factors which are the level of trustworthiness and the recommendation framing of a review. These factors are derived from earlier research which has found that trustworthiness and recommendation framing will influence the opinion of consumers and, therefore, their purchase behavior as well.
The findings emphasize the importance of online consumer reviews for services compared with products. Moreover, the level of trustworthiness of reviews will differently influence the purchase intention products and services. The second factor, recommendation framing, showed to have less than expected influence on the trustworthiness of the reviews and therefore purchase intentions of consumers. An affect have been found in that consumers doubt and reconsider their purchase intention more after reading a negative review of a credence good instead of a negative review about an experience good. Possibly causes of the founded difference between products and services are the risky characteristic of services and the fact that services are hard or impossible to evaluate after experiencing. Consumers thus could rely more on service reviews, since they are often the only information source.
TABLE OF CONTENTS
PREFACE ... 3
SUMMARY ... 4
1. INTRODUCTION ... 7
1.2. Word-of-Mouth ... 7
1.3. Services versus products ... 8
1.4. Problem statement & research questions ... 9
1.5. Relevance ... 10 1.5.1. Theoretical relevance ... 10 1.5.2. Managerial relevance ... 10 2. THEORETICAL FRAMEWORK ... 12 2.1. Online reviews ... 12 2.2. Trustworthiness of reviews ... 14 2.3. Recommendation framing ... 17 2.4. Conceptual model ... 19 2.5. Social influence ... 20 3. METHODOLOGY ... 21 3.1. Service domain ... 21 3.2. Research design ... 22 3.3. Procedure ... 23 3.4. Data collection ... 24 4. RESULTS ... 25 4.1. Sample ... 25 4.2. Manipulation check ... 25 4.3. Scale reliability ... 26
4.3.1. Scale reliability for social influence ... 26
4.3.2. Correlation social influence ... 28
4.3.3. Scale reliability eWOM variables ... 30
4.4. Hypotheses ... 31
4.5. Additional analysis ... 34
5. DISCUSSION ... 36
5.1. Conclusion ... 36
5.2. Implications for managers & researchers ... 39
6. LIMITATIONS & FURTHER RESEARCH ... 40
6.1. Limitations ... 40
6.2. Further research ... 40
APPENDIX A: FACTOR ANALYSIS SOCIAL INFLUENCE ... 46
Appendix A.1. Initial eigenvalues ... 46
Appendix A.2. Scree plots social influence ... 47
Appendix A.3. Factor loadings ... 48
APPENDIX B: FACTOR ANALYSIS SOCIAL INFLUENCE TOGETHER ... 50
Appendix B.1. Initial eigenvalues ... 50
Appendix B.2. Scree plot ... 51
Appendix B.3. Rotated component matrix ... 51
APPENDIX C: FACTOR ANALYSIS EWOM VARIABLES ... 52
Appendix C.1. Initial eigenvalues ... 52
Appendix C.2. Scree plot ... 52
Appendix C.3. Factor loadings ... 52
1. INTRODUCTION
“Human voice. Clear, honest and accurate communication. Word-of-mouth. Use it” Nancy White
The quote stated above indicates the enormous power of human voice and human contact. Word-of-mouth marketing is a broad type of human communication and is still highly valued by marketers. This study aims to gain insight into how this form of marketing exactly works in the field of online reviews.
1.2. Word-of-Mouth
Currently, the introduction of the Internet has made it possible to communicate in an online environment where a new – and more powerful – form of WOM has taken its place. Thorsten Hennig-Thurau et al. (2004) defines electronic WOM (eWOM) as “any positive or negative statement made by potential, actual, or former customers about a product or company, which is made available to a multitude of people and institutions via the Internet”. The fact that WOM was person-to-person communication does not always hold anymore in the field of eWOM, because the interactivity can be interrupted by time. Examples of eWOM are among other things electronically based forums (brands Websites such as www.nokia.forum.com and retailers Websites such as www.amazon.com), bulletin boards, newsgroups, personal blogs, social networking sites (such as Hyves and Facebook), and chat rooms. Research by Zheng et al. (2009) discovered that online comments had an impact on potential customers; 55% of readers indicated that they find the online comments useful and will probably take this into account when making purchase decisions.
The focus of this research is to concentrate on a specific form of eWOM, namely online reviews written by other consumers after using or buying a form of a product or service. Consumers read experiences and opinions from previous users from that specific product or service on Websites in order to attain information before purchasing. EWOM is a form of informational influence because you can read others expertise and have the opportunity to share experiences with the product or service.
1.3. Services versus products
mind (e.g. education), and processing of information (e.g. Internet banking or insurance). For consumers it might sometimes be difficult to evaluate the service quality or capability of service providers and therefore eWOM can become a credible and cost-effective way to gain information from other experienced customers. This is especially the case for credence goods; where it is difficult to evaluate the service after consumption. Examples are several forms of medical treatment and education. Furthermore, Brown and Reingen (1987) stated that WOM may be particularly important for services. Remarkable, less research has been done directed at the behavior of the consumer when buying services. Moreover, Murray & Schlacter (1990) showed that WOM is more important and influential in a services context instead of a product context.
1.4. Problem statement & research questions
Since there are significant differences between services and products and given the fact that eWOM is becoming more and more important nowadays, it would be interesting to link these concepts and to examine if the effect of eWOM is different for several purchase types (services vs. products) and to see how this will influence the purchase intention of consumers. Therefore, the following problem statement can be formulated:
How do online reviews affect the purchase intention of a consumer and how will this differ between product reviews and service reviews?
Previous research showed that the level of trustworthiness is an important antecedent of WOM (de Matos & Rossi, 2008) and that higher perceived credibility will result in higher eWOM adoption. Since this process might be influenced by the recommendation framing – positively or negatively framed – (Cheung et al., 2009) it is of importance and interest to take this into account while investigating the role of trustworthiness of reviews. Given that many previous studies ignored the type of product – services versus products – it would be of great interest to take a further look into this field. Therefore, the broad problem statement can be divided into several research questions.
Do online service reviews have a different effect on consumers than online product reviews?
What is the distinction between positive and negatively framed messages and how do they influence the consumer?
The objective of this study is to explore how consumers will be influenced after reading online reviews and to examine which factors might influence this process. Besides, the comparison will be made between service reviews and product reviews, since these have significantly different features and characteristics which influence the purchase decision of consumers. It is hoped that the results of this study will contribute to both the existing literature and practice in such a way that marketers can cater to this findings and that managers reach a more complete insight into the field of eWOM and social influence.
1.5. Relevance
1.5.1. Theoretical relevance
Because previous research has mainly focused on product-oriented issues (Murray, 1991), considerable less empirical research has been devoted to the behavior of consumers when making service purchase intentions based only on eWOM. Although empirical studies showed that customers are even more likely to rely on these interpersonal communications in the service context – because of the intangibility and experiential nature of service (Murray, 1991) – there is still a lack of knowledge about the factors influencing service-oriented issues compared with product-oriented issues. Therefore it is expected that this research will contribute to the existing literature of eWOM by filling the gap between online services and products.
1.5.2. Managerial relevance
1.6. Thesis structure
2. THEORETICAL FRAMEWORK
This chapter consists of a literature review regarding online reviews for products and services, the trustworthiness of online reviews and the recommendation framing of online reviews. The first paragraph starts with a deep analysis about the influence which product and service reviews have on purchase intentions of consumers followed by an investigation about the role of trustworthiness of these reviews. This chapter ends with an examination of the effect of recommendation framing on trustworthiness and purchase intention of online reviews.
2.1. Online reviews
When consumers have the intention to purchase a product or service they seek for information in order to gain knowledge about the product quality. Because of the growing popularity of the Internet during the last years, an increasing number of consumers tend to search for this information online. Online consumer reviews are becoming more and more important in consumers‟ purchase decisions. It can be defined as “a type of service of product information created by users based on personal usage experience” (Chen & Xie, 2008). Online product and service providers can invite users of their products and services to post personal evaluations on the sellers‟ Website. In 2007, figures showed that one person out of four (24%) will firstly search for reviews of other consumers before paying for a service (Comscore & The Kelsey Group, 2007). They incorporated services in the field of gastronomy, travel, automotive, home, and medical. From these services, consumers will mostly search for reviews of restaurants and hotels (41% and 40%). Another interesting result from the same source is that more than three quarters of review users reported that the review had significant influence on their purchase. Moreover, 97% of those people said to find the review accurate. Another example is that the product sales in the book market were significantly influenced by online consumer ratings (Chevalier & Mayzlin, 2006). Based on the figures above and other several studies (Sawhney & Eliashberg, 1996; Basuroy et al, 2003) there can be concluded that online consumer product and service reviews have an effect on the purchase behavior of consumers.
goods; you first have to experience the service before you can evaluate it. New upcoming service categories are evolving in the field of jobs offered online at people‟s homes, such as babysitters or plumbers (e.g., www.werkspot.nl) where customers can grade and evaluate the delivered service. These services are part of the credence good, where an evaluation is difficult or impossible after experiencing the service. Despite this upcoming eWOM on the service market, there is an enormous lack in the service context within the existing literature. While researchers have investigated the procedures and benefits of online product reviews, the difference with online service reviews of consumers have not been fully explored.
Earlier research showed that WOM is more important and influential within a services context than in a product context only (Murray & Schlacter, 1990). Products are relatively high in search qualities whereas services are low in search qualities. This is due to the fact that it is difficult to find information about the service in advance and therefore the risk to buy the service might be higher than with products. This is due to the fact of the IHIP (intangibility, heterogeneity, inseparability, and perishability) characteristics of services (Lovelock & Gummesson, 2004). Thus, consumers might not fully understand a service before purchasing and therefore WOM will become important and serves as an experienced source. Moreover, in the case of credence goods, the service is difficult or even impossible to evaluate at all (e.g., visiting a dentist). This results in a higher perceived risk when “purchasing” an experience or credence good instead of a product – a search good. A way of reducing this perceived risk by consumers is seeking information before purchasing. But since this is difficult or impossible for many services, WOM is the only information source on which to rely as well as past experiences in the case of experience goods. Literature suggests that WOM plays a more important role for services and is more influential for services than for products. Still, it is unknown if the same effect exists in an online environment.
H1a: The effect of reviews on purchase intention is higher in a service context than in a product context
When looking further in a services context only a difference exists in the meaning of experience goods and credence goods, as defined before in this chapter. Besides WOM, another possibility to gather information about the service is to rely on past experiences and evaluations, in the case of an experience good. Since this is impossible for credence goods we expect that consumers‟ value will be very high for online reviews. This results in the following hypothesis:
H1b: The effect of reviews on purchase intention is higher for credence goods than for experience goods
2.2. Trustworthiness of reviews
This chapter focuses on trust when consumers buy online products and services after reading several reviews from other consumers. Specifically, we would like to know whether a specific level of trustworthiness in reviews is necessary for consumers before making purchase intentions and if this trustworthiness is more important for product reviews or for service reviews. Is a certain amount of trust in the review needed for consumers before purchasing? And is the perceived trustworthiness of product reviews higher than for service reviews? Another important issue is the trustworthiness of the reviewer; the person who places the review on the Internet. There are thus two views of trustworthiness; the review itself and the reviewer.
search and purchases and customers can be more loyal online than offline (Shankar et al., 2003). Another reason why we expect trust to be important for eWOM is because of the risk which is involved when buying online products or services. Moreover, services are higher in perceived risk than products are (Murray, 1991). The concept of risk implies that many people make purchase decisions with some degree of uncertainty about the offered products or services. This is an important item since the study of Murray & Schlacter (1990) provided evidence that services evoke heightened risk perceptions.
The level of trust is dependent on different factors which consumers might experience. Martín and Camarero (2008) argued that these influencing factors are satisfaction with previous purchases, the Website security and privacy policies, and service quality. Additionally, product characteristics (e.g., product popularity) and consumer characteristics (e.g., Internet experience) affect the consumers‟ reliance on online reviews as well (Zhu & Zhang, 2010). Although there are contrary results in the literature, Zhu & Zhang (2010) demonstrated that reviews only influence product sales when consumers‟ reliance in reviews is sufficiently high. We therefore expect that the level of trustworthiness in reviews is playing a significant role within the process of purchase intention. Research by Nielsen (2007) showed that trust is high by online reviews posted by other consumers (61%) rather than by TV (56%) or radio (54%). This result is supported by Wilson & Sherrell (1993) who said that consumer-created information is likely to be more credible than seller-created information. These studies positively showed the understanding of trustworthiness in the opinion of others.
2010). It will help when consumers looking for reviews have insight into these different types of assertions in examining credibility. Furthermore, consumers must look for reviewers who place more than one assertion of experience online (Mackiewicz, 2010).
It is known that the role of trustworthiness in online reviews is critical before making purchases. Although many people are trustworthy against online reviews, people also believe (48% of respondents) that fake reviews were being created on consumer sites (Tynan, 2006). Another result is that 57% said that they would not buy a product or service when they find the review suspicious (Tynan, 2006). This again indicates the importance of the phenomenon trustworthiness. The research in online trust is extensive, although most research is focused at the role of product reviews and thus there is again a lack at research in service reviews.
The importance of trustworthiness in online reviews is remarkable in theory and therefore the hypothesis below will be formulated:
H2a: Consumers’ trust in online reviews will positively affect their purchase intention
We come to the conclusion that, because of the perceived risk when buying services, a specific level of trust in reviews is needed before making a purchase. We expect that this process might be more important for service reviews than for product reviews since product reviews do not have to be the only information source when buying products. This will lead to the following hypotheses.
2.3. Recommendation framing
Cheung et al. (2009) argue that recommendation framing refers to the valence of the eWOM; “whether a recommendation is positively framed (e.g., a praise message) or negatively framed (e.g., a complaint message)”. Additionally, there is reason to believe that the recommendation framing will influence the level of trustworthiness of the message. Since negatively messages will not contribute to the purchase intention of consumers, Doh & Hwang (2009) showed that a few negative messages might be useful for creating positive attitude toward trust of eWOM messages. This can be due to the fact that negative messages can be harmful but it is only effective when there are already several positive messages. The reason for this is that consumers might distrust the message if they found no negative information. This result will be supported by Mudambi & Schuff (2010) who found that for experience goods, reviews with extreme high ratings are less helpful than reviews with moderate ratings. It is thus likely that when consumers read too much positive reviews or too enthusiastic reviews, they will distrust the product or service at a certain level. Therefore we formulate the following hypothesis:
H3a: the trustworthiness of the online reviews is higher when there is a negative review between the positive reviews
Despite the lack of research towards online services, Zheng et al. (2009) based their research on the hospitality industry and argue that service providers must pay more attention to negatively framed messages and must provide service recovery procedures. In addition, research has been done to the effects of negatively framed eWOM (Luo, 2007) and online complaining behavior (Zheng et al., 2009). Since no comparison is been made of possible differences between services and products, it is interesting to further examine the eWOM processes within this field, which is part of the subject of this paper. No attention has been given in previous literature to possible differences for services and products when talking about recommendation framing. Since the nature of services can lead the consumer in more risky situations and more complex information search we will translate this to a services context specifically – where WOM and past experiences are the only information source on which to rely. We therefore expect that the effect of negative reviews is higher for services than for products:
H3b: The effect of negatively framed reviews on purchase intention is stronger in a services context than in a product context
Consistent with the earlier formulated argument about the difference between experience goods and credence goods we expect that the impact of negatively framed reviews is higher for credence goods than for experience goods. This is due to the fact that the review for a credence good is the only source on which to rely and when this message is not satisfying there is no option on which to fall back. The hypothesis will be as follow:
2.4. Conceptual model
Figure 1 displays the conceptual model in which the above formulated hypotheses are combined. This conceptual model is a research model which illustrates how the different hypotheses are positioned towards each other.
FIGURE 1: Conceptual model
2.5. Social influence
As said before, eWOM is a form of social influence. The information around a person is one of the most influencing factors which affect his behavior. This social influence exists of normative and informational influence. Normative influence is social pressure designed to encourage conformity to the expectations of others (Hoyer & MacInnis, 2007 p.408). Normative influences can occur in order to achieve rewards or to avoid punishment. Informative influences refer to the provision of credible evidence of reality (Burnkrant & Cousineau, 1975). These influences occur from people whose opinions are perceived as credible by other people. Social influence can be normative, informational or both; they are often hard to distinguish (Cialdini & Goldstein, 2004) because messages convey information both about the way things are (informational influence) and the way things ought to be (normative influence). Early research in this field found that in consumer decision making “individuals tended to conform to the group norm”. However when the intention of the group was to “go along” with the group, which resulted in restricted feelings, “the individuals tended to resist the group pressure” (Venkatesan, 1966), and that group members tended to choose the same brand as their group leader (Stafford, 1966). These results indicate that people close to each other will easily influence each other.
3. METHODOLOGY
For testing the previous mentioned hypotheses and conceptual model a research method will be developed and explained. The research objective was to examine differences between service reviews and product reviews for purchase intention and the role of trustworthiness and recommendation framing for this process. This chapter will first argue which services domains will be included and the choice of the research and then the method of collecting data will be discussed.
3.1. Service domain
This study will specifically focus on the online service industry, and therefore the respondents, which will be students, be confronted with questions and information concerning a specific product or service. One group of respondents will be asked questions about a search good, another group of respondents about an experience good and the last group about a credence good. All groups of respondents will be confronted with a scenario about the product or service. After that, they will read four consumer reviews in which the product or service is evaluated; positive or negative.
The experience good about which the respondents will be asked is a restaurant. The respondents must give their opinion about a specific restaurant – described in a scenario – on the basis of different consumer reviews. As described earlier, the study of Comscore & The Kelsey Group (2007) showed that most people who search for a review will do this before visiting a restaurant (41%). This indicates the importance of reviews for this category. Besides, this category is particular suitable for students, since they will regularly visit restaurants.
regularly. It is very important to offer a service with which the students are well-known because currently online reviews of dentists are not available yet.
In addition, besides the chosen credence good, respondents will be confronted with questions about a search good as well. The product type which will be used will be a laptop. Walter (2006) showed that music and technology are at the top four of most important topics surrounding WOM. In this way, possible differences between services and products can be discovered.
In order to select these three different purchase types it will not only be possible to conclude whether there is a difference in purchase intention between product reviews and service reviews but also to explore if there is a difference concerning trust between the three purchase types. Furthermore, we can explore if the recommendation framing will have different effects for the purchase intention or the level of trustworthiness of the reviews and reviewers within different purchase types.
Additionally, in order to explore if the three chosen purchase types are of similar importance for the respondents, a manipulation check will be executed simultaneously with the experiment. The level of involvement will be measured by means of three questions.
3.2. Research design
TABLE 1: Experimental design Purchase type
Recommendation framing
Positively framed search good reviews
Positively framed experience good reviews
Positively framed credence good reviews
Positive and negatively framed search good reviews
Positive and negatively framed experience good reviews
Positive and negatively framed credence good reviews
In total, we have six experimental conditions. Every condition will receive a different questionnaire dependent on the recommendation framing and the purchase type. There is chosen to perform an experimental design in order to infer causal relationships (Malhotra, 2007). The experimental design is a between subjects design. Before executing this experiment, a 24-factor questionnaire is used to identify influentials or opinion leaders.
3.3. Procedure
The main content of the experiment holds that respondents have read four reviews in which the framing and purchase type depends on the group in which the respondents are divided. While reading such reviews the respondents are thus situated in informative ingoing influence. The four presented reviews are accomplished by means of combining existing reviews and own completion. The reviewers are thus fictitious and are named like “Person 1”, “Person 2”, etc. After reading those reviews attentively, they will firstly be asked about the level trustworthiness and honesty of the four reviewers on a 7-point Likert scale varying from “very unreliable” to “very reliable”. This scale is based on the existing scale of Cheung et al. (2009). Related to this question, statements are showed about the trustworthiness of the reviews itself. This will be tested based on two statements which are also based on the scale of Cheung et al. (2009). Since many papers are using rather general questions and statements, these existing questions are adapted into this specific setting related to the reviews. These two statements will be measured on a 7-point Likert scale as well, varying from “totally disagree” to “totally agree”. The last two issues measure the general opinion of the product or service and their purchase intention. Both issues are measured the same way as explained above.
scales both normative and informative influences can be measured; an amount of 24 issues in total. Including this part in the experiment a more detailed perspective on social influence of the consumers is expected to discover. We think it is useful to identify influencers by means of these scales and if they are sensitive for ingoing influences as well. Before the main content of the experiment starts the source is been measured which consumers use when buying, using, or attaining the specific product or service which is a self-reported item that needs to demonstrate the use of online reviews. Additionally, the level of involvement with the product and services will be measured based on the existing scale of Knox & Wolker (2003) by means of three questions.
3.4. Data collection
4. RESULTS 4.1. Sample
In total, 197 respondents (N=197) participated in fulfilling the questionnaire. Probably due to the length of the questionnaire, 42 respondents did not complete the whole questionnaire. The sample consists of 32,5% males and 67,5% females. The average age of the respondents is 22,3 years and shows a spread of 16 to 27 years, from which most respondents are 23 years old. Furthermore the respondents possesses or have possessed 2,9 laptops from which two persons do not have a laptop. The respondents visit on average 12,8 times a restaurant in a year and they visit a dentist twice a year. There is one person who never visits a dentist. Additionally, it is interesting to see which information sources respondents use before purchasing the product or service. Table 2 provides an overview of the means for every information source.
TABLE 2: Means of information sources
Friends Family Website Comparison
site Online reviews Own knowledge Laptop 4.35 4.59 4.39 4.67 4.43 5.61 Restaurant 5.35 5 4.85 2.67 3.08 6.02 Dentist 4.9 5.61 4.36 3.07 3.10 5.30
From this table we can conclude that for laptops own knowledge and comparison sites are the most important information sources, for restaurants own knowledge and friends, and for a dentist family and own knowledge. Furthermore it is interesting to see whether there are significant differences between the three purchase types for every source. This can be done by means of an one-way ANOVA test which shows that there exists differences between the three purchase types for all sources (p < .05). Only using a Website is not different per purchase type (p > .05).
4.2. Manipulation check
name these variables “involvement”. Within table 3 the means of involvement per purchase type are showed.
TABLE 3: Means of involvement per purchase type
Purchase type Mean
Laptop 5.29
Restaurant 4.69
Dentist 4.41
From the table it becomes clear that respondents have the highest involvement with laptops, followed by restaurants and dentists. By means of an one-way ANOVA test we see that these differences are significant (p < .05), which means that the involvement by the respondents is not equal for the three purchase types. When looking specifically at Post Hoc Tests we see that only the involvement of the respondents for a restaurant and for a dentist is comparable (p > .05).
4.3. Scale reliability
4.3.1. Scale reliability for social influence
In order to measure the scale reliability for social influence the Bartlett‟s test of sphericity is used to see whether the factor analysis is appropriate. Social influence is been measured by means of ingoing and outgoing social influence The procedure for defining the number of factors includes a determination based on eigenvalues, the total variance explained and a scree plot. Table 4 shows the outcomes of the value of the KMO statistic and the significance based on Bartlett‟s test of spericity as well as the initial eigenvalues.
TABLE 4: KMO & Bartlett’s test of spericity
KMO Significance
Ingoing normative .871 .000
Ingoing informative .687 .000
Outgoing normative .915 .000
Outgoing informative .764 .000
only one factor can be extracted from the data since only the first factor has an eigenvalue of > 1.000. The total variance explained by one factor are for outgoing normative (70.913%) and informative (66.721%) above the minimum level of 60% whereas the total variance explained by one factor for ingoing normative (54.535%) and informative (57.982%) are very close by 60% and could therefore suggest two factors . Four scree plots can be found in the appendix as well and shows that the scree – referred to as the gradual trading off – begins after one factor and thus implies that one factor is most appropriate. At last, we could determine the number of factors based on the variance of each factor. Within appendix A there can be seen that for ingoing normative five variables have more than 5 percent of variance, for ingoing informative all four variables are > .5, for outgoing normative two variables are > .5 and for outgoing informative three variables are > .05. However the other approaches described above confirm that in this case one factor is most appropriate and therefore this analysis will be continued with one factor. To provide a complete overview, the factor loadings are given where can be seen which variables is highly loaded on the factor.
Additionally, the Cronbach alphas have been calculated for the different types of social influence. The in- and outgoing scales consist of normative and informative oriented questions. Within appendix A.3. there can be seen that the first questions of ingoing normative, ingoing informative, and outgoing informative score relatively low on the factor. Interesting is that this effect only becomes visible for the variables ingoing informative and outgoing informative social influence after calculating the Cronbach alphas if item deleted. Table 5 shows an overview of the alphas calculated for these separate scales.
TABLE 5: Cronbach alpha’s for all types of social influence
Variable Cronbach alpha
Ingoing normative .878
Ingoing informative .731
Outgoing normative .938
Outgoing informative .830
Furthermore, we conducted factor analysis in order to see the amount of factors when including all four types of social influence questions. When looking at the KMO statistic (.902) we see that the value is very large (> .5) and the Bartlett‟s test of sphericity is significant (p < .05). Factor analysis is thus appropriate in order to test the scale reliability. When applying the eigenvalue criterion we can conclude that four factors can be extracted since these four factors have an eigenvalue of > .1000. The total variance explained by three factors (64.353%) is above the minimum level of 60%. From the scree plot it is hard to see but the scree begins little after two factors. Furthermore, the variance of each factor only the first three variables are > .5. Based on the eigenvalues, it is most appropriate to continue with four factors. We have used the rotated factor matrix (Varmimax procedure) in order to simplify the interpretation (appendix B.3.). The matrix does not show clear separate factors for the four types of social influence. Ingoing and outgoing informative social influence have both high loadings on the second factor, for example, while both types of normative social influence are more clear. Looking at the fifth factor, which is showed within the rotated component matrix, we can see the deviation compared with the other factors and therefore there will be chosen not to continue with five factors. The fourth factor shows a very small deviation as well, but since several variables of ingoing normative have high loadings on factor four there will be chosen to continue this research with four factors.
4.3.2. Correlation social influence
When conducting a correlation test we see in table 6 that the subscales are significantly correlated.
TABLE 6: Correlation social influence
Innorm Ininform Outnorm Outinform
Innorm Pearson Correlation 1 .405 .731 .426
Sign. .000 .000 .000
Ininform Pearson Correlation .405 1 .250 .661
Sign. .000 .001 .000
Outnorm Pearson Correlation .731 .250 1 .549
Sign. .000 .001 .000
Outinform Pearson Correlation .426 .661 .549 1
Sign. .000 .000 .000
A scatterplot is been made in order to provide a complete picture of the heterogeneity of the sample. Within the four scatterplots below (figure 2) there can be seen how the respondents score on the different social influences.
FIGURE 2: Scatterplot of normative & informative social influence
The ingoing and outgoing normative influence are strongly correlated (r = .731, p < .05) which is showed in the left figure. The figure also shows that when outgoing normative influence is high the heterogeneity is high concerning the ingoing normative influence they report. The figure on the right shows a strong positive correlation as well (r = .661, p < .05) between ingoing and outgoing informative influence.
FIGURE 3: Scatterplot ingoing & outgoing social influence
informative influence increases as well, but slightly (r = .405, p < .05). The figure on the right shows that outgoing normative and outgoing informative influences are positively correlated (r = .549, p < .05). This figure also shows some heterogeneity in the left region. Some respondents are very susceptible for outgoing normative influence but score low for outgoing normative influences.
4.3.3. Scale reliability eWOM variables
Before testing the scale reliability for the different experiments, factor analysis will be used. When looking at the KMO statistic (.737) we see that the value is large (> .5) and the Bartlett‟s test of sphericity is significant (p < .05). There can be therefore concluded that factor analysis is here an appropriate analysis as well to test the scale reliability. Again, we applied the eigenvalue criterion from which one factor can be extracted since only the first factor has an eigenvalue of > .1000. The total variance explained by one factor is 81.300% which is far above the level of 60%. These results can be found in appendix C as well as the scree plot and the factor loadings. The scree plot shows that one factor is most appropriate as well when looking at the scree which begins after one factor. Within appendix C the factor loadings are given as well. Furthermore, when looking at the variance of each factor all three variables are > .5 but we will use one factor since the above mentioned approaches have showed that one factor is most appropriate.
TABLE 7: Cronbach alphas
Variable Cronbach alpha
Trust in review .842
General opinion .835
Purchase intention .808
From the table it can be concluded that the scales are reliable.
4.4. Hypotheses
There are two independent variables (purchase type versus recommendation framing) which will be divided in different groups (search good/experience good/credence good and all positively framed reviews/positive and one negatively framed review. Therefore for several hypotheses an ANOVA 2-way test is been conducted. The confidence level used for this test is 95%. For other hypotheses a regression analysis is been conducted since we want to explain different variables.
H1a: The effect of reviews on purchase intention is higher in a service context than in a product context
H1b: The effect of reviews on purchase intention is higher for credence goods than for experience goods
For the purchase intention, the ANOVA showed a significant main effect for the purchase type (F = 58.031, p < .05). Specifically, respondents have higher purchase intention after reading a service review (M = 3.801) than after reading a product review (M = 2.100). This result provides support for H1a. Furthermore, when looking specifically at the type of service, the ANOVA showed a significant main effect for purchase type as well (F = 38.015, p < .05). Respondents have higher purchase intention after reading a review of an experience good – a restaurant – (M = 4.260) than after reading a review of a credence good – a dentist – (M = 3.400) and therefore does not support H1b.
H2a: Consumers’ trust in online reviews will positively affect their purchase intention H2b: The effectiveness of trustworthy reviews on purchase intention will be higher in a
service context than in a product context
Testing the above mentioned hypotheses we want to explore whether there is a relationship between the dependent variable, which is purchase intention, and several independent variables, which are trust, trust × purchase intention, and purchase type. The latter two variables apply to the second hypothesis. Based on this, we conducted a multiple regression analysis in which the dependent variable assumingly can be explained by the level of trustworthiness in online reviews and the purchase type.
The overall model is significant (p < .05) and thus the independent variables explain at least a part of the dependent variable. When looking at the adjusted R2, we see that the purchase
intention can be explained for 54,4 % by the level of trustworthiness in online reviews and the purchase type (R2 = .544, F = 58.232, p < .05). Not all separate dependent variables attend to this result; only the level of trustworthiness in online reviews adds .431 to this equation. This result is in line with H2a. Table 8 shows an overview of the betas and the p-values.
TABLE 8: Betas and p-values of multiple regression analysis
Beta P-value
Trust × purchase type .387 .095
Purchase type .014 .945
Trust in review .431 .000
Furthermore, the table shows that trust × purchase type is weak significant (p < .10) and that it adds .387 to this equation. This small interaction effect is in line with H2band although not significant, the test shows a non-significant trend in the direction of H2b. Additionally, purchase intention does not differ for the type of purchase only (beta = .014).
TABLE 9: Betas and p-values of multiple regression analysis including “involvement”
Beta P-value
Trust × purchase type .392 .091
Purchase type .015 .940
Trust in review .431 .000
Involvement .044 .457
The small interaction effect is higher than before (.389), but still weak significant (p < .10) and therefore the result of H2b does not change.
H3a: The trustworthiness of the online reviews is higher when there is a negative review between the positive reviews
H3b: The effect of negatively framed reviews on purchase intention is stronger in a services context than in a product context
H3c: The effect of negatively framed reviews on purchase intention is stronger for credence goods than for experience goods
For all three hypotheses which are mentioned above an ANOVA test is been performed. First of all, the ANOVA did not show a main effect for recommendation framing on the level of trustworthiness of the online reviews (F = 1.881, p > .05). The respondents did trust positive reviews more (M = 4.6) than they trust positive reviews combined with a negative review (M = 4.313). Although early evidence is been found in previous research for this hypothesis, we have to conclude that the result is not in line with H3a and therefore the hypothesis will be rejected.
The ANOVA showed that the main effect for recommendation framing on purchase intention was not significant (F = .619, p > .05). Therefore we could not provide support for H3b as well.
Despite the fact that the overall main effect is not significant, the effect of negatively framed reviews on purchase intention (F = .641, p > .05) is stronger for a credence good (M = 3.328) than for an experience good (M = 4.058). This interaction effect shows that H3c can be therefore supported.
4.5. Additional analysis
In addition to testing the hypotheses some interesting additional analysis have been performed as well. Despite the fact that there was no elaboration within the theoretical analysis about the different types of social influence we have tested the sensitivity of the respondents for different types of social influence and if this results corresponds with the purchase intention after reading the four reviews. This test has been executed by means of a correlation. The most important and interesting finding shows that respondents sensitive for ingoing informative social influence scores high on purchase intention after reading the reviews as well (r = .169, p < .05). Furthermore, participants who are sensitive for ingoing normative influences also scores high on the purchase intention (r = .198, p < .05). There was no correlation between outgoing social influence and the purchase intention after reading the reviews.
Another interesting item is the perceived trustworthiness and honesty of the persons who have placed the reviews – the reviewers. Since the respondents must evaluate all four types of reviews we have to measure within subjects and therefore we have performed an ANOVA repeated measures in order to discover differences in the opinions of participants. Table 10 shows the means for the perceived trustworthiness and honesty created by the reviewers.
TABLE 10: Means perceived trustworthiness and honesty of reviewers
Trustworthiness Honesty
Reviewer 1 4.28 4.65
Reviewer 2 4.23 4.68
Reviewer 3 4.56 4.99
Reviewer 4 4.71 4.91
But since reviewer 3 was half of the time negative in his review, the means can be different. In table 11, the means of reviewer 3 will be given for both his positive and his negative review.
TABLE 11: Means reviewer 3
Trustworthiness Honesty
Positive 4.53 4.68
Negative 4.61 5.28
5. DISCUSSION
In this section, conclusions are made based on the data which will be related to the theoretical findings. An answer will be given on the problem statement. Subsequently, this chapter ends with a discussion about recommendations of online reviews for managers and researchers which are based on the results of this research.
5.1. Conclusion
This analysis and much previous research showed that online reviews affect the behavior of consumers. Since this research focuses specifically on the role of purchase type – services and products – an important conclusion is that consumers‟ intention to buy online is higher after reading a service review instead of a product review. This is possibly due to the characteristics of services, which implies risky and uncertain situations. Therefore, online reviews are especially important for the services industry. Within this research three different purchase types are chosen in order to test whether and where differences exist. Expected was that the level of involvement is about equal among the participants; students. In the beginning of this research we performed a manipulation check in order to test whether the three purchase types included the same level of involvement. Since this manipulation check showed that this level of involvement is not equal for all respondents, it could contribute to the rejection of several hypothesis of this research. But first a conclusion will be given for each hypothesis.
As expected, results showed that when consumers trust online reviews their purchase intention will increase. When testing if consumers do trust service reviews more than product reviews we can conclude that this hypothesis is partly accepted. After including the variable “involvement” within the correlation test we noticed a higher non-significant trend in the direction of this hypothesis but still insufficient to accept the hypothesis completely.
With this hypothesis the second research question can therefore partly be answered. We thus can conclude that trustworthiness of reviews is very important for consumers before making a purchase. Moreover, when consumers do trust the review, the purchase intention will be slightly higher in a services context.
Furthermore, the effect of negatively framed reviews on purchase intention is stronger in a product context than in a services context, while the opposite was expected. The same cause as described earlier can be attributed here; when respondents evaluated more reviews, the result could have been different. These last hypotheses will answer our third research question; this study shows to a certain extent differences in beliefs of consumers after reading positive or negative reviews. Thus we can conclude that online reviews do not only influence the purchase intention of consumers but influence their level of trustworthiness as well. These conclusions give an answer on the problem statement.
At last several interesting additional analysis have been performed and there can be concluded that participants sensitive for social influences react sensitively for online reviews as well. Additionally, we noticed some heterogeneity between the participants which implies that participants differ in their susceptibility for both normative and informative influences, and ingoing and outgoing social influences. Thus it is not the case that when participants are highly susceptible for informative influences, they are highly susceptible for normative influences as well. Another result from the additional analysis is that participants rated reviewer 3 – thus the reviewer as a person – as trustworthy and honest. Since reviewer 3 is half of the time negative about the purchase type, we ascribe this result to the negative message. Here again, further research is needed since participants did not evaluated the negatively framed reviews in general as more trustworthy than positively framed reviews which thus does not influence the purchase intention, which is very remarkable. This can be due to the fact that participants evaluate reviewers individually and become more aware of the true message than they have to evaluate all messages simultaneously.
5.2. Implications for managers & researchers
Since online reviews are a powerful tool, both positive and negative, managers must decide whether they want to provide consumers the opportunity to share opinions and experiences in the first place. When managers want to offer this service they have to be aware of the content of the reviews. Managers could also choose to manage this process themselves; removing negative or unreliable reviews. But they have to realize the other side as well. Many consumers think that certain reviews are posted by marketers, which can have negative results too. When looking specifically at service providers, they can gain more insight in their potential consumers thanks to these researches. Because there has been shown that reviews are very important for services, service providers can easier choose to provide consumers the opportunity to share experiences. Furthermore, managers have gained more insight into how consumers influence each other and how they are influenced by other consumers.
6. LIMITATIONS & FURTHER RESEARCH
This research brought several insights into the field of online reviews for services and products. Nonetheless, there are limitations which might be interesting for further research.
6.1. Limitations
This study has been executed with care and precise information; nevertheless several points of criticism can be ascribed to this writing. First of all, my study focuses only on a very specific form of eWOM. In order to create a complete view of how eWOM works and what influences eWOM, other forms must be taken into consideration as well. Examples are forums, blogs, social networking sites and chat rooms. Secondly, since this study has been limited to only two influencing factors – level of trustworthiness and recommendation framing – other influencing factors might be of importance as well, such as the Websites‟ reputation, user experience with online reviews, and earlier experience with the product or service. It was not possible to take into account all influencing variables because of the limited time. Another limitation is that the perspective of the reviewer is not taken into account. For example: the reviewer‟s thoughts about placing the review, his knowledge, and his opinion. Another point of attention is the amount of reviews in the questionnaire. In this study the reviews are limited to the amount of four and probably, the respondents needed to read more reviews in order to create different effects. Since the length of the questionnaire was already rather long, more reviews could have made the respondent to decide to quit with it.
6.2. Further research
7. REFERENCES
Basuroy, S., S. Chatterjee & R.S. Abraham (2003) How Critical are Critical Reviews? The Box Office Effects of Film Critics, Star Power, and Budgets. Journal of Marketing, 67 (4), 103-117.
Bearden, W.O., R.G. Netemeyer and J.E. Teel (1989) Measurement of Consumer
Susceptibility to Interpersonal Influence. Journal of Consumer Research, 15 (4), 473- 481.
Beason, L. (1991) Strategies For Establishing an Effective Persona: an Analysis of Appeals to Ethos in Business Speeches. Journal of Business Communication, 28 (4), 326-346. Bone, P.F. (1995) Word-of-Mouth Effects on Short-term and Long-term Product Judgments.
Journal of Business Research, 32 (3), 213-223.
Brown, J.J. and P.H. Reingen (1987) Social Ties and Word-of-mouth Referral Behavior. Journal of Consumer Research, 14 (3), 350-362.
Burnkrant, R.E. & A. Cousineau (1975) Informational and Normative Social Influence on Buyer Behavior. Journal of Consumer Research, 2 (3), 206-215.
Chen, Y. & J. Xie (2008) Online Consumer Review: Word-of-Mouth as a New Element of Marketing Communication. Management Science, 54 (3), 477-491.
Cheung, M.Y., C. Luo, C.L. Sia & H. Chen (2009) Credibility of Electronic Word-of-Mouth: Informative and Normative Determinants of On-line Consumer Recommendations. International Journal of Electronic Commerce, 13 (4), 9-38.
Chevalier, J.A. & D. Mayzlin (2006) The Effect of Word of Mouth on Sales: Online Book Reviews. Journal of Marketing Research, 43 (3), 345-354.
Cialdini, R.B. & N.J. Goldstein (2004) Social Influence: Compliance and Conformity. Annual Review of Psychology, 55 (1), 591-621.
Comscore & The Kelsey Group (2007) Online Consumer-Generated Reviews Have
Significant Impact on Offline Purchase Behavior, ComScore Inc, (November 29), available at:
www.comscore.com/Press_Events/Press_Releases/2007/11/Online_Consumer_Revie ws_Impact_Offline_Purchasing_Behavior
Doh, S. & J. Hwang (2009) How Consumers Evaluate eWOM Messages. Cyberpsychology and Behavior, 12 (2), 193-197.
Communication. Marketing Science, 23 (4), 545-560.
Hoyer, W.D. & D.J. MacInnis (2007) Consumer Behavior. Fourth Edition. Houghton Mifflin Company, Boston, New York.
Jager, W. (2010) Two-dimensional Simulated Social Interaction: some data, ideas and
problems. Paper submitted to the World Conference on Social Simulation, September 2010, Kassel, Germany.
Kanouse, D.E. (1984) Explaining Negativity Biases in Evaluation and Choice Behavior: Theory and Research. Advances in Consumer Research, 11 (1), 703-708.
Knox, S. & D. Walker (2003) Empirical Developments in the Measurement of Involvement, Brand Loyalty and their Relationship in Grocery Markets. Journal of Strategic Marketing, 11 (4), 271-286.
Lovelock, C. & E. Gummesson (2004) Whither Services Marketing? In Search of a New Paradigm and Fresh Perspectives. Journal of Service Research, 7 (1), 20-41. Luhn, R. (2009) Can You Trust Online User Reviews? PC World, 27 (1), 30-32.
Luo, X. (2007) Consumer Negative Voice and Firm-Idiosyncratic Stock Returns. Journal of Marketing, 71 (3), 75-88.
Mackiewicz, J. (2010) Assertions of Expertise in Online Product Reviews. Journal of Business & Technical Communication, 24 (1), 3-27.
Malhotra, N.K. (2007) Marketing Research. An Applied Orientation. Fifth edition. Pearson Prentice Hall, New Jersey, USA.
de Matos, C.A. & C.A.V. Rossi (2008) Word-of-mouth Communications in Marketing: a Meta-analytic Review of the Antecedents. Journal of the Academy of Marketing Science, 36 (4), 578-596
Maheswaran, D. & J. Meyers-Levy (1990) The Influence of Message Framing and Issue Involvement. Journal of Marketing Research, 27 (3), 361-367.
Martín, S.S. & C. Camarero (2008) Consumers Trust to a Web Site: Moderating Effect of Attitudes toward Online Shopping. CyberPsychology & Behavior, 11 (5), 549-554. Moorman, C., R. Deshpandé & G. Zaltman (1993) Factors Affecting Trust in Market
Research Relationships. Journal of Marketing, 57 (1), 81-101.
Morgan, R.M. & S.D. Hunt (1994) The Commitment-Trust Theory of Relationship Marketing. Journal of Marketing, 58 (3), 20.
Mudambi, S.M. & D. Schuff (2010) What Makes a Helpful Online Review? A Study of Customer Reviews on Amazon.com. MIS Quarterly, 34 (1), 185-200.
Journal of Marketing, 41 (9/10), 1173-1202.
Murray, K.B. (1991) A Test of Services Marketing Theory: Consumer Information Acquisition Activities. Journal of Marketing, 55 (1), 10-25.
Murray K.B. & J.L. Schlacter (1990) The Impact of Services versus Goods on Consumers‟ Assessment of Perceived Risk and Variability. Journal of the Academy of Marketing Science, 18 (1), 51.
Nielsen Consumer Report (2007) Word-of-Mouth: the Most Powerful Selling Tool: Nielsen Global Survey.
Patti, C.H. & C.H. Chen (2009) Types of WOM Messages: Information Search and Credence- Based Services. Journal of Promotion Management, 15 (3), 357-381.
Sawhney, M.S. & J. Eliashberg (1996) A Parsimonious Model for Forecasting Gross Box- Office Revenues of Motion Pictures. Marketing Science, 15 (2), 113.
Shankar, V., A.K. Smith & A. Rangaswamy (2003) Customer Satisfaction and Loyalty in Online and Offline Environments. International Journal of Research in Marketing, 20 (2), 153.
Stafford, J.E. (1966) Effects of Group Influences on Consumer Brand Preferences. Journal of Marketing Research, 3 (1), 68-75.
Sundaram, D.S., K. Mitra & C. Webster (1998) Word-of-Mouth Communications: A Motivational Analysis. Advances in Consumer Research, 25 (1), 527-531. Taylor, J. (2003) Word of Mouth Is Where It‟s At. Brandweek (June 2), 26.
Thorsten Hennig-Thurau, F., K.P. Gwinner, G. Walsh & D.D. Gremler (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. Tynan, D. (2006) This Blog for Hire. PCWorld (December 22).
Trusov, M., R.E. Bucklin & K. Pauwels (2009) Effects of Word-of-Mouth Versus Traditional Marketing: Findings from an Internet Social Networking Site. Journal of Marketing, 73 (5), 90-102.
Venkatesan, M. (1966) Experimental Study of Consumer Behavior Conformity and Independence. Journal of Marketing Research, 3 (4), 384-387.
Wilson, E.J. & D.L. Sherrell (1993) Source Effects in Communication and Persuasion Research: A Meta-Analysis of Effect Size. Journal of the Academy of Marketing Science, 21 (2), 101.
Wirtz J. & P. Chew (2002) The Effects of Incentives, Deal Proneness, Satisfaction and Tie Strength on Word-of-Mouth Behaviour. International Journal of Service Industry Management, 13 (2), 141-162
Zheng, T., H. Youn & C.S. Kincaid (2009) An Analysis of Customers‟ E-Complaints for Luxury Resort Properties. Journal of Hospitality Marketing & Management, 18 (7), 718-729.
APPENDIX A: FACTOR ANALYSIS SOCIAL INFLUENCE
As mentioned in chapter 4.3.1. This appendix includes a factor analysis consisting of a determination based on the eigenvalues, the total variance explained, the variance of each factor, and on the scree plots. Furthermore, an overview of the factor loadings is been given.
Appendix A.1. Initial eigenvalues
Ingoing normative social influence
Component Initial eigenvalues
Total % of variance Cumulative %
1 4.363 54.535 54.535 2 .980 12.246 66.781 3 .660 8.255 75.036 4 .570 7.124 82.160 5 .443 5.541 87.701 6 .397 4.962 92.663 7 .329 4.116 96.779 8 .258 3.221 100.000
Ingoing informative social influence
Component Initial eigenvalues
Total % of variance Cumulative %
1 2.319 57.982 57.982
2 .966 24.154 82.136
3 .457 11.437 93.573
4 .257 6.427 100.000
Outgoing normative social influence
Component Initial eigenvalues
Total % of variance Cumulative %
Outgoing informative social influence
Component Initial eigenvalues
Total % of variance Cumulative %
1 2.669 66.721 66.721
2 .841 21.018 87.739
3 .291 7.264 95.003
4 .200 4.997 100.000
Appendix A.2. Scree plots social influence
Ingoing normative Ingoing informative
Appendix A.3. Factor loadings
Ingoing normative social influence
Component
1
I rarely purchase the newest products and brands until I am sure my friends approve of
them. .598
It is important that others like the products and brands I buy .751
When buying products, I generally purchase those brands that I think others will
approve of .812
If other people can see me using a product, I often purchase the brand they expect
me to buy .711
I like to know what brands and products make good impressions on others .765
I achieve a sense of belonging by purchasing the same products and brands that
others purchase .796
If I want to be like someone, I often try to buy the same brands that they buy .733 I often identify with other people by purchasing the same products and brands they
purchase .722
Ingoing informative social influence
Component
1
To make sure I buy the right product or brand, I often observe what others are buying
and using .310
If I have little experience with a product, I often ask my friends about the product .857 I often consult other people to help choose the best alternative available from a product
class .901
I frequently gather information from friends or family about a product before I buy .823
Outgoing normative social influence
Component
1
My friends are influenced by my acceptance of their purchasing the newest products and
brands .769
My friends think it is important that I like the products and brands they buy .739 When buying products, my friends generally purchase those brands they think I like .877
If my friends see me using a product, they often purchase the same brand .847
My friends like to know what brands and products make good impressions on me .820 My friends achieve a sense of belonging by purchasing the same products and brands as I
purchase .905
My friends often try to buy the same brands that I buy .906
My friends often identify with me by purchasing the same products and brands as I
Outgoing informative social influence
Component
1
To make sure to buy the right product or brand, my friends often observe what I am
buying and using .492
If my friends have little experience with a product, they often ask me about the product .897 My friends often consult me to help choose the best alternative available from a product
class .913
APPENDIX B: FACTOR ANALYSIS SOCIAL INFLUENCE TOGETHER
This appendix includes a factor analysis of all four types of social influence questions together.
Appendix B.1. Initial eigenvalues Component Initial eigenvalues
Total % of variance Cumulative %
Appendix B.2. Scree plot
Appendix B.3. Rotated component matrix
