ONLINE COMMUNITIES OF INTEREST:
THE KEY TO SUCCESS
A research conducted for
L.A. Staal
MSc Ba Business Development
Rijksuniversiteit Groningen
Student number: 2046342
University supervisor: F.D. Streefland
Abstract
This paper identifies the success factors regarding attracting and sustaining members for communities of interest and identifies creative ways to create return on investment for this type of communities. Constructs of the technology adoption theory are combined with the constructs from community literature to find ways to improve the attraction of new members for communities of interest. A
previously tested model for online communities in general is adjusted and tested for the business case of this study Trybes. Results of a survey to test these two proposed models showed significant
influences. Compatibility, observability and trialability are significant influencers for the intention to adopt a community of interest. Information quality and social usefulness are significant influencers for member loyalty for the case study included in this research. These findings form the fundament for a redesign. The redesign provides ways to improve the intention to adopt communities of interest and increase member loyalty for the case study. Continuously, this research identifies a design for creating return on investment. A final action plan provides steps to be taken to for the problem owner in this research to create a successful community of interest.
Key words: Communities of interest, technology adoption theory, community member needs, member loyalty, return on investment
Word of appreciation
From the end of February till half July I was writing these Master Thesis at Getlogic, a
webdevelopment company in Groningen. The time that I have spent at their office was very pleasant and gave me a lot of insights in the development of complex online products. I want to thank Edwin Tiben, CEO of Getlogic for his good supervision during this period. The brainstorming sessions we had every two weeks helped me getting new insights. Besides this I want to thank all employees of Getlogic for their support during this period.
Many gratitude goes to university supervisor Frank Streefland. To be honest I do not think I was an easy student to supervise. The meetings we had, were not always easy, but when looking back I learned a lot from his feedback and critical questions about my work.
Furthermore many thanks to Hans van der Bij. Without his good explanations and advises it would have been very difficult to finish this thesis within six months.
Last but not least, special thanks to Frederik van der Veen, my boyfriend, who always supported me during this stressful period…
Table of contents 1. Introduction ... 5 1.1 Academic interest ... 5 1.2 Business interest ... 6 1.3 Research methodology ... 7 2. Problem definition ... 7 2.1 Management question ... 8
2.2 Preliminary problem validation ... 8
2.3 Research question ... 11
2.4 Research approach ... 11
3. Analysis ... 12
3.1 Input from business setting ... 12
3.2 Theory ... 13
3.2.1 Definition ... 13
3.2.2 Community types ... 14
3.2.3 New member participation ... 15
3.2.3.1 Conceptual model intention to adopt ... 21
3.2.4 Member loyalty ... 21
3.2.4.1 Conceptual model loyalty ... 23
3.2.5 Conceptual model ... 24
4. Data collection ... 24
4.1 Survey ... 24
4.2 Measurement development ... 26
4.3 Validity and Reliability ... 26
5. Results ... 27
5.1 Results intention to adopt ... 27
5.2 Results member loyalty ... 30
6. Redesign ... 31
6.1 Setting specifications ... 32
6.2 Generation of redesign ... 33
6.2.1 Redesign new members ... 33
6.2.1.1 Compatibility ... 33
6.2.1.2 Observability ... 34
6.2.2 Redesign member loyalty ... 35
6.2.2.1 Information quality ... 35
6.2.2.2 Social usefulness ... 35
6.2.3 Return on investment ... 35
6.2.3.1 Letting members choose advertisements ... 36
6.2.3.2 Blogs ... 36 6.2.3.3 Affiliate links ... 37 6.3 Managerial implications ... 38 7. Discussion ... 39 References ... 41 Appendix ... 45
Appendix 1 Interviews entrepreneurs online communities ... 45
Appendix 2 email member support ... 49
Appendix 3 Survey ... 50
1. Introduction
The increasing popularity and the wide possibilities of Internet stimulated the development of new media in the nineties. In the early years of the 21st century, new media developed even further. Internet became more and more a holistic platform for interaction, used by the corporate environment to co-create value (Harrison & Barthel, 2009). This new phase in the development of Internet is called Web 2.0 following on the previous phase Web 1.0 (O’Reilly, 2005). Web 1.0 was characterized by being very informative, but merely one-dimensional. Information was only transferred from the company to the consumer(O’Reilly, 2005). With an increasing interest in relationship marketing and companies trying to build long-term relationships with customers, co-creating of value became of vital importance (Harrison and Barthel, 2009). Social media like online and mobile communities promises to do just that. The success of popular social communities like Facebook and the Dutch social
community Hyves made entrepreneurs realize that communities could have a huge scope. As a result different sorts of communities were popping up like brand communities and communities focused on niche markets. Loyal members are the driving force behind the success of communities (Porter, Donthu, MacElroy and Wydra, 2011; Chi, 2011).
1.1 Academic interest
Many different types of communities can be distinguished (Hummel and Lechner, 2002; Wenger, 2004) for which no specific success factors exist yet. Previous studies mainly provide success factors for online communities in general (Leimeister, Sidiras and Krcmar, 2006; Irriberri and Leroy, 2009; Toral, Martinez-Torres, Barrero and Cortes, 2008; Lin and Lee, 2006; Lin, 2008). Leimeister et al. (2006) conducted a Delphi study among 20 experts on online communities to judge previously founded success factor from other studies. Their findings provide recommendations for building and managing online communities. Many different types of communities were included, from gaming communities to customer communities, but outcomes are generalized to all these types and no distinction is made. Iriberri and Leroy (2009) conducted a Meta analysis on the success factors for designing the components of online communities. They included many different types and conducted a general life cycle as a guideline to design an online community. Iriberri and Leroy (2009) created an overview with success factors per stage in their life cycle, and indicated to which types of communities the success factor can be adapted. Toral et al. (2008) analyzed determinants of success according to the social network perspective. This means that they analyzed the influence of network cohesion, network structure and network centrality on success. Their findings show that network structure is the most important. Toral et al. (2008) used online communities related to open source software to come to these findings.
Motivations of people to join an online community are represented in several studies (Iriberri and Leroy, 2009; Jin, Park and Kim, 2010; Porter, Donthu, MacElroy and Wydra, 2011). Iriberri and Leroy give a schematic overview in how to attract members for different types of online communities including some for communities of interest. Jin et al. (2010) approach commitment in online communities from a social exchange perspective. They found that a perceived social benefit
significantly influences affective commitment. Jin et al. (2010) do not distinguish different community types in their analysis. Porter et al. (2011) found different ways to foster engagement in online brand communities. To fill in part of the academic gap, this study aims to provide factors leading to new members for online communities of interest.
To create success in attracting members for communities of interest, interesting combinations between different streams of theoretic perspectives are made in this research. Moore and Benbasat (1991) followed on the theory of Rogers (1983) to create six attributes that influence the adoption of an innovation. The technology adoption theory is combined with community literature about success factors to set hypothesis. These hypotheses include the expected factors leading to new members for communities of interest.
In the field of member loyalty in online communities, the same phenomenon exists. While member loyalty for online communities in general is represented several times in literature (DeLone and McLean, 2003, Lin and lee, 2006, Lin, 2008), member loyalty for specific types of online communities stays out. Lin and Lee (2006) searched for constructs, which influence member loyalty in online communities based on the model of DeLone and McLean (2003). They used community members and leaders from many different types of communities to measure their constructs. They found that social and technical factors influence loyalty within online communities. Lin (2008) continued on this study by adding different determinants in this category. She also used online communities in general to collect findings for instead of specific types. This research aims to provide insights in member loyalty for specific type of online communities named, communities of interest.
Another gap to fill in with this study is creating return on investment for communities of interest. While some previous studies focused on creating return on investment with help of advertising on social media and with brand communities (Porter et al., 2011, Chi, 2011) no research is done about the creation of return on investment for communities of interest. Thereforethis research also aims to provide the first step in the creation of return on investment for communities of interest.
1.2 Business interest
This Master thesis is written for Getlogic, an online web development company specialized in the development of web applications, e-commerce platforms, mobile applications and online service and hosting. The main sources of income of Getlogic are development fees based on the time and effort it takes to create final applications for clients. One of the web applications they offer to their clients is building and designing online communities. Most often, these are communities of interest. Getlogic also has their own community, named Trybes. Unfortunately, Trybes and community projects of clients of Getlogic, do not generate the success which was aimed for by the initiators. The expected numbers of community members are not reached and return on investment is not created sufficiently. Therefore the academic interest of this research perfectly matched the business problem found at Getlogic.
1.3 Research methodology
To fill in the shortcoming in the literature in the area of communities of interest, the business case of Getlogic will be solved with help of academic literature. This type of research is called academic problem solving. By reflecting on these findings it is expected that academic gaps can be filled in. Van Aken, Berends and Van der Bij (2007) designed a process for problem solving studies. This research uses some guidelines out of their book to structure the analysis. They use the regulative cycle of Van Strien (1997) to design a business problem solving study from which the first steps will be included in this research.
Figure 1. Regulative cycle (Van Strien, 1997)
This study starts with the problem mess and problem definition described in the next section. The analysis and diagnosis can be found in an extensive problem validation section, which starts with the identification of the possible causes of the problem with help of theory and community user input. The problem causes are validated with help of a quantitative analysis. Results of this analysis form a base for the redesign section in which a plan of action for Getlogic is presented to eliminate the problem causes. Getlogic can implement the action plan, which will take a certain period. This research ends after presenting the action plan to Getlogic. After the implementation period, it is recommendable for Getlogic to intervene and evaluate the implications of the changes due to the implementation of the action plan out of this research.
2. Problem definition
Van Aken et al. (2007) explain that a problem-solving project typically starts with a problem mess consisting of issues, opinions, judgments, interests and power in the organization. The problem mess must first be structured and categorized in order to define the underlying problem in the problem
definition phase. It is important that the organizations’ perceived problem is investigated to make sure that it is sufficient enough to be a real problem. In this chapter the research objective is presented that provides the basis for this research project.
2.1 Management question
In 2005 the online community platform Trybes was developed. Getlogic owns the community platform Trybes. On this platform, specific groups can create their own online community for existing offline groups such as sport teams and student clubs. The difference with communities like Facebook and Hyves is that Trybes focuses on niche groups, while Facebook and Hyves focused on a broader target market. On Trybes groups with a specific connection, like a football team, could interconnect online and facilitate their existing real life group. On voetbalteams.net, a random football team can create their own community with Trybes and for example organize trainings, sign up for a match, upload pictures, make a team set up for a match and start discussions. Other examples of communities created via Trybes are jaarclubs.net, disputen.net and studentenhuizen.net. In 2005 the former owner of Getlogic invested about 200.000 euro to build the platform and to launch it in the market. They aimed to have more than one million users after a few months and then sell the data generated from users to companies like Vodafone to create return on investment. Subsequently Vodafone could target users via the community and provide suitable marketing and sales techniques directly to these
community users. Unfortunately user numbers did not reach the expected high levels and Getlogic quickly ran out of money. Therefore, they were forced to continue with their other business and they did nothing to create return on investment. Trybes continued with a couple of thousand users.
In 2010 the current owner of Getlogic took over the company and Trybes was included in the deal. Currently Trybes only cost money for Getlogic and does not generate return on investment. Therefore the owner would like to have a plan for this community. His question is to develop a suitable strategy for Trybes to create return on investment, which can also be applied to other communities for clients of Getlogic, because they have the same problem. Table 1 gives a schematic overview of Trybes.
Table 1 Schematic overview Trybes
Trybes Community platform with for example voetbalteams.net, jaarclubs.net and disputen.net
Target group Students
Strategy to attract users
Large marketing campaign started in Groningen with flyers, advertisements and community launch during Keiweek. Purpose Create return on investment and make profit
ROI strategy Sell user data and advertisements
Problem Forecasted user numbers are not reached, user numbers stagnate currently and no return on investment
2.2 Preliminary problem validation
The owner of Getlogic named a lack of return on investment (ROI) as the main problem for Trybes and his question was to develop a strategy to generate ROI. As mentioned earlier, Getlogic also builds
communities for clients. Unfortunately, these communities are not successful as well. This is at first sight not the problem of Getlogic, they were only asked to build the community and they fulfilled this job. But when their clients do not sell their community in the market and create commercial success, no success stories about Getlogic as a community development company will be spread. The other way around, when communities build by Getlogic do generate success in the market, Getlogic can benefit from this as well, because their name is linked to this successful community. Besides this, when no community success is generated at client’s communities, no need for further expansion of the community is needed, so Getlogic looses their client. To avoid this, Getlogic wants to give advice to their clients on how to generate success for their communities. Currently they already try to do this, but they base this on gut feelings, instead of proven facts. To provide better advice in the future and co-create commercially successful communities, a closer look at the problems of the communities of their clients will be given in this section.
Communities of the clients of Getlogic are generally structured around a certain topic. The first one to include here is a healthcare community, the second one a community about fishing and the third one a community about energy. By three semi-structured interviews (see appendix 1), an overview of the objectives, strategies and problems of the communities is outlined. Below a schematic overview can be found with the most relevant findings.
Table 2 Schematic overview Wingez
Wingez Launched in February 2011, healthcare community
Target group Healthcare professionals and people who are interested in healthcare topics in the Netherlands
Strategy to attract users
With help of social media (they use Twitter to reach people), after training sessions of the NIGZ refer to content on Wingez, creating interesting topics themselves on Wingez, creating specific content which can be found easily with search engine Google
Purpose - Complementary product to NIGZ clients
- Promote trainings NIGZ
- Create return on investment and make profit ROI strategy Selling software license
Selling educational courses of NIGZ
Problem User amount stay out and return on investment uncertain Table 3 Schematic overview Topvisser
Topvisser Launched in June 2010, community for fishermen Target group Fishermen from the Netherlands
Strategy to attract users
Worth of mouth
Purpose - Bringing fishermen together
- Organize and monitor fishing events for fishermen - Create return on investment and make profit ROI strategy Selling advertisements
Problem User numbers are declining and no sales person for selling
Table 4 Schematic overview Syntropolis
Syntropolis Launched in June 2012, community for energy professionals Target group Energy professionals and people interested in energy worldwide Strategy to attract
users
Launch community on worldwide energy conference and arrange energy ‘rock stars’ to create first content on the community
Purpose - Bringing energy people from over the whole world together to discuss about energy and promote the Energy Delta Institute - Create return on investment for shareholders
ROI Sponsoring and content/advertisement auctions
Problem Forecasted user numbers stay out, no return on investment
After conducting the interviews with community holders, a deeper insight in their problems around community success was provided. It turned out that the problem is not only that return on investment cannot be created. This was only a result of the underlying problem. To go to the core of the problem we had to take one step back. All communities including Trybes experience difficulties in attracting and keeping community members, therefore the problem starts with getting community members and keeping them, which later should result later in return on investment. Therefore this research starts to solve the three most important similar problems of Trybes and the communities of clients of Getlogic, which are:
1. Not enough new members 2. Not enough loyal members 3. No return on investment
A more specified research goal could be formed after the interviews to give better answers to the business problem of Getlogic. A strategy to attract and sustain members should be formed, so that return on investment can be generated. This is needed to fulfill the main goal of this study, which is to
give an advice to Getlogic about the way to exploit communities of interest successfully in terms of new members, loyal members and return on investment.
In the desired new situation, new members and loyal members should lead to return on investment, but this goes not automatically, because members do not pay for using the community. In other words they make use of a free community, so community members do not directly create ROI. Indirectly community members can generate incomes for the community of interest (Chi, 2011). Members ‘pay’ with their participation by generating information about their interests and providing their email address (Chi, 2011). When having many members in a community with specific interest in a certain topic, the community becomes more favorable to advertise on for marketers who promote products related to these interests, because marketers are willing to promote their products or services to a specific target market only with the likelihood of sales (Chi, 2011; Porter et al., 2011). In other words, the more new and loyal members with specific interest for a certain topic, the more income can be generated from advertisement sales.
When the goals of this study are met, Getlogic can advice their clients how to exploit their
communities successfully and implement the action plan out of this study in their own community, Trybes. To reach this goal, the causes of the not satisfying user numbers should be identified and eliminated and creative ways for earning return on investment should be designed.
2.3 Research question
Resulting from the management question and the preliminary problem validation, a research question for this study can be formed.
‘’Which factors influence the success of online communities of interest with regard to attracting and sustaining members and how can these communities create return on investment?’’
In the research question a specific community type is used, communities of interest. The Theory section gives a deeper explanation about the characteristics of this specific community type.
2.4 Research approach
As mentioned earlier the goal of this research is to advice Getlogic how to exploit a community successfully in terms of members and return on investment. To reach this goal, success factors to attract and sustain users have to be identified and creative manners for return on investment for communities of interest have to be formed. This is done, by splitting this study in two parts. The first part analyzes the factors leading to new and loyal members. The second part analyzes possible ways of creating return on investment for communities of interest. During the first part, theory and a
discussion with Trybes members, former Trybes members and potential Trybes members form the base for the conceptual model. The proposed relationships will be tested in a quantitative analysis amongst students and young professionals, including Trybes members. A redesign section will outline a business advice, resulting in improved strategy for attract members for communities of interest and sustain members for Trybes. Besides this, the second part of the problem, create return on investment for communities of interest will be analyzed here. Creative solutions will be designed which are based on sufficient new members and loyal members after implementing the redesign based on the
3. Analysis
This chapter provides a literature review combined with a discussion with current, new and ex users of Trybes to identify potential causes for the problem in this research. Together this result in a conceptual model which will be tested in the validation part of this section. This will be the fundament for the next chapter: the redesign.
3.1 Input from business setting
Besides theoretical factors that could influence the problem this research identified practical factors. An email was send to Trybes members and ex-Trybes members to start a discussion about the use of Trybes and the reason for not using it anymore. This e-mail can be found in appendix 2. One person came by the office of Getlogic to discuss about Trybes, one person called and three e-mails were received. Besides Trybes members and ex-members, people who did not know Trybes were approached to have a look at the community and give their opinion.
The most important question posed to ex-Trybes members was why they stopped using it. Most answers were related to team members who also did not use it anymore. The Ex-Trybes members now use facebook groups to discuss with their team. They prefer to have not too many accounts to remember.
The most important reasons why Trybes members are still using Trybes is because in this manner they stay in touch with their team and because of the unique features. A frequent Trybes member said
‘’Our student house uses the tool to keep track on who pays what, very handy!’’. They use the special
feature of a grocery house list in Trybes. Another member said ‘’I stay in touch with my commission
members of a couple of years ago, and on Trybes we share good memories and pictures of that period’’. Another Trybes member of voetbalteams.net named that they use the feature to make an
online team set up before the match in Trybes, so that everybody in the team could have a look on that on forehand.
Three persons who do not use Trybes yet, but were asked to make an account on Trybes to see what it looks like, named also reasons why they would use Trybes and reasons why they would not use Trybes. A discussion with these trial members turned out to a couple of reasons for their intention to use/not use Trybes. They would use Trybes because it has many functions for a team on first sight. Besides this they observed that the design of the community looks professional. They named that they would not use Trybes, because many functions can also be done in Facebook, the social network of which they already have a membership, like uploading pictures and share information. The most important reasons are summarized in table 5.
Table 5 summary input from business setting
Trybes Members (3) Ex-Trybes members (2)
Trial persons (3) Why use Trybes? - I keep in touch with
my jaarclub via Trybes
- I use Trybes because of the unique features, like making a set up for the football match
- Because of the many functions it has for a group
- Because of the professional design
Why not use Trybes?
- My team stopped using it
- I use Facebook groups now with my team
- I have too many user accounts to keep track on (>5)
- Because a lot of features are overlapping with social networks such as Facebook
3.2 Theory
Next to the input from business setting, literature is taken into account related to the potential factors leading to success of communities with regard to attract and sustain users. This paragraph starts with a definition for online communities, followed by literature about attracting and sustaining community members.
3.2.1 Definition
Intuitively everyone seems to understand the concept of ‘’online community’’, but there is no agreed definition (Toral, Matinez-Torres, Barrero and Cortes, 2009). This is due to the multidisciplinary nature of this topic, which can be analyzed from different perspectives including psychology, marketing, information systems and innovation. Many authors have been analyzed the principles of online communities. For instance, some authors (Amin and Roberts, 2008; Albors et al., 2008) highlight the connection of online communities with the social learning theory and communities of practice
developed by Wenger (1998), while others are focused on their relation with knowledge sharing (Kuk, 2006), knowledge creation (Lee and Cole, 2003) and innovation models (Von Hippel and von Krogh, 2003). Another important perspective to communities is denoted to motivation of people participating in online communities (Schroder and Holze, 2010). Schroder and Holze (2010) investigated the similarities and differences between definitions and clustered the main characteristics of communities given in these definitions. The most important ones are taken into account to form a definition for this study. First a community is an online communication technology, which means that members of a community communicate via the internet. The second characteristic is virtuality, which indicated the irrelevance of physical co-location of members. Furthermore collectivity is important when it comes to
a definition for communities, because in communities people are grouped into a critical mass or social group. Next interaction plays an important role within communities. There are different types of interaction in communities, like many-to-many or relational interaction, but some kind of interaction is required. Besides this communities show relationships, like personal relationships, mutually reinforcing relationships and friendships. Within a community social exchange is also an important characteristic, because of the continuous social processes going on. Communication and exchange is the next important characteristic, because regular communication and exchange of information is important for online communities (Schroder and Holze, 2010). Summarized from above the important ingredients for a suitable definition are:
- Online communication technology - Virtuality
- Collectivity - Interaction - Social exchange - Communication
For an optimal definition for this study also the community type should be taken into account. The next section will outline various community types.
3.2.2 Community types
Wenger (2004) identified five online community types based on their reason for use: communities of place, communities of interest, communities of purpose, communities of circumstance and
communities of practice.
Communities of place were identified, as online communities in which members share a common physical location such as a neighborhood or region (Black & Hughes, 2001). They can exist totally offline, totally online or they can serve as a complement to face-to-face interactions. Online
communities of place are typically formed to complement the offline local community. Communities of interest are identified as communities in which members share a common personal interest (Wenger 2004). While it is likely that the first communities of interest are formed within communities of location, the uptake of information, new communication technologies, increased mobility and the passing of time have resulted in them becoming increasingly divergent (Blaug et al. 2006). Communities of purpose are identified as communities that focus on a particular issue, such as women’s rights or global warming. Membership is typically interdisciplinary and, rather than sharing expertise, the interactions of members are focused on achieving a specific goal (Wenger 1999). Communities of circumstance are identified as communities in which members share a position, circumstance or life experience, such as a disability or a personal loss, rather than a profession or interest (Cummings, Heeks & Huysman 2003). Membership is typically made up of individuals with a particular illness, condition or experience that connects them. Finally, communities of practice are identified as groups
of people who are enthusiastic about a particular area, or domain, and who interact to share and learn more about their area of interest (Wenger, 2004). It has been argued that communities of practice have three key characteristics. Firstly, members share a real commitment to the domain, which implies shared competence. Secondly, a desire to learn from each other underpins the interactions. As a consequence, they build relationships and engage in shared activities (Wenger 2004). Finally, group members are practitioners with a combined repertoire of resources, experiences and perspectives (Wenger 2004).
In their extensive literature review, Irriberri and Leroy (2009) also name different community types. First communities based on the need they fulfill. A community can be of interest, relationship, fantasy or transaction. At the same time, communities are classified by geographic characteristics, that is, formed by members in close proximity, or by demographic characteristics, that is, formed for or by people of specific age, gender, life style, or ethnicity, by topical characteristics, such as specific interests, hobbies or past times, or by activities such as shopping, financial investment, or gaming. Hummel and Lechner (2002) identified three genres of communities. These genres are games, interest or knowledge. Both Iriberri and Leroy (2009) and Hummel and Lechner (2002) outline that communities of interest are structured around a certain topic with or without an existent interest groups in real life. Wenger (2004) also stated that when members sharing a common personal interest can be seen as a community of interest. These community characteristics matches the different communities of Trybes which are also structured around a specific topic with member sharing a common personal interest such as sports, student clubs or school. Other communities important for the problem owner of Getlogic also focus on communities with a certain topic such as energy, healthcare or fishing.
Therefore these communities can be placed under the community type: communities of interest.
After determining the community type of this study, a suitable definition for this research can be formed in combination with the definitions and characteristics found in the previous section. The leading definition for this study therefore is: An online community of interest is a virtual communication
technology with a collective interest for a certain topic aiming for interaction, which results in relationships and social exchange.
3.2.3 New member participation
The lack of new members is the greatest obstacle to success for online communities (Ginsburg and Weisband, 2004). It is important for each community to have member participation (Porter et al., 2011). This section analyses literature to find the factors leading to successfully attracting members in online communities of interest. It combines online community literature with technology adoption literature for innovations, because this is not done before in the field of online communities and it is expected to deliver interesting results. The reason for this expectation is that online communities of interest are seen as an technical innovation in this research. Within the information technology literature success factors for general online community participation have been studied tremendously from different point of views. Using a socio-technical perspective, Preece (2001) conceptualized two
dimensions for, sociability and usability, as the determinants of community success in general. Jin, Park and Kim (2010) outline that if a firm provides the right online community environment for its members, more members will join, interact and participate in the online community. They argue that online community attributes are useful in determining effectiveness of a firm. This is another
perspective than looking for in this study, because Jin et al. (2010) assume communities as complementary to a firm.
Most studies talk about online communities in general, while as seen in the previous section, there are so many different community types. Simultaneously, the current business problem, which plays an important role in this study, is about how to attract and commit members for a specific community type like communities of interest. Therefore in this research success factors are formed for this specific type particularly, by specifying literature findings from the main category they belong to: online communities in general. These success factors are combined with theory about the adoption of
information technology innovations. In this theory potential adopters’ perceptions of innovations can be measured. The reason for using the technology adoption theory is that we want to know when new members are likely to adopt a community of interest.
Information systems researchers integrated various perspectives and created research agendas to initiate a more focused and controlled empirical study of online communities (Gupta and Kim 2004; Lee, Vogel and Limayem, 2003; Li 2004). The focus shifted to members’ needs and requirements. (Arnold, et al., 2003; Kling and Courtright, 2003; Stanoevska-Slabeva and Schmid, 2000). Porter et al. (2011) also focused on the needs and motivations of potential members for online communities. Although the community type Porter et al. (2011) focus on are brand communities, their findings might be relevant for this research, because there are some similarities. In case of brand communities a marketer attracts users and in this research companies are also the initiators of communities instead of members who set up a community. In other words, both types need a commercial strategy to reach potential members. In case of brand communities, members are seeking for their needs to fulfill around their interest for a certain brand. In case of communities of interest members are seeking to fulfill their needs around their interest for a certain topic. In both cases people with shared interest come together in the community. Concluding from this there is a shared goal for
marketers/entrepreneurs of communities of interest, namely attracting members and for members, namely topic or brand related information. Therefore findings of Porter et al. (2011) are used here to form a starting point in this study to understand the motivations for community participation better and anticipate on this more to attract members. The following model shows a schematic overview of the findings of Porter et al. (2011).
Figure 3 Various needs that members fulfill via virtual communities (Porter et al., 2011) Stage 1 is of relevance for this phase of research, because factors leading to attracting users for communities are required here. Therefore looking at the needs and motivations for potential members to fulfill in a community might helpful. Stage 2 and 3 might be of relevance in the second part of this research, dependent on the outcomes of the problem validation. In stage 1 Porter et al. (2011) investigated how firms foster effective engagement in online communities. They start with the needs that members of communities fulfill via online communities. The most relevant needs for this study are outlined below.
• Enjoyment à online community members are gratified by achieving flow states while interacting with others by having control over their experience with a community.
• Belongingness à online community members desire a sense of attachment to a community, as a whole, and are gratified by having their contributions to the community respected by others.
• Status/influence à online community members seek status and influence among others within a community.
To make the right decision about the success factors to include in this research for communities of interest, the findings of Porter et al. (2011) will be compared with finding of other researchers. Iriberri and Leroy (2009) identified in their meta-analysis a life cycle for communities for which existent literature strengthens the cycle with success factors. In this cycle they also analyzed factors of growth leading to attracting members for communities of interest. Findings of Ginsburg and Wiesband (2004) out of this analysis show that there are three main factors to attract community members for
• Existence of an offline customer club as starting advantage • Real life status symbols
• Sending numeric goals for contribution
To continue, the second literature will be introduced now, the technology adoption theory. Rogers (1983) first identified the following five general attributes that influence the adoption of an innovation.
Relative Advantage: the degree to which an innovation is perceived as being better than its precursor Compatibility: the degree to which an innovation is perceived as being consistent with the existing values, needs, and past experiences of potential adopters
Complexity: the degree to which an innovation is perceived as being difficult to use Observability: the degree to which the results of an innovation are observable to others Trialability: the degree to which an innovation may be experimented with before adoption
Moore and Benbasat (1991) added the attribute ‘’image’’ to these list. Image is defined as ‘’the
degree to which use of an innovation is perceived to enhance one’s image or status in one’s system’’. Preece (2001) argued that sociability is about social contact and self esteem that people are searching for in online communities. Besides this Porter et al. (2011) argues that status and influence are
motivational needs for potential community members to join. He outlines that online community members seek status and influence among others within a community. Findings of Ginsburg and Wiesband (2004) showed that status symbols are a reason to attract members in a community of interest. Status symbols can be also placed under ‘image’, which strengthens the expectation that image might have a positive influence on the intention to adopt a community of interest. Therefore the following hypothesis is created:
H1: Image positively influences the intention to adopt a community of interest
The adoption attribute compatibility is in line with the success factor that Iriberri and Leroy (2009) name as the existence of an offline customer club as starting base. This is because the community of interest complements in this way an existing group and subsequently creates the need to join for other members of the team. Trybes members and ex-Trybes members also name their offline team also as an important reason for joining the online community of interest. Based on this compatibility with existing valeues is expected to influence the intention to adopt a community of interest in a positive way.
H2: Compatibility positively influences the intention to adopt a community of interest
Relative advantage is the degree to which the community of interest will be perceived as better than its precursor or competitive communities. Social media, which have the goal of creating and
maintaining social relationships or friendship, can be seen as the precursor of online communities of interest (Irriberri and Leroy, 2009). As seen in the discussion with Trybes members, unique features of communities of interest could make a difference when talking about relative advantage. In this light we mean a difference compared to social media, the precursor of online communities of interest.
Therefore relative advantage over social media might be important for communities of interest to attract new members. To test whether relative advantage over a precursor results in the intention to adopt a community of interest the following hypothesis is designed.
H3: Relative advantage positively influences the intention to adopt a community of interest
Complexity is the degree to which the community of interest is perceived as being difficult to use. Ex-Trybes members also indicate that they dislike the amount of user accounts they have and therefore are inclined to stop using their account for communities of interest. Remembering many passwords can be perceived as being difficult and as a result members can perceive the community of interest as being difficult to use. Ginsburg and Weisband (2004) argue that user friendliness leads to the
likelihood that more members will join the community. Concluding from this, making it complex to use and log into the community of interest could scare of new members. Therefore it is expected that perceived complexity on forehand leads to less members for communities of interest.
H4: Complexity negatively influences the intention to adopt communities of interest
Observability is the degree to which the results and usefulness of a community of interest are observable to others (Rogers, 1983). In this case when members are aware of the usefulness of a community of interest on forehand, they might be more likely to adopt the community. As shown in the member attraction success factors of Ginsburg and Wiesband (2004), sending numeric reasons to join will positively influence the attraction of new members. By doing this potential members become more aware of the possibilities of the community of interest and can in this way better ‘observe’ the
usefulness of the community. As said before, according to the adoption theory observability positively influences the intention to adopt a new information system. To test whether this is also the case for communities of interest, the following hypothesis is created.
H5: Observability positively influences the intention to adopt a community of interest
Trialability is the degree to which the community of interest can be experimented with before adoption. In some communities of interest included in this research, non-members can see and use a lot of features of the community without making an account, while on others, people first have to make an account to see activity. One of Porter’s (2011) member needs was enjoyment. If potential members enjoy a trial version of a new community of interest, the chance to subscribe for the community might increase. Another scenario for experimentation is testing a community via a friend or relative. To go back to the community needs of Preece (2001), sociability is important. Friends or relatives of potential
members who already use the community of interest are able to letting others try the community. The advantage of this scenario is the existent social connection, which might positively influence the intention to adopt. Concluding from above, trialability might influence the intention to adopt a community of interest positively. Therefore the following hypothesis is designed.
H6: Trialability positively influences the intention to adopt a community of interest
Currently, developers of social networking sites refocus their efforts on promoting vertical social networking sites for members with similar personal interests within their social network (Schroder and Holze (2010). Facebook introduced Facebook Groups, which is currently the most well-known
example of a vertical social networking site. These vertical social networking sites behave in the same way as traditional communities of interest in which for example pet lovers commune (Irriberry and Gondy, 2009). As shown in the previous paragraphs Facebook Groups is indicated, by both, the literature and ex-Trybes users as threat for Trybes. Since Trybes belongs to the category of
communities of interest, this might count as well for communities of interest in general. Therefore it is expected that the use of Facebook Groups has a significant influence on the relationships between the above mentioned factors and the intention to adopt communities of interest. This moderator will be tested as well in the analysis.
H7a: Whether someone uses Facebook Groups has a negative influence on the relationship between image and the intention to adopt a community of interest
H7b: Whether someone uses Facebook Groups has a negative influence on the relationship between compatibility and the intention to adopt a community of interest
H7c: Whether someone uses Facebook Groups has a negative influence on the relationship between relative advantage and the intention to adopt a community of interest
H7d: Whether someone uses Facebook Groups has a negative influence on the relationship between complexity and the intention to adopt a community of interest
H7e: Whether someone uses Facebook Groups has a negative influence on the relationship between observability and the intention to adopt a community of interest
H7f: Whether someone uses Facebook Groups has a negative influence on the relationship between trialability and the intention to adopt a community of interest
3.2.3.1 Conceptual model intention to adopt
To visualize the above, formed hypotheses figure 4 is formed. The 6 hypothesizedconstructs might influence the intention to adopt a community of interest. Facebook Groups is the moderator influencing the relations between the 6 constructs and the intention to adopt.
Figure 4 conceptual model intention to adopt
3.2.4 Member loyalty
Sustaining community members remains a challenge (Porter, 2011). Loyalty creates a stable pool of members and thus it plays a significant role in expanding the community (Lin, 2008). To learn more about how loyalty arises, literature about the creation of loyal members will be included here.
Preece’s (2001) principals, sociability and usability also have an influence on loyalty in communities. Sociability refers here to how well an online community supports human–human interaction. Usability is primarily concerned with how much an online community facilitates people’s interaction with technology (human–computer interface). In another perspective of online community research, psychology researchers focused on members’ relationships and attachments within online communities. Blanchard (2004) and Blanchard and Markus (2004) studied sense of community including feelings of belonging, safety, and attachment to the group. When these feelings are present, members develop lasting relationships with other members feel attachment to the community,
and perceive the online community as a source of social and emotional support. Basically, all factors can be divided in two main categories, social and technical.
Building on the work of DeLone and McLean (1992), Lin (2008) identified four fundamental factors leading to customer loyalty. The model integrates the influences of both technical and social factors on the success of online communities. Lin (2008) names four main factors influencing member loyalty.
1. System quality 2. Information quality
3. Trust
4. Social usefulness
In fact the model of Lin (2008) measures member loyalty for online communities in general. This part of the research aims to find the success factors in terms of member loyalty in case of Trybes, which is a community of interest. Because of the inclusion of both technical and social factors in Lin’s (2008) constructs, these factors will be used as a baseline, but adjusted to the Trybes case. Besides this items with no significance in the study of Lin (2008) will be reconsidered to include here.
Although, Lin (2008) supposed that system quality, which measures technical success, and
information quality, were leading to member satisfaction and continuously to member loyalty, she did not found a really strong relationship. Other studies (Toral, Martinez-Torres, Barrero and Cortes, 2010; Iriberry and Leroy, 2009) found that system quality directly influences commitment to a community. Hew (2009) strengthens the expectation about system quality, he names community technology as an indicator for loyalty in online communities. Therefore a direct relationship is expected between system quality and member loyalty. In case of Trybes the technique to build an online community of interest is very important. When having a good functioning system, members might be more likely to be loyal to the community. To test whether system quality indeed has a positive influence on member loyalty for Trybes, the following hypothesis is created.
H8: System quality positively influences member loyalty for Trybes
Iriberry and Leroy (2009) also denoted information quality as a success factor for online community growth might be of high relevance. If information quality is high, a complete and structured overview of information is provided in a community (Lin, 2008). Porter et al. (2011) strengthen findings of Iriberry and Leroy (2009) by outlining that online community members find value in a community that provides access to qualitative information that helps them learn, solve problems, and make decisions. Other studies also argue that information quality positively relates to the growth of an online community (Sangwan, 2005; Leimeister and Krcmar, 2003; Zhang and Hiltz, 2003). This assigns the relevance of testing a direct relationship to expect between information quality and member loyalty. To test this for Trybes, the following hypothesis is designed.
H9: Information quality positively influences member loyalty for Trybes
Trust is leading via sense of belonging to member loyalty according to Lin (2008). In case of Trybes, members are already connected offline and already have some kind of belongingness together. Besides this many authors indicate trust as direct influencer for member loyalty (Andrews, 2001; Leimeister et al., 2005; Iriberry and Leroy, 2009;). Therefore it can be argued that there might be a direct relation between trust and member loyalty. Besides Lin (2008), Iriberry and Leroy (2009) and Leimeister et al. (2005) also denote trust as success factor for growth in online communities. Building
trust amongst members positively influences the growth of online communities in general (Iriberri and Leroy, 2009) and therefore is worth it to test this for more niche communities like Trybes. Therefore the following hypotheses is designed.
H10: Trust positively influences member loyalty for Trybes members
Social usefulness refers to the perceived support in online communities in terms of receiving respect, recognition and approval from other members (Lin, 2008). In Lin’s (2008) model, social usefulness influences member loyalty via sense of belonging. Because later studies (Porter et al., 2011; Jin, Park and Kim, 2010) found evidence for a direct connection between social usefulness and member loyalty, this will be tested as well for communities of interest. Porter et al. (2011) outline that online community members want to achieve self-awareness and are gratified by the emotional and cognitive connection with the community, as a whole, as well as their ability to express such connection. Porter et al. (2011) also states that online community members are willing to help others within a community, especially those with whom they have developed a personal connection. In case of Trybes communities are build with existing offline teams, which already assumes a personal connection. It is likely that these groups want to extent their relationship building in their community. Besides this Jin, Park and Kim (2010) found a strong (.65) relationship between perceived social benefit and affective commitment in communities, which also creates more fundament for the expectation that social usefulness might directly influence member loyalty within Trybes. Therefore the following hypothesis is designed.
H11: Social usefulness positively influences member loyalty for Trybes 3.2.4.1 Conceptual model loyalty
The above hypotheses are shown in the conceptual model below. The four constructs supported by literature are expected to have all a positive influence on member loyalty for Trybes.
3.2.5 Conceptual model
Summarized from the problem definition and the theory analysis the conceptual model for this research can be visualized as shown in figure 6. To continue, intention to adopt and member loyalty will be tested in a quantitative analysis (two black boxes). After this a redesign is made to increase the intention to adopt and member loyalty. The redesign section also outlines creative solutions to create return on investment with new members and loyal members (red box).
Figure 6 Conceptual model
4. Data collection
This section will outline how the data for this study is collected. First the method will be explained, followed by an explanation of the sample. Continuously, the preconditions for measurement development are outlined and finally the reliability and validity of this research is explained.
4.1 Survey
To find the significant constructs for the intention to adopt, a quantitative research is conducted amongst students and young professionals. The reason for this is that this group (18-28 years old) is the first generation grown up with social media and several types of online communities, and thus is
communities of interest included in this research. Another reason for using this group can be found in literature, several studies argue that students and young professionals resemble the normal
population of Internet users (Hackbarth and Grover, 2003; McKnight, Choudhury and Kacmar, 2002). To test if the constructs have a significant influence on the intention to adopt a community of interest, data was collected using an online survey. 298 students and young professionals filled in the survey completely.
To test loyalty amongst users of Trybes, Trybes members were included. 108 Trybes members completely filled in an online survey to test if the constructs influence member loyalty significantly.
These Trybes members also filled in the questions about the intention to adopt communities of interest, so these 108 Trybes members are part of the 298 respondents of the adoption part. To avoid bias, Trybes was not mentioned in the invitation for the survey and Trybes questions were posed in the second part of the survey. In other words, Trybes members could not know that this research was related to Trybes. This means that Trybes members first filled in the ‘’intention to adopt questions’’ for communities of interest in general and afterwards, if they filled in that they were Trybes member, the loyalty questions. Members who were not Trybes member could not see questions about Trybes.
To ask attention for the survey, an email was send to large group of students and young professionals. About 7000 people received the e-mail. In this database also Trybes members were included. A cover email explained the purpose of this study (without mentioning Trybes) and promised to ensure that the responses were kept confidential. When clicking on the survey link, a description of online
communities of interest was given with some examples of possible communities of interest. The survey can be found in appendix 3. Table 6 and 7 shows the demographic characteristics gender, age and education level of the respondents. The control variables are tested in the regression analysis, but show no significant influence on the results.
Table 6 demographics gender and age
# Answer Response %
1 Man 115 39%
2 Vrouw 183 61%
Total 298 100%
Table 7 demographic education level # Answer Response % 1 VMBO 3 1% 2 HAVO 4 1% 3 VWO 31 10% 4 MBO 12 4% 5 HBO 45 15% 6 WO 199 67% 7 Anders 4 1% Total 298 100% 4.2 Measurement development
All independent variables were measured using multiple items, which were gathered in the survey using a seven-point liker type scale (rating from 1=strongly disagree to 7=strongly agree). The constructs were primarily adapted from previous studies and modified to collect the data about communities of interest and Trybes. Appendix 4 lists all scales used to come to each question.
The dependent variable of the first model, intention to adopt, is measured by asking how likely the chance is that they will adopt a community of interest in the coming year. Answer possibilities were in percentages, where 0% is not likely at all and 100% is very likely.
The dependent variable of the second model, member loyalty is also gathered using the seven-point liker type scale (see appendix 4).
4.3 Validity and Reliability
When the output of the survey was generated, the reliability and the validity of the constructs of both models are tested. Therefore the items of both models are put into a factor analysis. The rotation method of the factor analysis is performed with the Varimax with Kaiser Normalization for both the adoption study and the loyalty part. For the adoption items the rotation is converged in 6 iterations, for the loyalty items the rotation is converged in 5 iterations.
After the factor analysis a Cronbach’s Alpha analysis is performed for each scale separately to find the degree to which items measure the same underlying construct (Blumberg, Cooper, and Schindler, 2005). To be reliable, the Cronbach’s Alpha should exceed 0,7 (Nunnally, 1978). If items were below this score they should be deleted. Sometimes a score between 0,6 and 0,7 can be accepted since this value is still acceptable for exploratory research or if the construct does not have many items (Hair, Black, Babin, Anderson and Tatham, 2006). A high score of the other constructs should compensate
between the scales within both the adoption and loyalty study. Pearson is chosen because we use continuous scales here, which means the answer possibilities are one number without decimals (1-7). A two-tailed test is done because not all relations are expected to be positive. The correlation
coefficients from this test indicate if multicollinearity exists between the variables. If this is the case, two different scales are too much the same, which negatively influences further analysis. To avoid multicolinearity, all correlation coefficients should be below 0,750 (Klijn. 2007). After this a multiple regression analyses is performed. This analysis tests if the hypotheses can be accepted or should be rejected with help of the Beta value and variance inflation factors (VIF). VIF should be below 10,0 (Blumberg et al., 2005).
5. Results
In this section the outcomes of the factor analyses for the constructs of both models are discussed. Continuously, the Cronbach Alpha and Correlation coefficients are outlined. Finally the outcomes of the multiple regression analysis are presented.
5.1 Results intention to adopt
From the factor analyses of the adoption model it seems that some questions of different constructs seems to overlap. Therefore some had to be adjusted to generate valid scales to use for the multiple regression analysis. From the questions, which measure the image, image 4 is removed from the analysis, because this question generated overlap with other constructs and each question can only measure one construct. Furthermore compatibility 3 has been deleted to make this scale valid for this construct for the same reasons. Complexity also showed some overlap with other scales and to make this scale valid complexity 4 has been deleted. Subsequently, for observability, items 3 and 4 are deleted to create a valid scale. Relative advantage also had overlapping. To make this valid, relative advantage 2 and 4 are removed from further analysis. The trialability scale reduced after the factor analysis to 1 item, which makes this scale not very strong. The final rotation of the factor analysis is displayed in table 8.
Tabel 8 Factor analysis constructs intention to adopt communities of interest Raw Rescaled Component Component 1 2 3 4 5 6 1 2 3 4 5 6 Image 2 1,159 ,898 Image 3 1,130 ,886 Image 1 1,143 ,858 Compatibility 2 1,324 ,872 Compatibility 1 1,249 ,842 Compatibility 4 1,168 ,737 Complexity 3 ,896 ,824 Complexity 2 ,870 ,815 Complexity 1 ,809 ,712 Observability 1 1,353 ,835 Observability 2 1,238 ,759 Relative Adv 1 1,437 ,927 Relative Adv 3 ,771 ,608 Trialability 3 1,608 ,954
From table 8 can be seen that all items have a high score now with their factor (,608-,954). This is a valid score, because it should be above ,400 (Klijn, 2007). Before the actual scales are created the reliability of the potential items of the scales are tested with Cronbach Alpha. The outcomes of the Cronbach Alpha are in table 8. The items of relative advantage had a low score (<,600). Therefore it has been decided to remove relative advantage 3 from the analysis.
To continue the correlation between variables is tested to see how the different variables correlate together. Table 9 shows correlation coefficients together with the Cronbach Alpha (diagonal bold numbers). The correlation coefficients vary between ,071 and ,604. This is all below ,750, which indicated that no multicollinearity exists.
Table 9 Correlation Coefficients and Cronbach Alpha Adoption model
Mean Std 1 2 3 4 5 6 7 1 Intention to adopt 33,8389 26,6703 α - 2 Image 2,6376 1,17521 ,267 α ,888 3 Compatibility 3,9508 1,36543 ,604 ,359 α ,873 4 Complexity 5.4966 ,94191 ,367 ,050 ,441 α ,769 5 Observability 4,2500 1,51605 ,503 ,212 ,593 ,562 α ,850 6 Relative adv. 4,21 1,550 ,232 ,210 ,336 ,230 ,238 α - 7 Trialability 3,39 1,686 ,338 ,259 ,319 ,218 ,337 ,071 α -
The first objective of this statistical analysis was to find the variables, which significantly influence the intention to adopt a community of interest. This is done by hypothesis testing through a multiple regression analysis. Table 10 shows the outcomes of the multiple regression analysis.
Table 10 Multiple Regression Analysis Adoption model
Non Std B Sig Std Beta VIF
Image Compatibility Complexity Observability Relative advantage Trialability 1,218 8,087 1,575 2,991 0,332 1,909 0,280 0,000* 0,319 0,007* 0,690 0,014 0,054 0,414 0,056 0,170 0,019 0,121 1,226 1,846 1,549 1,923 1,160 1,198 R2 0,416 Adjusted R2 0,404 F 34,538
To start analyzing the model, the R2 and the adjusted R2 are introduced. When these values are high, for example ,900, multicollinearity exist between the variables (Klijn, 2007). In this analysis, these values are relatively low (,416 and ,404) which forms a good starting point for further regression analysis. Besides this VIF is largly below 10,0 which makes the results of the regression highly interpretable. Because of the sample of 298, which is relatively low for this type of research, a high significance level is set. Only hypotheses with a significance of 95% are accepted. This means that the significance of the regression analysis should be below ,050 to be accepted. As shown in table 10, compatibility, observability and trialability have significance values of 0,000, 0,007 and 0,014
respectively. Therefore H2, H4 and H6 can be accepted and H1, H3 and H5 should be rejected. The standardized beta (std beta) shows the influence the independent variables have on the dependent variable. The std beta of compatibility is 0,414 and the std beta of observability is 0,170 and the std beta of trialability is 0,121, which means that compatibility has a higher influence on the intention to adopt communities of interest than observability and trialability.
It was expected that Facebook Groups (FG) should influence the relationship between the
independent and dependent variables in the adoption model. Therefore a multiple regression analysis with the moderator FG is done. In this analysis the independent variables and moderator are reduced by their means. Continuously, the independent variables are multiplied by FG. Expected was that FG negatively influences the relationships between the independent and dependent variable, but the outcomes presented in Table 11 do not confirm this. Therefore H7a-f are rejected.
Table 11 Mean and Moderator Multiple Regression adoption model
Non Std B Sig Std Beta VIF
IMAGE-MEAN COMPATIBILITY-MEAN COMPLEXITY-MEAN OBSERVABILITY-MEAN RELATIVE ADVANTAGE-MEAN TRIALABILITY-MEAN
FACEBOOK GROUPS (FG)-MEAN FG*IMAGE FG*COMPATIBILITY FG*COMPLEXITY FG*OBSERVABILITY FG*RELATIVEADVANTAGE FG*TRIALABILITY 1,184 8,067 1,367 3,109 0,480 1,839 0,893 -0,242 0,634 0,651 1,337 -1,433 -0,251 0,309 0,000* 0,426 0,006* 0,604 0,021 0,751 0,925 0,816 0,851 0,628 0,504 0,895 0,52 0,413 0,048 0,177 0,024 0,116 0,015 -0,005 0,015 0,012 0,034 -0,031 -0,007 1,282 1,846 1,795 2,001 1,031 1,236 1,061 1,291 2,071 1,989 2,388 1,060 1,375 R2 0,419 Adjusted R2 0,393 F 20,78422
5.2 Results member loyalty
From the factor analyses of member loyalty model it also seems that some questions of different constructs seem to overlap. Therefore some had to be adjusted to generate valid scales to use for the multiple regression analysis. From the questions, which measure trust, trust 1 is removed from the analysis, because this question generated overlap with other constructs and each question can measure one construct. Furthermore social usefulness 3 has been deleted to make this scale valid for this construct for the same reasons. Information quality also showed some overlap with other scales and to make this scale valid information quality 2,3 and 5 has been deleted. Subsequently, for system quality, items 1, 2 and 4 are deleted to create a valid scale. This scale reduced after the factor
analysis to 1 item, which makes this scale not very strong. The final rotation of the factor analysis is displayed in table 12.
Tabel 12 Factor analysis constructs member loyalty Trybes
Raw Rescaled Component Component 1 2 3 4 1 2 3 4 Trust 2 1,446 ,930 Trust 3 1,251 ,896 Social Usefulness 1 1,253 ,836 Social Usefulness 2 1,422 ,878 System Quality 3 1,093 ,834 Social Usefulness 3 1,165 ,634 Information Quality 4 1,237 ,821 Information Quality 1 1,320 ,871
From table 12 can be seen that all items have a high score now with their factor (,634-,930). This is a valid score, because it should be above ,400 (Klijn, 2007). Before the actual scales are created the
reliability of the potential items of the scales are tested with Cronbach Alpha. The outcomes of the Cronbach alpha are in table 13.
To continue the correlation between variables is tested to see how the different variables correlate together. Table 13 also shows correlation coefficients. The correlation coefficients vary between -,026 and ,631. This is all below ,750, which indicated that no multicollinearity exists.
Table 13 Correlation Coefficients and Cronbach Alpha Loyalty model
Mean Std 1 2 3 4 5 1 Member Loyalty 6,7664 2,96387 α ,612 2 Information Quality 7,0278 2,62076 ,500 α ,680 3 Social usefulness 9,2430 4,05351 ,631 ,383 α ,747 4 Trust 9,4393 2,73405 ,307 ,182 ,269 α ,831 5 System quality 4,02 1,304 ,021 ,093 ,093 -,026 α -
In the regression analysis (see table 14), the values R2 and adjusted R2 are acceptable (,493 and ,473 raspectively) which forms a good starting point for further regression analysis. The VIF scores are all largly below 10,0 which is a good fundament for the regression. The significance should be 95% for the hypotheses to be accepted. This means that the significance of the regression analysis should be below ,010. As shown in table 14, information quality and social usefulness have significance values of 0,000. Therefore H9 and H10 can be accepted and H8 and H11 should be rejected. The std beta of information quality is 0,295 and the std beta of trust is 0,120, which means that information quality has a higher influence on member loyalty for Trybes than trust.
Table 14 Multiple Regression Analysis Loyalty mode
Non Std B Sig Std Beta VIF
Information quality System Quality Trust Social usefulness 0,332 -0,112 0,131 0,359 0,000* 0,488 0,105 0,000* 0,295 -0,049 0,120 0,491 1,187 1,016 1,090 1,238 R2 0,493 Adjusted R2 0,473 F 24,813 6. Redesign
To achieve the desired outcome of this research several redesign steps are needed. First, the factors leading to the intention to adopt and loyal members followed from the quantitative analysis will be translated into ideas to attract and sustain more members. Second, ways to create return on investment with new members and loyal members will be introduced here with help of theory. The intention to adopt does not mean automatically new members. Therefore potential members first have to use the community of interest (see figure 6.1). By improving the factors leading to the intention to adopt the likelihood that potential members will adopt the community of interest will increase, which
