Public Usage Behavior in Online Social Networks qwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbn

(1)

qwertyuiopasdfghjklzxcvbnmqwerty

uiopasdfghjklzxcvbnmqwertyuiopasd

fghjklzxcvbnmqwertyuiopasdfghjklzx

cvbnmqwertyuiopasdfghjklzxcvbnmq

wertyuiopasdfghjklzxcvbnmqwertyui

opasdfghjklzxcvbnmqwertyuiopasdfg

hjklzxcvbnmqwertyuiopasdfghjklzxc

vbnmqwertyuiopasdfghjklzxcvbnmq

wertyuiopasdfghjklzxcvbnmqwertyui

opasdfghjklzxcvbnmqwertyuiopasdfg

hjklzxcvbnmqwertyuiopasdfghjklzxc

vbnmqwertyuiopasdfghjklzxcvbnmq

wertyuiopasdfghjklzxcvbnmqwertyui

opasdfghjklzxcvbnmqwertyuiopasdfg

hjklzxcvbnmrtyuiopasdfghjklzxcvbn

mqwertyuiopasdfghjklzxcvbnmqwert

yuiopasdfghjklzxcvbnmqwertyuiopas

dfghjklzxcvbnmqwertyuiopasdfghjklz

Public Usage Behavior

in Online Social Networks

Product Usage Behavior Under the Aspect of

Need for Uniqueness and Susceptibility to

Interpersonal Influence

23 June 2014

Anke Hauptmann

(2)

Public Usage Behavior in Online Social Networks

Product Usage Behavior Under the Aspect of

Need for Uniqueness and Susceptibility

to Interpersonal Influence

University of Groningen

Faculty of Economics and Business MSc Marketing Intelligence

Master Thesis 23 June 2014

1st supervisor: Dr. H. Risselada 2nd supervisor: Dr. J. E. M. van Nierop

(3)

Management Summary

In online social networks, an increasing trend in sharing product usage by default is observable. This trend is not only attached to applications such as games on Facebook but also to products which automatically connects with the profile in an online social network of the user and post the person's achievements. However, this trend might not be as valuable as intended. The intention behind this trend is the increase of word-of-mouth and therefore the increasing distribution of these products. But several personal traits can be contradictive towards this effect. Two of such characterizes are the Need for Uniqueness and Susceptibility to Interpersonal Influence. Need for Uniqueness is described as the drive to differentiate from other individuals with the behavior and products. When the product usage behavior becomes public through sharing, a person with a high Need for Uniqueness might be afraid of imitators. In contrary, Susceptibility to Interpersonal Influence is defined as how influenceable an individual is to the reference group or other individuals. A person with a high Susceptibility to Interpersonal Influence has the drive to correspond to the norms and expectations of the reference group. But this usually holds only when the behavior is observable. Hence, when a person wants to use a product which is not corresponding to the norms of the reference group but is shared by default, the person might risks the status within the social group. Overall, the present paper answers the research question how usage behavior is affected when it becomes public in online social networks under the aspect of Need for Uniqueness and Susceptibility to Interpersonal Influence.

(4)

Preface

During my internship at the research project UCIT, I initially get to know what research in the field of marketing contained and was familiarized with the topic of opinion leadership to which I dedicated my Bachelor thesis. After getting the opportunity to present on two conferences and publish two papers in the proceedings, I found the perfect studies for my field of interest: Marketing Intelligence at the University of Groningen. After only one year of studies, I know so much more than I thought I would learn. The year passed by way too fast and I am on one side sad to leave this place and on one side I am proud to have achieved my current situation.

The topic of my Master thesis derived from the research topic of opinion leadership. When I analyzed literature for my research team, I found the contradiction of opinion leaders being on one side highly unique people but also willing to share while studies on Need for Uniqueness intended that high Need for Uniqueness individuals are less likely to share information on innovations. Another influence I faced during browsing through gaming social networks in which game usage was publicly disclosed. As I know from word-of-mouth, some of my friends and my own experience, some gaming communities are rather strict and that for example in a World of Warcraft community people who play life simulation games such as TheSims are not highly acknowledged. Hence, I combined these influences to the present work.

I want to use the opportunity to thank the people who supported me on my way to this point. First of all, I want to thank my first supervisor Dr. Risselada for his helpful comments and advice. Also I want to thank Dr. Eggers for his support. I thank my fellow students who gave me advise and supported me during the semester on my other courses so that I had enough time to write this work. Last but not least, I want to thank my family for enabling me to this studies and supporting me on my way.

Groningen, June 2014

(5)

Table of Content

1. Introduction ... 1

2. Theoretical Framework ... 4

2.1 Online Social Networks and Sharing... 4

2.2 Public and Private Usage in Online Social Networks ... 5

2.3 Need for Uniqueness ... 6

2.4 Social Influence and Susceptibility to Interpersonal Influence ... 9

2.5 Conceptual Model ... 12 3. Research Design ... 14 3.1 Methodology ... 14 3.2 Design ... 14 3.3 Research Procedure ... 16 4. Findings ... 17 4.1 Preliminary Analysis ... 17

4.2 Results of the Simple Conjoint Analysis ... 18

4.3 Results of the Conjoint Analysis concerning Need for Uniqueness ... 22

4.4 Results of the Conjoint Analysis concerning Susceptibility to Interpersonal Influence ... 24

4.5 Discussion ... 27

5. Conclusion ... 32

References ... 34

Appendix ... 36

Appendix 1: Screenshots of the Conducted Survey ... 36

Appendix 2: Interaction Effects ... 37

(6)

2

List of Figures

Figure 1: Screenshot of Facebook game application ... 1

Figure 2: Screenshot of Steam profile ... 1

Figure 3: Screenshot of a profile of Svpply ... 1

Figure 4: Conceptual model of the relationship between susceptibility to interpersonal influence and need for uniqueness on usage behavior ... 13

Figure 5: Boxplot of the values of Susceptibility to Interpersonal Influence ... 17

Figure 6: Boxplot of values of Need for Uniqueness ... 17

Figure 7: Graph of the answers to the question "If product usage would be shared by default" ... 18

Figure 8: Attribute importance per class of model 4 ... 20

Figure 9: Introduction page of the online survey ... 36

(7)

3

List of Tables

Table 1: Attributes and their levels of the choice-based conjoint experiment ... 14

Table 2: Utility and importance of attributes in the simple model ... 19

Table 3: Overview of the model fit measures for the simple conjoint analysis models with 1 to 5 classes ... 19

Table 4: Overview of the model fit measures for the simple conjoint analysis models with 1 to 5 classes ... 19

Table 5: Level utilities of the attributes per class of model 4 ... 21

Table 6: Utility and importance of the model including the interaction effect of Need for Uniqueness and Sharing Product Usage ... 22

Table 7: Utilities and P-values of the models including the interaction effects of the components of Need for Uniqueness and Sharing Product Usage ... 23

Table 8: Utility and importance of the model including the interaction effect of Need for Uniqueness and Product Uniqueness ... 23

Table 9: Utilities and P-values of the models including the interaction effects of the components of Need for Uniqueness and Product Uniqueness ... 24

Table 10: Utility and P-value of the model including the interaction effect of Need for Uniqueness and Reference Group Product Usage ... 24

Table 11: Utility and P-value of the models including the complex interaction effects of Need for Uniqueness ... 24

Table 12: Utility and importance of the model including the interaction effect of Susceptibility to Interpersonal Influence and Sharing Product Usage ... 25

Table 13: Utilities and P-values of the models including the interaction effects of the components of Susceptibility to Interpersonal Influence and Sharing Product Usage ... 25

Table 14: Utility and P-value of the model including the interaction effect of Susceptibility to Interpersonal Influence and Reference Group Product Usage, and the model including the interaction effect of Susceptibility to Interpersonal Influence and Reference Group Acceptance ... 26

Table 15: Utilities and P-values of the models including the interaction effects of the components of Susceptibility to Interpersonal Influence and Reference Group Product Usage ... 26

Table 16: Utilities and P-values of the models including the interaction effects of the components of Susceptibility to Interpersonal Influence and Reference Group Acceptance ... 26

Table 17: Utility and P-value of the models including the complex interaction effect of Susceptibility to Interpersonal Influence ... 27

Table 18: Overview of supported hypotheses ... 31

Table 19: Overview of calculated interaction effects ... 37

(8)

1

1. Introduction

On many online social networks, sharing information about the usage behavior of products becomes more and more implemented in various tools and processes such as online purchase processes, game applications (see figure 1) and online gaming profiles (see figure 2). The reason for increasing the online visibility of usage behavior of these tools is to trigger word-of-mouth and, hence, influence the follower or friends of user

in online social networks to use these tools, too. As explored in many studies, word-of-mouth is a powerful and efficient source of influencing customers and to spread awareness (Vilpponen, Winter, Sundquist, 2006: 64). It emerges when people communicate with each other. Word-of-mouth is created in online social networks, communities, on blogs, due to product-reviews and more (Reichelt, 2013: 58-59). However, evolving issues due to social influence for instance are perceived as less important when implementing such features as sharing by default.

Online social networks like Facebook (see figure 1), gaming networks such as Steam (see figure 2), but also the upcoming trend of online social marketplaces like Svpply (see figure 3) include features that share user behavior like achievements in applications (Facebook, Steam), playing a game (Steam), or owning a certain products (Svpply, Steam) with the community. The user can sometimes adjust the visibility of such usage information disclosure by changing the privacy settings and the visibility settings of the account on many social networks. However, many users do not use these settings because of unawareness of such options, carelessness or other motives.

Information disclosure online has been examined closely also due to legal discrepancies like privacy issues (Strater, Lipford, 2008: 111). Also the sharing behavior of users has been analyzed in detail (Raban, Rafaeli, 2007; Strater et al., 2008; Hsua, Lin, 2008). As it becomes more and more

Figure 1: Screenshot of Facebook game application

Figure 2: Screenshot of Steam profile

(9)

2 often that information of user's product usage are shared by default, this thesis will provide further insights about the influence on behavior when usage is displayed in online social networks by default. Especially, usage behavior can be influenced when a person has some personal characteristics which make the individual sensitive to public awareness of his/her actions. Such a trait is individuality which is also referred to as the Need for Uniqueness. People with a high Need for Uniqueness want to be different from others and, hence, prefer products with which they can identify and are unique in a certain way. This kind of individuals may not want to display their product usage online because others could copy their behavior and would decrease their distinctiveness.

But social influence showed to be influential on sharing and usage behavior. Social influence plays an important role on usage behavior. People tend to act according to the norms and trends of their social reference group in order to gain or maintain a certain social status within the group but also to be accepted and appreciated (Hsua et al., 2008: 67). Social groups vary in their characteristics and there are some which are stricter, where the members have to do certain things or to own certain products. For example in a community of people who identify themselves by playing strategic and shooter games like League of Legends or Counter Strike, a member who wants to play a casual life simulations such as the Sims might risk to lose his/her status within the group if it is known to other members. However, how much individuals are influenced by their reference groups depends on their social Susceptibility to Interpersonal Influence (Bearden, Netemeyer, Haws, 2011: 136).

Through sharing usage information, a private product usage becomes public. This implies that the usage behavior could change under certain personal traits as Need for Uniqueness or Susceptibility to Interpersonal Influence. Hence, the research question which will be answered in the course of this paper is:

How is usage behavior affected when it becomes public in online social networks under the aspect of Need for Uniqueness and Susceptibility to Interpersonal Influence?

(10)

(11)

4

2. Theoretical Framework

2.1 Online Social Networks and Sharing

The term online social networks is described by a myriad of definitions. However, they all have in common that an online social network is a web-based service which enables individuals to connect themselves with others, to have a profile and to view profiles of other individuals within the system (Kneidinger, 2010: 49). Depending on the platform, many different tools are offered to manage the own identity, to search for information and people, to get in contact with others, to manage these contacts, to get informed about the own social network and to communicate with others (Kneidinger, 2010: 50-51).

In order to participate in social interactions on such networks, it is necessary to disclose personal information in form of sharing (Strater et al., 2008: 111) which is understood in the following as publishing information in form of a comment or a post in online social networks. Due to lack of nonverbal stimuli in an online environment (Reichelt, 2013: 56) and its thereby created anonymity (Mühlenbeck, Skibicki, 2008: 17), information disclosure is a key tool to manage the own identity of a user online and to bond with other users (Strater et al., 2008: 111). Sharing behavior can occur on a regular basis, but also spontaneous (Raban et al., 2007: 2368), it varies in the amount of information shared (Gross, Acquisti, 2005: 73) and the visibility of the shared information to other users (Gross et al., 2005: 79). Because of the given anonymity, the interpersonal communication changes between people. It is easier in online social networks to exceed social limits compared to the offline environment. People are friendlier online, but also ruder and more explicit. Additionally, information exchange can be asynchrony due to the freedom in time and place. It is also limitless in its form, content, identity and environment (Mühlbeck et al., 2008: 18).

Motivations for sharing information online are egoistic and altruistic motives. Egoistic reasons are given when the information sharing is socially or economically rewarded (Hsua et al., 2008: 66-67). A social reward could be gained reputation (Hsua et al., 2008: 72) or structural embedding in the online social network (Goes, Lin, Au Yeung, 2014: 3). But also the sole pleasure of disclose information to others or to get approval for the own behavior can be an egoistic motive (Cheema, Kaikati, 2010: 554). Altruistic motivations are defined as willing to provide information in order to increase the welfare of others without any personal returns (Hsua et al., 2008: 66-67).

(12)

5 owned information. Moreover, if the information was attributed with the user's personal identification the information was shared more frequently.

A similar finding was observed in a study conducted by Cheema et al. (2010: 559-600). It was differentiated between publicly used products for which the consumption can be seen by others, and privately used products (Cheema et al., 2010: 554-555). An example for these product categories can be a jacket for a publicly used good and toilet paper as a privately used product. The consumers were segmented according to their Need for Uniqueness to which will be referred to in chapter 2.3. The study showed that consumers with a low Need for Uniqueness are more likely to discuss information about a product and to recommend buying a product, thus to share information about a product, when they do own it. This applies also for high Need for Uniqueness consumers when the product is privately used, and when product details are discussed (Cheema et al., 2010: 559-600).

These results show that ownership positively influence the likelihood of sharing information about a product. As ownership is a requirement of product usage, it can be inferred that ownership would not affect the usage behavior additionally if the use would be public by default.

2.2 Public and Private Usage in Online Social Networks

Usage can be differentiated between public and private use. The term usage is understood as the use of an owned good which can be a product or a service. Public usage in an online social network is hereby defined as a usage which is shared online in a social network. For example, on Facebook when playing a social network game, it is sometimes offered to share certain achievements online. In this case sharing an achievement represents the actual usage of this game by a user. Another example is Steam where it is shown on the user's profile what game is played in this very moment or the last time the user played a particular game (see figure 2). Contrary to public usage, private usage is not shared online. Examples are watching a TV show without sharing this information online, wearing clothes without mentioning it at an online social platform, or not sharing an achievement of a social network game.

When usage is observed by others in online social networks the usage behavior of the person is influenced (Goes et al., 2014: 15). Hence, it can be inferred that usage behavior varies between products used in public and products used in private, especially thought that privately used products become publicly used products by sharing.

(13)

6

2.3 Need for Uniqueness

A personal attribute which influences the product usage behavior as well as the sharing behavior is the Need for Uniqueness. The Need for Uniqueness influences the product preference of a person (Cheema et al., 2010: 554) and is defined as the desire to be different relative to others through acquisition, usage, and disposition of consumer goods (Tian, Bearden, Hunter, 2001: 52). The purpose of Need for Uniqueness is the development and the enhancement of the own self-image and social image (Tian et al., 2001: 52). The concept originates of the Theory of Uniqueness which contains the need of an individual to perceive him/herself as being different from others. If others become too similar, a individual with a high Need for Uniqueness behave self-distinguishing in order to reclaim his/her self-esteem (Tian et al., 2001: 52). There are many different ways to express one's uniqueness like displaying the own possession, having special style of interpersonal interaction or being an expert in a different domain. Individuals vary in expressing their uniqueness through consumer behavior and ownership (Tian et al., 2001: 52). However, the motivation to be unique varies across people and situations (Lynn et al., 1997: 602-603).

The Need for Uniqueness derives of three dimensions: Creative Choice Counter-Conformity, Unpopular Choice Counter-Conformity, and Avoidance of Similarity (Bearden et al., 2011: 42). Creative Choice Counter-Conformity is given when individuals make different choices than others but which should be considered as good choices by others. Unpopular Choice Counter-Conformity is specified as the different choice of products or brands which are socially risky and differ from the group's norms. Such behavior leads usually to a lowered social image, but due to the possibility that such goods become socially accepted over the time is increasing the chance that the individual becomes an innovator or an opinion leader. Both Counter-Conformities lead to an enhanced self-image. The Avoidance of Similarity is given when individuals with a high Need for Uniqueness avoid and devalue goods that are perceived as a commonplace. Individuals with a high Need for Uniqueness even dispose and discontinue the use of goods to avoid similarity in order to create a unique self-image. This behavior requires a willingness to change the own consumer behavior and preferences (Tian et al., 2001: 52-53).

(14)

7 to buy a public product. Hence, Need for Uniqueness has a negative influence on the willingness to recommend public products (Cheema et al., 2010: 560).

Ownership and the Need for Uniqueness in interaction have a significant influence on sharing information about public products. If other individuals would own the recommended public product, the uniqueness of a product would be decreased and, therefore, its attractiveness (Cheema et al., 2010: 558). High Need for Uniqueness individuals are also less likely to adopt a publicly consumed product which is already adopted by a friend. Probably they want to be first adopters (Cheema et al., 2010: 556) which would be coherent with the higher level of innovativeness high Need for Uniqueness individuals usually have (Lynn, Harris, 1997: 603-604).

Low Need for Uniqueness individuals are not affected by the product category of publicly or privately consumed products (Cheema et al., 2010: 557) and are willing equally to recommend a product and to discuss product details (Cheema et al., 2010: 560). The willingness to discuss product details is not significantly affected by the Need for Uniqueness if the product is owned by the discussion participants. It has shown that ownership of a product increases the willingness to discuss details about the owned product (Cheema et al., 2010: 559). A reason for this result could be that people gain social capital by discussing product details and that product detail discussion is perceived less as a process that leads to product adoption than product recommendation (Cheema et al., 2010: 561-562).

Based on the findings of Cheema et al. (2010), it can be assumed that individuals with a high Need for Uniqueness are less willing to share the usage in public of an owned product because product sharing could be perceived as an engagement in word-of-mouth and therefore as an engagement of product recommendation. People with a low Need for Uniqueness in contrary are not negatively affected in their usage behavior when it becomes public given that product ownership is given. This leads to the following hypothesis.

H2: The Need for Uniqueness affects usage behavior negatively if the usage is shared in an

online social network.

(15)

8 Goods which are scarce and unique are perceived as effective status symbols. Materialism reflects the sole importance of possessing material goods (Lynn et al., 1997: 603).

Individuals with a strong desire for unique products tend to buy and use scarce, customized, outmoded and innovative products as well as they like to buy in small, unique retail stores. Innovative products are usually new products which do not often gain immediate and popular acceptance. Hence, they are first adopted by small groups of innovators and early adopters. It is found that self-attributed Need for Uniqueness and the tendency to be an innovator are positively related (Lynn et al., 1997: 603-604).

An individual with such characteristics might be sharing product usage less willingly because of the risk of imitation (Cheema et al., 2010: 554). This risk could be altered if the used product is scarce or is highly individual due to product customization. Hence, it also seems to be likely that individuals with a high Need for Uniqueness would share the use of such a product in order to gain or display a certain status (Lynn et al., 1997: 603). Therefore, it can be assumed that the uniqueness of a product is positively related to the Need of Uniqueness and could alter the relationship between the Need for Uniqueness and public usage behavior.

H3: The uniqueness of the used product reduces the effect Need for Uniqueness has on public

usage behavior.

As discussed earlier, individuals with a high Need for Uniqueness are less likely to adopt a product which is already adopted by a friend (Cheema et al., 2010: 556). Hence, in order to be perceived as different, an individual with a high Need for Uniqueness would be less willing to share the usage of a certain products when the reference group is using this product as well.

H4: The product usage of the reference group enhances the effect Need for Uniqueness has on

public usage behavior.

Regarding the effect the reference group's usage of a product might has in relation with Need for Uniqueness on product sharing, an individual with a high Need of Uniqueness would not want to share the usage of this particular product either and rather not use the product even if the product is unique, since using this unique product would not lead to a differentiation between the high Need for Uniqueness individual and the reference group.

H5: The product usage of the reference group reduces the moderation effect of the uniqueness

(16)

9

2.4 Social Influence and Susceptibility to Interpersonal Influence

There are three causes which lead to an action of a person: The person is a very active individually and acts without being influenced, external events to the focal situation affect the person and lead to an action of this person, or the individual is influenced by friends, family and other people to do something (Goyal, Bonchi, Lakshmanan, 2010: 4). Hence, social influence plays an important role in a person's behavior. For instance, a study showed that students are more likely to have a profile with adapted privacy settings in an online social network if their friends have such profiles, too (Hargittai, 2010). Interaction with other people influences a person (Goes et al., 2014: 2). This applies also for the online environment (Goes et al., 2014: 15).

In the process of innovation adoption, interpersonal influence has an important role in the later stages of the process and leads to the behavioral change (Kim, 2007: 16-17). Some people exert a higher level of influence on other individuals and are therefore called opinion leader (Goes et al., 2014: 2). But also early adopters, innovators (Kim, 2007: 16-17), experts, consumers with a high customer life time value and heavy users (Haenlein, Libai, 2013: 73-74) are more influential than others. Hence, the level of influence of people varies due to their personality and their behavior. As described earlier in 2.1, through the absence of non-verbal stimuli in online social networks, a general anonymity is created which influence positively social identity and the social behavior of users in an online social network (Reichelt, 2013: 56). The actual behavior of a user is more in the focus than the user's personality. Also prejudice and stereotypes are less existent due to the limited information provided by the actual user (Reichelt, 2013: 56). Thereby, users are more oriented towards the norm in an actual situation and more linked to a certain group of other users (Reichelt, 2013: 56).

(17)

10 A reference group is a group to which a person is oriented and of which a person influenced by. The influence of a reference group is defined as the influence which originates by a real or imaginary individual or a group and affects the evaluation, the aspiration and the behavior of the influenced individual (Bearden et al., 2011: 140). The motivation of reference group influence on an individual has consists out of informational influence and normative influence (Bearden, Netemeyer, Teel, 1989: 474).

Informational influence describes the tendency of information acceptance from others due to the enhancement of an individual's knowledge about the environment or the ability to manage some facet of the environment such as product purchase (Bearden et al., 2011: 140). This kind of influence derives by the search for information from experts or conclusions due to observations of other's behavior. This leads to the process of internalization of the information to increase a person's knowledge about the facet of environment. Informational influence affects the consumer decision process concerning the evaluation and choice of goods (Bearden et al., 1989: 474).

Normative influence can be divided in utilitarian influence and value expressive influence (Bearden et al., 1989: 474). The utilitarian influence of a reference group is reflected by the compliance with others regarding their expectations in order to perceived mediation of rewards or punishments of visible behavior. Another motive is the gain of rewards and the avoidance of punishments (Bearden et al., 2011: 140). The value expressive influence is the desire of a person to enhance or support his/her self-concept for others (Bearden et al., 2011: 140). This influence is reflected by the identification process when a person adopts a behavior or opinion of others (Bearden et al., 1989: 474).

The reference groups provide norms for the behavior of the member and has expectations for the role of a member within a group (Dressler et al., 2009: 7). Also the behavior of such a peer can change the member's understanding of a focal behavior. The peer influences a member in several ways. It can raise awareness to a product or a feature, persuade individuals to change their expectations and enhance imitation, as well as social learning (Aral, 2010: 2). Peer influence is consisting out of flexible influence mechanisms. It is casual driven but also dynamically evolving due to loops of feedback (Aral, 2010: 2-3). Group influence can also emerge indirectly. For example, if many members of a group start buying sport equipment in order to lose weight, others may not buy such products but start a diet or get a membership in a gym (Aral, 2010: 3).

(18)

11 of a product. If a product is costly, the advice of the group is requested and, hence, very influential (Aral, 2010: 4). Consumer Susceptibility to Interpersonal Influence is a personal characteristic that is defined as the need to identify with others or enhance a person's image for others by a person's purchase behavior (Bearden et al., 2011: 136). It varies across individuals and is related to other characteristics such as self-esteem or intelligence. It is found that individuals with a low interpersonal confidence are more susceptible to peer influence. Susceptibility to Interpersonal Influence as the motivation of reference group consists out of normative and informational influence (Bearden et al., 2011: 136) which are referred to as Normative Susceptibility and Informational Susceptibility in the following. However, if a person is relative susceptible in one situation, the person tends to be susceptible in other social situations as well. Interpersonal influence manifests like reference group influence through informational or normative influence (Bearden et al., 1989: 473-474). Concerning the usage behavior, it can be assumed that the Susceptibility to Interpersonal Influence affects the product usage positively if it becomes public in an online social network because the similarity to others can be displayed on this way.

H6: Susceptibility to Interpersonal Influence affects product usage behavior positively if it is

shared in an online social network.

According to this hypothesis, it is also likely that a person with a high Susceptibility to Interpersonal Influence will use a product publicly if the reference group uses the product as well.

H7: The product usage of the reference group is positively related to Susceptibility to

Interpersonal Influence when the product is used publicly.

As mentioned earlier, the role within a group and norms are providing a guideline for a member's behavior. According to the Social Identity Theory, people classify themselves into social categories and behave according to their assumed and identified roles. Also the expectations of other members of the group influence a member's behavior. This social identification is an important factor for group formation, and for members to perceive themselves as such (Hsua et al., 2008: 67).

(19)

12 2008: 67). However, Hsua et al. (2008: 72) found that social norms do not influence the user's intention to share information.

Regarding this information, an individual with a high degree of Susceptibility to Interpersonal Influence would act according these norms and, therefore, would not use a product visibly to the reference group if the usage would not be accepted by the group. This leads to the following hypothesis:

H8: The acceptance of the usage of a product by the reference group determines the direction

of the effect Susceptibility of Interpersonal Influence has on public usage behavior.

The distinction between the actual usage and the acceptance of the usage of a product is important as these two factors are not always jointly occurring. It seems likely that if the product usage is accepted by the group, group members use the product. However, this has not to be the case. Sometimes a certain behavior is accepted by the reference group, but group members do not behave this way. For example in some groups it is acceptable to have an artificial hair color such as pink or green, however, none of the group members has such a hair color at the moment. Vise versa, that a group does not accept a certain behavior, leads not necessarily to the avoidance of this behavior of a group members. However, this leads usually to social punishment and to a lower status within the group (Stieglitz, 2008: 75).

2.5 Conceptual Model

Product usage behavior might change if it becomes public in online social networks by default. As Cheema et al. (2010) have shown in their study, the information sharing behavior changes if products are used privately or publicly. As mentioned in the introduction, providers of online social networks tend to implement features by which private usage becomes public usage. In the previous chapters, it is shown that many characteristics of a person and its environment influence a person's consumer behavior, especially if the behavior is observable by other individuals. The degree of Need for Uniqueness of a person, as well as the degree of Susceptibility to Interpersonal Influence of a person has strong influence on a person's behavior. In this paper, the product usage behavior is the dependent variable and shows the characteristics of being public or private. This means that it is either shared or not shared. The Need for Uniqueness and the Susceptibility to Interpersonal influence are the independent variables (see figure 4). Both characteristics consist out of multiple dimensions. The Need for Uniqueness has three dimensions and the Susceptibility to Interpersonal Influence includes two dimensions.

(20)

13 group uses a certain brand or product, a highly susceptible person would tend to adopt this brand or product in order to gain a positive self-concept, to enhance a positive self image for the reference group, to obtain reward or to avoid punishment. The same refers to the avoidance of a product. According to the utilitarian influence, this applies especially if the behavior of a person is public and therefore observable. Regarding this reasoning, it is expected that the Susceptibility to Interpersonal Influence of a person is positively related to product usage (see figure 4). As discussed in 2.4, two factors are moderators for the relationship of Susceptibility to Interpersonal Influence and public usage behavior: The acceptance of the reference group and the product usage of the reference group (see figure 4). It is expected that both moderators are influencing the relationship between Susceptibility to Interpersonal Influence and public product usage positively.

Individuals with a high Need for Uniqueness tend to desire unique products. As the risk derives by public usage that other people imitate a highly individual person, it can be assumed that such individuals do not want to share usage behavior and would, thus, change their behavior by using a different online social network or a different but similar product or brand. Hence, Need for Uniqueness has a negative effect on usage behavior if it is shared online by default (see figure 4). As described in chapter 2.3, the trait Need for Uniqueness is influenced in its effects by the uniqueness of the product, as well as the product usage of the reference group. This means that the relationship of Need for Uniqueness to the product usage behavior is moderated by the product characteristics concerning the uniqueness of a product and the product usage of the reference group (see figure 4). It is expected that the uniqueness of the product is reducing the negative effect Need for Uniqueness has on public usage behavior whereas the product usage of the reference group enhances this negative effect.

(21)

14

3. Research Design

3.1 Methodology

In order to test the hypotheses, an experimental environment needs to be created in which the moderating variables can be manipulated. The independent variables are personal characteristics and cannot be manipulated therefore. Also a control group in which the product usage is not shared needs to be created. An efficient way to create such situations is a choice-based conjoint experiment. This method enables the selection of attributes which represents in this case factors of the situations and also attribute levels which describe specific characteristics of the attributes like the color of a product and in this case for instance that friends use the product as well (Malhotra, 2010: 702). It also includes the characteristics an experiment provides. With the choice-based conjoint analysis it can be determined how relative important factors are for the respondent's decision and the utilities of the attribute levels (Malhotra, 2010: 701). Additionally, the personal characteristics of the respondents need to be measured on the degree of Need for Uniqueness and the Susceptibility to Interpersonal Influence.

The chosen method offers many advantages such as a fast and efficient way to simulate the needed experimental environment in which the moderating variables can be manipulated. This method also has its downsides. In order to avoid the influence of the preference of a certain product or product type, the product description was rather basic and therefore more abstract which caused some problems for respondents in understanding the task.

3.2 Design

The conducted online survey contained a choice-based conjoint experiment, two scales testing Need for Uniqueness and Susceptibility to Interpersonal Influence, as well as questions concerning demographical data. As Facebook is the most popular online social network, an example looking similar to Facebook was created in the beginning (see Appendix 1, Figure 9). Given that interaction effects were tested, a fractional design was

chosen.

The choice-based conjoint analysis consisted out of 10 sets per respondent in order to avoid a fatigue effect of the respondent. In each set, three alternatives were offered. Four attributes with each two levels were chosen (see table 1). They included the moderating variables, as well as the attribute Sharing Usage Behavior which is

Attributes Levels Sharing Usage Behavior Not Shared Shared Product Uniqueness Ordinary Product Unique Product Reference Group Product Usage

Friend do not use the product

Friends use the product

Reference Group Acceptance

Community does not accept usage

Community accepts usage Table 1: Attributes and their levels of the choice-based

(22)

15 based on the shared and not shared conditions of the experiment. Given this number of attributes and the related number of levels, 16 combinations of stimuli were possible. As 3 attributes per choice set were used, 560 different choice sets could be provided. As 10 sets per respondents were provided, a fractional factorial design was given. A screenshot of a set of the choice-based conjoint experiment can be found in appendix 1.

The choice-based conjoint experiment is orthogonal and balanced. The balance of a conjoint analysis referrers to the equal distribution of all levels of attributes. It showed that some levels were provided less often than others. The maximum deviation is about 0.6% and the minimum deviation is about 0.2%. This deviation is very small and, hence, the choice-based conjoint experiment is still balanced. An orthogonal conjoint analysis is applied when all combination of the attributes' levels are equally shown in the survey. In this survey, some variants of alternatives were shown more often than others. However, the deviation is about 0.3% meaning that this conjoint analysis is orthogonal as the deviation is very small.

Next to the alternatives, a No Choice Option in form of a question was provided ("If you could decide yourself, would you want to share the product usage in your chosen situation?"). This question was asked in order to provide further insights on how willingly product usage is shared.

After the sets of the conjoint analysis, the scale to measure the Susceptibility to Interpersonal Influence of Bearden, Netemeyer, and Teel was provided (Bearden et al., 2011: 136). The scale contained 12 items which referred to the Informational (4 items) and Normative Susceptibility (8 items). It is measured in a 7-point rating scale from "strongly disagree" to "strongly agree". The scores are summed. The possible range is a value of 12 to 84 (Bearden et al., 2011: 136).

In the following, the scale of Tian, Bearden, and Hunter was used to measure the degree of Need for Uniqueness (Bearden et al., 2011: 42). This scale contained 31 items and provides insights on the degree of the dimensions of Need for Uniqueness: Creative Choice Counter-Conformity (11 items), of Unpopular Choice Counter-Conformity (11 items), and of Avoidance of Similarity (9 items). It is measured in a 5-point Lickert-type scale from "strongly disagree" to "strongly agree". To calculate the score, the average is taken. In the literature, for the Need for Uniqueness a mean of 2.60 is observed, as well as a range of 1.06 to 4.55 (Bearden et al., 2011: 124).

(23)

16 The sample size should be minimum 200 respondents. Hence, 50 values can be used to estimate the utilities for the four attributes. The target population should include equally female and male participants and also consist out of some non-students in order to be representative and generalizable. This survey aimed to target especially European respondents in order to avoid a possible invalidity of scales due to different cultural backgrounds. As the topic is about product usage in online social networks, each respondent needs to participate in an online social network otherwise the respondent might not be able to recognize the provided situations.

3.3 Research Procedure

First of all, the sample was analyzed its representativeness and on respondents who did not participate in online social networks. Respondents who did not meet this requirement were excluded from the data set. The scale of the Need for Uniqueness and the scale of Susceptibility to Interpersonal Influence were evaluated. Subsequently, the simple conjoint analysis was estimated meaning that the four attributes and their levels were evaluated without any additional variables. Afterwards, the simple conjoint analysis was conducted by creating more classes (segments) in order to get further insights.

(24)

17

4. Findings

4.1 Preliminary Analysis

A sample size of n = 205 is collected. Hence, the sample size is sufficient. The sample consists out of 49.3% women (n = 101) and 50.7% men (n = 104). The average age is 26.41 years. Most participants are from Germany (64.4%), thus, most participants are European. About 11.71% come from other continents. 69.8% of the sample are students. 30.2% of the respondents are employed or have a different professional status than student. Most of the respondents are frequently using online social networks (17.6% once per day, a 50.2% few times per day, 14.6% always online). 2.9% (6) of the participants does not use online social networks. Hence, the respondents

which do not use online social networks are excluded from dataset hereafter. All in all, the collected sample is representative.

The Susceptibility to Interpersonal Influence has a mean value of 40.75. The values of the respondents range from 12 to 74 (see figure 5). The difference to the maximum score possible is 10. The Normative Susceptibility which consists out of 8 items shows a mean of 23.04 a minimum of 8 and a maximum of 49. The maximum value achievable is 56. The mean of the Informative Susceptibility to Interpersonal Influence is 17.72. The minimum value is 4 whereas the maximum value is 28. This range corresponds to the maximum range possible. As no mean scores are reported in the literature (Bearden et al., 2011: 137), it is to be assumed that the distribution of the Susceptibility to Interpersonal Influence is representative as the given ranges meet the possible ranges.

The Need for Uniqueness shows a overall mean of 2.69, a maximum of 4.52 and a minimum of 1 (see figure 6). This shows that the observations concerning the Need for Uniqueness in the conducted survey are representative, as in previous studies a mean of 2.60 was observed, as well as a range of 1.06 to 4.55 (Bearden et al., 2011: 124). Concerning the Creative Choice Counter-Conformity a mean of 2.82 and a range from 1.00 to 4.73 is observed. The Unpopular Choice Counter-Conformity shows a mean of 2.72, a minimum of 1.00 to 4.64 and for the Avoidance of Similarity a mean value of 2.50 is calculated. The range for Avoidance of Similarity is from 1.00 to 4.56.

Figure 5: Boxplot of the values of Susceptibility to Interpersonal Influence

(25)

18 8.01% 39.74% 25.13% 8.51% 18.62%

If product usage would be shared by

default

_{I would continue using the}

product. I want it to be shared.

I would continue using the product. I don't care. I would search for another way to use the product. I would stop using the product.

I don't know.

The statement "If product usage would be shared by default" showes that few want their product usage to be shared (8.0%) and that some would stop using a product (8.5%). Most of the participants would continue using the product (47.7%) of which most do not care if the usage is shared (39.7%). 25.1% of the respondents would search for another way to use a product and 18.6% do not know what they would do.

4.2 Results of the Simple Conjoint Analysis

The conjoint analysis of the four attributes Sharing Usage Behavior, Product Uniqueness, Reference Group Product Usage and Reference Group Acceptance has a squared Log-Likelihood of 4091.0123 and a Log-Likelihood of -1981.1030 with 5 parameters. The overall model is significant with a P-value of 1.4e-694 (< 0.05). The overall R² is at 0.1055. As the R² is lower than 0.6, the model is not strong (Malhotra, 2010: 697). The Hitrate of the prediction model is 53.32%. The prediction error is about 0.4668. This means that more than half of the cases are correctly classified.

All four tested attributes are significant (Sharing Usage Behavior: p = 6.5e-47 < 0.05; Product Uniqueness: p = 9.1e-7 < 0.05; Reference Group Usage: p = 1.6e-7 < 0.05; Reference Group Acceptance: p = 8.0e-35 < 0.05). However, the No Choice Option is insignificant (p = 1.00 > 0.05) and hence does not influence the product usage choice (see table 2).

The model shows that Sharing Usage Behavior (37.86%) and Reference Group Acceptance (31.89%) are the most important attributes. Whereas Sharing Usage Behavior shows a negative influence on the choice (Product usage is shared: -0.4195), Reference Group Acceptance effects the choice of product usage positively (Group accepts the product usage: 0.3533). However, Sharing Usage Behavior has a stronger effect on the choice. Both, Product Uniqueness (Unique product: 0.1384) and Reference Group Product Usage (Friends use the product, too: 0.31483), have a rather low importance in the product usage choice (see table 2).

The No Choice Option was provided as the question if the respondent would want to share the product usage. Overall, in 69.70% of the cases the participants did not want to share the product

(26)

19 usage. The No Choice Option showed to have a negative effect on the choice of product usage meaning that the utility of "Yes, I want to share my product usage" is about -0.0484 and meaning that not sharing product usage is more in favor of the respondents. However, the effect is insignificant and can only be used as a support for found effects (see table 2).

Attribute Level Utility Wald P-Value Importance Sharing Usage

Behavior

Not Shared 0.4195 206.9031 6.5e-47 37.86%

Shared -0.4195

Product Uniqueness

Ordinary Product -0.1384 24.1102 9.1e-7 12.49%

Unique Product 0.1384

Reference Group Product Usage

Friend do not use the product -0.1483 27.4958 1.6e-7 13.39% Friends use the product 0.1483

Reference Group Acceptance

-0.3533 151.5401 8.0e-35 31.89% Community accepts usage 0.3533

No Choice Option

Usage should not be shared 0.0484 0.0000 1.00 4.37% Usage should be shared -0.0484

Table 2: Utility and importance of attributes in the simple model

In order to get further insights in these factors and their influence on respondents' product usage choice, classes were created. As the No Choice Option showed to be insignificant, it is excluded in further calculations concerning the segmentation. In table 3 to 4, the model fit measures are shown. All models are significant with a p-Value blow 0.05. The CAIC (Consistent Akaike's Information Criterion) which is a good comparison between models due to its included penalization for the number of parameter is lower for Model 4 than for the other models. For the other information criterions, the values are decreasing the more classes are added. However, the classification error is 0.1308 for Model 4 which is lower than for the models with 3 and 5 classes and the R² of Model 4 is at 0.3008 which is an improvement compared to Model 1. Therefore, model 4 with 4 classes is selected.

Number of Classes

LL BIC(LL) AIC(LL) AIC3(LL) CAIC(LL) Npar Model 1 1 -1981.1030 3983.3793 3970.2061 3974.2061 3987.3793 4

Model 2 2 -1868.3837 3784.4071 3754.7674 3763.7674 3793.4071 9

Model 3 3 -1839.9644 3754.0351 3707.9288 3721.9288 3768.0351 14

Model 4 4 -1820.4679 3741.5085 3678.9357 3697.9357 3760.5085 19

Model 5 5 -1805.8677 3738.7747 3659.7354 3683.7354 3762.7747 24 Table 3: Overview of the model fit measures for the simple conjoint analysis models with 1 to 5 classes

L² df p-value Class.Err. R² Model 1 3962.2061 195 6.3e-694 0.0000 0.1055 Model 2 3736.7674 190 1.2e-650 0.0734 0.2243 Model 3 3679.9288 185 3.6e-642 0.1327 0.2742 Model 4 3640.9357 180 2.2e-637 0.1308 0.3009 Model 5 3611.7354 175 1.2e-634 0.1582 0.3174

(27)

20 Model 4 is overall significant (p = 2.2e-637 < 0.05). The same holds for the four tested attributes (Sharing Usage Behavior: p = 3.2e-35 < 0.05; Product Uniqueness: p = 1.7e-5 < 0.05; Reference Group Usage: p = 9.4e-12 < 0.05; Reference Group Acceptance: p = 7.1e-9 < 0.05).

In the first class Unique Users of Model 4, Product Uniqueness (45.85%) and Reference Group Acceptance (41.17%) have a high influence on the product usage choice (see figure 8). These two important attributes have a positive effect on the product choice, meaning that for unique products (0.284) and the community accepts the product use (0.255) the utility is positive (see table 5). The Ordinary User class in contrast has a high concern in Sharing Usage Behavior (46.72%), but also takes Reference Group Acceptance (25.28%) and Reference Group Product Usage (18.03%) into account for product usage decisions (see figure 8).The effect of Sharing Usage Behavior is negative and has a strong impact on product usage choice (Shared: -1.5779). The other attributes show positive relationships (see table 5). The Private User class values Sharing Usage Behavior (64.75%) even more than the Ordinary User class. The second important attribute is Reference Group Product Usage (25.02%) (see figure 8). As in the Unique User class and the Ordinary User class, Sharing Usage Behavior has a negative effect. However, Reference Group Product Usage shows a negative relationship as well, meaning that the utility for the level "Friends use the product" is -0.5132 (see table 5). The Public User class shows a higher importance for Reference group Acceptance (35.89%). But also the other attributes are relatively important and show no lower importance than 15% (see figure 8). Besides Product Uniqueness, all attributes have a positive relationship to the choice of product usage. Exceptional is that the Public User class is the only class which shows a positive utility on shared product usage (Shared: 0.4786) (see table 5).

Figure 8: Attribute importance per class of model 4 46.72% 64.75% 20.65% 45.85% 9.96% 8.42% 15.84% 18.03% 25.02% 27.61% 41.17% 25.28% 35.89% 0% 20% 40% 60% 80% 100% Unique User Ordinary User Private User Public User

Attribute Importance per Class

Sharing Usage Behavior Product Uniqueness

(28)

21

Attribute Level Utility

(Unique User) Utility (Ordinary User) Utility (Private User) Utility (Public User) Sharing Usage Behavior Not Shared 0.0321 1.5779 1.3283 -0.4786 Shared -0.0321 -1.5779 -1.3283 0.4786 Product Uniqueness Ordinary Product -0.2840 -0.3366 0.1727 0.3671 Unique Product 0.2840 0.3366 -0.1727 -0.3671 Reference Group Product Usage

Friend do not use the product -0.0484 -0.6091 0.5132 -0.6399 Friends use the product 0.0484 0.6091 -0.5132 0.6399

-0.2550 -0.8539 -0.0373 -0.8318 Community accepts usage 0.2550 0.8539 0.0373 0.8318

Table 5: Level utilities of the attributes per class of model 4

(29)

22

4.3 Results of the Conjoint Analysis concerning Need for Uniqueness

First of all, the interaction effect of Need for Uniqueness and the attribute Sharing Product Usage was tested (see table 6). It is found that this interaction effect is significant (p = 0.0046 < 0.05). The overall model is significant as well (p = 1,5e-693 < 0.05) and shows an R² of 0.1071 which is an improvement of the model compared to the simple one-class model. All attributes beside the No Choice Option are significant as well. The direction of the relationships and the distribution of the importance of the attributes to the respondents are similar to the simple model. The interaction effect of Need for Uniqueness and Sharing Product Usage has a positive relationship (0.1110) to the product usage choice (see table 6). The interaction effect of Need for Uniqueness and Sharing Product Usage enhances the direction of Sharing Usage Behavior as the utilities are adding up. This means that when a high Need for Uniqueness is given, the probability that the individual would decide for a product usage which is shared decreases even more. Additionally, the attribute Sharing Usage Behavior becomes even more important as the importance of the interaction effect adds to the importance of Sharing Usage Behavior.

Behavior

Not Shared 0.4221 208.1849 3.4e-47 32.07%

Shared -0.4221

Reference Group Product Usage

Interaction

Effect Sharing Product Usage x Need for Uniqueness

0.1110 8.0547 0.0046 16.86%

Table 6: Utility and importance of the model including the interaction effect of Need for Uniqueness and Sharing Product Usage

(30)

23 effect of Sharing Product Usage and Avoidance of Similarity is insignificant with a P-value of 0.21 (> 0.05) (see table 7).

Attribute Interaction Effect Utility Wald P-Value Interaction Effect Sharing Product Usage x Creative Choice

Counter-Conformity

0.0578 2.7379 0.098

Interaction Effect Sharing Product Usage x Unpopular Choice Counter-Conformity

0.0733 3.3844 0.066

Interaction Effect Sharing Product Usage x Avoidance of Similarity

0.0416 1.6006 0.21 Table 7: Utilities and P-values of the models including the interaction effects of the components of Need for

Uniqueness and Sharing Product Usage

The interaction effect between Product Uniqueness and Need for Uniqueness is significant (p = 0.0086 < 0.05)(see table 8). The overall model is also significant (p = 8.5e-694 < 0.05) and is better than the simple model as the R² value is 0.1084. The four attributes are significant as well(see table 8). The interaction effect of Product Uniqueness and Need for Uniqueness shows a positive relationship (0.1019). Given that the utility of the interaction effect adds to the utility of the related attribute, this means that the higher the Need for Uniqueness, the higher the preference for unique products. Also the importance of Product Uniqueness increases when the interaction effect is added. The distribution of the other attributes' importance is similar to the simple conjoint model.

Behavior

Not Shared 0.4209 207.5740 4.6e-47 32.77%

Shared -0.4209

Usage should not be shared -0.0187 0.0000 1.00 1.46% Usage should be shared 0.0187

Interaction

Effect Product Uniqueness x Need for Uniqueness

0.1019 6.9218 0.0086 15.87%

Table 8: Utility and importance of the model including the interaction effect of Need for Uniqueness and Product Uniqueness

(31)

24

Attribute Interaction Effect Utility Wald P-Value Interaction Effect Product Uniqueness x Creative Choice

Counter-Conformity

0.1143 11.3067 0.00077

Interaction Effect Product Uniqueness x Unpopular Choice Counter-Conformity

0.0275 0.4832 0.49

Interaction Effect Product Uniqueness x Avoidance of Similarity

0.0524 2.6425 0.10 Table 9: Utilities and P-values of the models including the interaction effects of the components of Need for

Uniqueness and Product Uniqueness

However, the interaction effect between Reference Group Product Usage and Need for Uniqueness shows to be insignificant (p = 0.28 > 0.05) (see table 10).

Attribute Interaction Effect Utility Wald P-Value Interaction Effect Reference Group Product Usage x Need

for Uniqueness

-0.0424 1.1897 0.28 Table 10: Utility and P-value of the model including the interaction effect of Need for Uniqueness and

Reference Group Product Usage

Additionally, more complex interaction effects are tested including two attributes and the Need for Uniqueness (see table 11). It shows that both, the interaction effect between Sharing Product Usage, Product Uniqueness and Need for Uniqueness (p = 0.15 > 0.05), and the interaction effect between Product Uniqueness, Reference Group Usage and Need for Uniqueness (p = 0.51 > 0.05), are insignificant. The interaction effect between Sharing Product Usage, Reference Group Product Usage and Need for Uniqueness is marginal significant (p = 0.099 < 0.1). The utility of this interaction effect is 0.0635 (see table 11).

Attribute Interaction Effect Utility Wald P-Value Interaction Effect Sharing Product Usage x Product

Uniqueness x Need for Uniqueness

-0.0543 2.0708 0.15

Interaction Effect Product Uniqueness x Reference Group Product Usage x Need for Uniqueness

-0.0253 0.4336 0.51

Interaction Effect Sharing Product Usage x Reference Group Product Usage x Need for Uniqueness

0.0635 2.7199 0.099 Table 11: Utility and P-value of the models including the complex interaction effects of Need for Uniqueness

4.4 Results of the Conjoint Analysis concerning Susceptibility to

Interpersonal Influence

(32)

25 Acceptance: p = 5.0e-35 < 0.05). The No Choice Option is insignificant like in the other tested models (p = 0 1.0 > 0.05). The interaction effect has a positive influence on the product usage choice, as it is 0.0098 (see table 12). This means that the effect of Sharing Usage Behavior is enhanced. The underlying effect of Sharing Usage Behavior is negative which implies that if product usage is shared, it has a negative effect on the product usage choice. This underlying effect is enhanced by the interaction effect of Sharing Usage Behavior and Susceptibility to Interpersonal Influence given that the utility of the interaction effect is added to the underlying effect. For instance, a person with a low Susceptibility to Interpersonal Influence is less likely to avoid product usage if the product is shared in online social networks. Also the importance of Sharing Usage Behavior, as the importance value of the underlying effect and the related interaction effect are adding up.

Behavior

Not Shared 0.4222 207.1532 5.7e-47 29.81%

Shared -0.4222

Interaction

Effect Sharing Product Usage x Susceptibility

0.0098 17.5405 2.8e-5 22.89%

Table 12: Utility and importance of the model including the interaction effect of Susceptibility to Interpersonal Influence and Sharing Product Usage

In the following, the components of Susceptibility to Interpersonal Influence are tested on an interaction with Sharing Product Usage (see table 13). It showed that the interaction effect of Normative Susceptibility with Sharing Product Usage is significant (p = 1.9e-7). However, the interaction effect between Sharing Product Usage and Informational Susceptibility is insignificant (p = 0.7 > 0.05). Like the interaction effect of the overall Susceptibility to Interpersonal Influence, the utility of the interaction effect of normative Susceptibility and Sharing Product Usage is positive with a value of 0.0157 (see table 13) which means that a high Normative Susceptibility enhances the given utility of Sharing Product Usage.

Attribute Interaction Effect Utility Wald P-Value Interaction

Effect Sharing Product Usage x Normative Susceptibility 0.0157 27.1350 1.9e-7 Interaction

Effect Sharing Product Usage x Informational Susceptibility

0.0021 0.1464 0.70 Table 13: Utilities and P-values of the models including the interaction effects of the components of

(33)

26 The other tested interaction effects of Susceptibility to Interpersonal Influence, the interaction between Susceptibility to Interpersonal Influence and Reference Group Product Usage (p = 0.73 > 0.05), and the interaction effect of Susceptibility to Interpersonal Influence and Reference Group Acceptance (p = 0.81 > 0.05), showed to be insignificant (see table 14).

Effect Reference Group Product Usage x Susceptibility -0.0008 0.1158 0.73 Interaction

Effect Reference Group Acceptance x Susceptibility -0.0005 0.0552 0.81

Table 14: Utility and P-value of the model including the interaction effect of Susceptibility to Interpersonal Influence and Reference Group Product Usage, and the model including the interaction effect of Susceptibility

to Interpersonal Influence and Reference Group Acceptance

Both interaction effects are also tested spitted into the components of Susceptibility to Interpersonal Influence. The interaction between Reference Group Product Usage with Normative Susceptibility is insignificant (p = 0.16 > 0.05) while the interaction between Reference Group Product Usage with Informational Susceptibility is marginal significant (p = 0.074 < 0.1). The utility of this effect is 0.0097 (see table 15). As the utility of the interaction effects adds to the utility of the related attribute, the utility of Reference Group Product Usage is enhanced by the utility of Informational Susceptibility.

Effect Reference Group Product Usage x Normative Susceptibility

-0.0041 1.9848 0.16

Interaction

Effect Reference Group Product Usage x Informational Susceptibility

0.0097 3.1950 0.074 Table 15: Utilities and P-values of the models including the interaction effects of the components of

Susceptibility to Interpersonal Influence and Reference Group Product Usage

Both interaction effects of the components of Susceptibility to Interpersonal Influence and the Reference Group Acceptance are insignificant (see table 16).

Effect Reference Group Acceptance x Normative Susceptibility

-0.0036 1.4629 0.23

Interaction

Effect Reference Group Acceptance x Informational Susceptibility

0.0087 2.6663 0.10 Table 16:Utilities and P-values of the models including the interaction effects of the components of