A conceptual model to design partially virtual communities

Proceedings of the 2012 IEEE 16th International Conference on Computer Supported Cooperative Work in Design

A Conceptual Model to Design Partially Virtual Communities Francisco Gutierrez, Nelson Baloian, Sergio F. Ochoa

Gustavo Zurita Control Management and Information Systems Department Universidad de Chile Santiago, Chile [email protected]

Computer Science Department Universidad de Chile Santiago, Chile {frgutier, nbaloian, sochoa}@dcc.uchile.cl

Typically, these communities are closed, small (less than 100 people) and have one or more community managers. Their Web supporting platform is not the center of the community and the level of security required for the interactions is low, since members know each other. In PVCs the social computing platform helps trigger face-to-face interactions among participants, which is highly positive for the psychological health of the participants [8, 13] and also to improve the trust and commitment among community members [25]. Since a PVC involves partially virtual participants, it inherits several features of a virtual community, as for example, its lifecycle. Studies of member participation in virtual communities indicate that users tend to leave a community when a certain number of conditions are satisfied, leading finally to the end of its lifecycle [17, 12]. Therefore, the need of captivating and keeping active users arises as a problem that can be critical in the community lifecycle. Moreover, users also pass through a specific lifecycle timeline [14, 23] that dynamically affects participation within a community. An understanding of the factors related to ownership and member participation within a virtual community will assist organizations to extract the full potential from them [11]. Furthermore, communities also need to understand the competitive, regulatory and strategic implications of positioning themselves in a certain category, as then they can develop an understanding of the forces that influence and operate in that environment [19]. In this paper we seek to understand and identify relevant factors in order to improve the problem of designing successful partially virtual communities. By successful we mean a community that is healthy, growing and sustainable through time. We define strategies that aim to keep, relatively stable, the gap between users entering and leaving the community at a certain time period. In fact, in online communities where users do not know each other, the design of the technological platform that supports the community may directly impact how it is designed. Therefore, with this study, we aim to define a conceptual model that will allow designers to establish a frame for developing successful partially virtual communities, as well as guidelines for community managers in order to help them maintain and evolve communities through time. Section 2 presents a literature review of the research related to online communities and member lifecycle models,

Abstract— Today, millions of users around the world log into Web platforms to participate in online communities. Some of these communities group people that interact frequently in a face-to-face manner, such as the community of a course or a neighborhood. These communities are partially virtual because their members interact through both, a physical and a virtual space. These partially virtual communities have particular features and needs, and to the best of our knowledge, there are no guidelines concerning their modeling and evolution. This paper addresses the challenge of supporting the design of partially virtual communities. In particular, the article proposes a conceptual model to help design these communities, based on a literature review, a previous field study and the authors’ experience. This model is expected to help designers and community managers establish the basis for developing and maintaining sustainable communities in order to ensure a certain level of activity through time. Keywords: partially virtual communities, community design, participation, motivation, community management.

I.

INTRODUCTION

Millions of people participate on a daily basis in one or more of the many kinds of virtual communities that exist within the Web. These social platforms are dedicated to gathering people together within a specific context, such as business Websites, Web 2.0 educational tools, professional social networks, and social gaming applications, among many others [4, 6]. In fact, this is an increasingly complex target where people all around the world gather and interact together. As an example, Facebook, the largest social network today in the world, consists of more than 800 million active users, with more than 50% of them logging on in any given day [1]. The social platform of Facebook, YouTube and Wikipedia are among the top 10 most-visited sites on the Web [2]. Many taxonomies have been proposed to classify these communities [11, 19]. All of them consider just the interactions the participants have through the Web supporting platform. However there exist some communities where the participants know each other and have the chance to interact frequently through a virtual and a real scenario. We have called these communities partially virtual communities (PVC). Examples of PVC are the community of a school/university course, of an office or even a small neighborhood.

978-1-4673-1210-3/12/$31.00 ©2012 IEEE

685

as well as factors that might impact the success of a community. It also discusses the problem of motivation and participation in these communities. Section 3 presents the model for designing successful PVC, and it also describes its main components. Section 4 shows the results of an initial case study (a proof of concept) we carried out to test the validity and pertinence of the components considered in the proposed model. Section 5 presents the conclusions and future work.

Brandtzaeg and Heim state in [3] that there are several reasons why community users decrease or stop their participation over time, e.g. lack of interesting people, low quality content, low usability, harassment, or timeconsuming community, among others. This gives us an idea of how community managers should lead the community, as well as design principles to keep in mind when planning to build a community. Finally, it is worth pointing out the case of lurkers. This kind of users may or may not be considered a problem, depending on the perspective from which this behavior is being judged and the goals of those making the judgment [22]. Lurkers are users that typically consume useful information from the community and also ask questions of its members, but do not support the needs of others. Various studies on online communities indicate that lurkers represent 80–90% of a virtual community population [6, 18].

II. RELATED WORK Various aspects of online communities have been studied from several perspectives: computer science, psychology, and sociology [6]. All of them provide interesting insight on the designing of successful communities. The following sections address the related work from a computer science point of view, but considering the psychological and social aspects involved in this kind of product.

C. Design Principles When designing a social Website, one of the first issues to take into consideration is the usage lifecycle. Porter presents in [20] a model that consists of a set of stages people go through when using software: unaware, interested, first-time use, regular use, and passionate use. This can be linked to the idea of passing over hurdles such as awareness, sign-up, return visits and emotional attachment. Preece and Shneiderman present in [23] a set of guidelines for designing robust virtual communities. They state that users are shy at first and do not interact in an appropriate way with the platform. The longer the users are engaged, they go through a natural evolution process, from reader to contributor, to collaborator and finally to leader. When taking this into account, interfaces should include well-thought-out usability and sociability features, such as adapting to the general context of the community, and easy access to relevant content through navigation or search. Kim presents in [14] nine design strategies that sustain the basis of successful communities, such as defining a purpose, building extensible gathering places, or creating meaningful and evolving member profiles. This can be considered as a whole as designing for growth and change, creating and maintaining feedback loops and empowering members through time.

A. Virtual Community Lifecycle Understanding the lifecycle of a community and identifying the needs of both users and managers in each stage is crucial for encouraging user participation and designing and maintaining successful virtual communities through time. Mousavidin and Goel present a conceptual model that depicts the general elements related to the life of a virtual community [17]. They state that the community lifecycle is influenced by four different factors: socially shaped aspects (such as critical mass), individually demonstrated characteristics (and how they are related to participation), technologically facilitated features and external influences. Iriberri and Leroy state in [12] that online communities evolve in stages, and each one presents distinct characteristics and needs: inception, creation, growth, maturity and death. Therefore, different social and technology features are needed depending on the development stage of the community. Understanding this aspect of a community helps community managers to detect needs and intervene in the community when it is required. B. Success Factors and Metrics When trying to understand the success factors and related metrics within virtual communities, we fall into an integrated viewpoint considering both social and technical features. In [12] the authors state that a community will be successful when its users develop an identity towards it, as well as develop a well-worked-out technical platform matching its users’ needs, considering both sociability and usability. This can be measured by tracking the size, level of participation and number of contributions within the community at a specific time period. Lin claims in [16] that both member satisfaction and a sense of belonging are determinants of member loyalty within a community. Linked to this same idea, Preece states in [21] that sociability and usability are key issues within any kind of virtual community. Sociability is concerned with developing software, policies and practices to support social interaction online. On the other hand, usability is concerned with how intuitive and easy is for individuals to use and interact with a product.

D. Motivation and Participation in Virtual Communities One of the key issues when designing successful virtual communities is to define a strong motivation and participation strategy. In fact, participation (in terms of quantity and quality) is often considered as one of the main metrics when analyzing the success of a virtual community. Some of the main motivational factors for encouraging participation are [20]: identity, uniqueness, reciprocity, reputation, sense of efficacy, control, ownership and attachment to a group and fun. Kohl et al. state in [15] that leaders of robust, sustainable virtual communities, find ways to strengthen their members’ sense of social identity and motivate their participation in the community activities. The drivers that stimulate communities are: the leaders’ involvement, the level of offline interaction

686

and the perceived usefulness, while the challenges remain in the communication, motivation and leadership levels. Recognition counts among the numerous factors that contribute to the success of virtual communities [5]. There are three different forms of perceived recognition: identity, expertise and tangible recognition. Tedjamulia et al. present in [24] a model for motivating content contributions to online communities. They state that feedback is the process of measuring performance, comparing it to a standard, and reinforcing action through incentives over a period of time. These incentives may either intrinsically or extrinsically motivate the community. A different approach of tackling the motivation and participation issues is carried out in [7] and [10]. The first one states that motivations for participating vary according to the personality type of users, where different attitudes towards the community affect the way users interact with the platform and with others. The second one claims that individuals’ needs largely determine their participation and intention. Moreover, identification and their perception of trust also influence their behaviors to a certain extent. The authors in [9] propose a model for triggering participation in virtual communities, considering intrinsic motivation and highlighting both the value of quality and quantity of contributions. The approach carried out consists of providing feedback to community members in different ways, in order to allow them to rank themselves within the community and enable interaction with others. III.

In fact, external factors (context, target and privacy and legal issues) lay within the whole process of design and development, since they characterize the community and its surrounding environment. Internal factors (governance, participation, platform development and feedback loops), on the other hand, refer to specific elements that have to be considered when designing, developing and maintaining the community. These parts, as well as their respective constituent components, will now be discussed in turn. A. External Factors These factors refer to the surrounding environment of the community. They define its context of application and sustain the internal factors: 1) Context: We propose to classify the whole universe of virtual communities in 5 groups, according to the main topic or context they are related to: social, business, educational, professional, gaming. This classification relates to the fact that different kinds of communities have different structures, social cohesion and interaction patterns among users. Besides, we have to consider the technical environment that acts as support of the virtual community: desktop-based, mobile-based and desktop-mobile communities. This also defines the kind of interaction between members. For example, mobile devices support hardware capabilities such as geolocation sensors, that may add a new layer of information to the social and interest graphs that lay behind desktop-based virtual communities. Finally, we can also define a specific taxonomy of virtual communities according to the relative existence of one or more community managers, and whether or not the people that form the group already know each other. The idea of considering the context of a specific community relates to the fact that different communities have different structures and dynamics. Therefore, both the social and technical backgrounds define the first stage on how the different features that build up a PVC have to be designed. 2) Target: Every community also responds to a specific target of people (ethnographical analysis). Different kinds of users may require some specific usability or accessibility considerations that may impact dramatically in the community design and in participation. Also, different kinds of users have different kinds of needs that have to be addressed when designing both the social features and the motivation and participation strategies that would enable interaction within the community. 3) Privacy and Legal Issues: Members have to know beforehand what is allowed and what is not when working in a specific kind of community. This ensures a certain frame of formality and states basic norms of behavior and community standards that may impact heavily in participation. In fact, every community has to rely on a known set of rules that users need to know beforehand in order to ensure a proper ambiance. Moreover, these

A MODEL OF SUCCESSFUL VIRTUAL COMMUNITY

As a result of the literature review presented in the previous section, as well as observation and a previous field experience [9], we have conceived a model for designing successful partially virtual communities (Fig. 1):

Figure 1. Model of successful partially virtual community

The constructs in the model are grouped into two integrated areas, according to their relative place with the interaction within the community. In this model, we can identify two categories: external factors (blocks in solid line) and internal factors (blocks in dashed line). The former are individually related to the global structure of the community, and the latter are grouped together as factors that build up the design strategy to be considered when developing a community.

687

responsibilities go in both senses, as designers and community managers should offer proper services to users.

considering the members that are within the social graph of a specific user. This would turn out into a network effect that will sustain the community through time. With the intention of evaluating one of the core components of the model we propose in this paper, we will focus on developing a community and analyzing the response on its lifecycle when affecting its members’ motivation and participation. The idea of working with these variables comes from the need to quantify the relative impact of a followed participation strategy within the validity of the proposed model. In fact, the other core components (context, target, privacy policy, legal issues, governance design, and technological platform) are definitions that are at global scope when designing a community: they cannot be easily affected over time in a concrete way, even though they might change as the community grows or dies. We will be working on a community formed by people who are already familiar with each other, as well as with a community manager that will act as moderator. Its size will be consequently small and the design will be supported for short-time participation. We will be exploring the evolution of its lifecycle, from the very beginning until its end. With this community, we are interested in carrying out a proof of concept in order to study the community lifecycle and quantifying members’ participation. We will also analyze behavioral patterns that might appear.

B. Internal Factors These factors refer to specific points to be considered when designing, developing and maintaining a virtual community: 1) Community Governance: Designers should consider beforehand how they plan to enable users’ interaction within the community and how they plan to lead the process. Some examples of specific tasks to be undertaken are the definition of specific roles that go along with the member lifecycle, peer moderation, external moderation by a community manager, user empowerment, among others. In fact, virtual communities currently tend to lack strong governance structures. Taking this into consideration may dramatically improve the user experience within the group. 2) Motivation and Participation Strategy: The community has to be built around a strong motivation and participation strategy in both quantity and quality of contributions, following the members’ lifecycle. If either users or contributions fail to reach a critical mass, the community will die. Eventually, such a strategy may lead users from passive readers to active regular contributors. Some strategies for enabling this are discussed in [9]. They consider adding a ranking system, giving regular feedback about each member’s participation and including moderation of contributions to ensure quality. 3) Platform Development: A successful virtual community has to consider interface usability design as a key factor when developing the physical support of the group. The Website that supports the community should be ideally HTML5, CSS3 (Web design and development), as well as WCAG (Web accessibility) compliant. In fact, these technologies are considered as recommendations made by the World Wide Web Consortium (W3C), the main international standards organization on Web development. 4) Feedback Loops: In order to maintain participation and interaction within the community, designers and community managers should enable and favor feedback loops. A feedback loop is defined as the causal path that leads from the initial generation of the feedback signal to the subsequent modification of the event. In terms of participation and virtual communities, one of the many possible ways of achieving this is by enabling on-site notifications, where each user gets real-time information about what is currently happing within his/her social and interest graphs, as well as sharing relevant content within the whole community. Also, a way to enable feedback loops is to ensure community rituals (such as welcoming new members) through time. This can have a dramatic impact on promoting users’ identity. For example, by allowing notifications and self-expression in the form of a personal profile, as well as providing a targeted flow of information

IV. PROOF OF CONCEPT The goal of this proof of concept is to obtain initial concrete data on how participation strategies affect the community lifecycle and how members react to it. In fact, we consider this to be one of the fundamental issues that have to be undertaken before designing a PVC. That way, it will be easy to carry out further field research on other related topics afterwards, such as, how to design proper community governance strategies, and the impact of this model in different contexts. A. General Context We designed and developed a PVC that was on service for 15 days. As a support technology, we developed the platform using Elgg, an open-source PHP framework intended for developing online communities. The community ran under an Apache-MySQL environment. In order to adapt the Website to our needs, we developed some additional features, such as an improved ranking system, including an algorithm for modeling and quantifying participation. Before its launch on service, a usability test was performed in order to ensure a certain level of quality, by having direct input on how real users might use the proposed system and comparing it to its intended purpose. The community we developed was used as a support for an introductory course on Information Technology, at the Business School of Universidad de Chile. The group consisted of 43 students, ages 21-24 years old. Users were free to sign up and use the supporting community, intended as a tool for discussing topics that were already treated in the regular lectures of the course.

688

profiles, which corresponds to their current level of participation. This tag is shown next to the nickname of each community member and is visible all through the platform. Through this kind of modification, we aim to measure the effect of this label and the kind of information it represents over the value of participation of each user over two consecutive observation periods. At first, we set up these checkpoints to be daily, alongside with the publishing of the new discussion topics by the community managers. We found out there was enough data to infer the behavioral patterns of users within this time period. All the users see their levels change at the same time, once a day. According to the level of participation, we classify community members into three groups: those who had either a LOW, MEDIUM or HIGH level of participation. As it was stated before, users may up or down their current levels: we put into practice a security band feature in order to lower the effects of frustration when users grade down. This consists of defining a small gap around the limits for grading down, so a user that sees his/her participation declining will keep their current level and a warning is placed into action. The fact that users may grade down their levels of participation responds to the fact that users might feel at risk of not participating as hard as the others. However, users do not know about these rules and some of them learnt them by practice. We plugged a functionality into the community allowing users to rate the discussion topics created by the other community members. Each user has the right to vote for one topic created by other members utilizing a scale from 0 to 5 points. This vote is linked to the personal perception (such as, for example, pertinence, usefulness, among others) that users have over a specific discussion topic. At a final stage of observation, we stopped publishing new discussion topics as community managers. The idea of this stage is to evaluate the sustainability of the community and the behavior of the different kinds of members. Table 2 shows how the proposed model is applied in the developed community that served as proof of concept:

B. Design Issues We asked students to create a nickname that would identify them throughout the Website. This was done in order to motivate shy students to participate in the community, as well as try to keep the real identity of the users within the community anonymous. However, when signing up, we required the students to validate their identities with their university email address, in order to document internal tracking of what has been done in the community. Each day we proposed two different discussion topics linked to the contents treated in the regular course. Students were also free to propose their own. In order to ensure a certain level of participation and commitment with the community, we established an initial goal of 6 participations during a week. We tracked both logs and participations of all users in the community on a daily basis. As a participation strategy, we chose to add a special tag to all the users, corresponding to their current level of participation during the observation period. This kind of interaction sticks to one of the core factors that build up the proposed model. The tags used were: LOW, MEDIUM and HIGH, according to their quantity and quality of contributions. Students could either up or grade down their ranking. We also developed a safety band feature in order to prevent users getting frustrated when they seemed to lower their category. It is worth pointing out that there was only intrinsic motivation in the model of community we developed. This strategy was employed to avoid the influence of external rewards, such as grades or credits in the course, which would eventually affect the pertinence of the objectives we wanted to test with this field research. In order to model a function that measures participation for each user, we considered a list of pre-defined tasks to be performed through the site. We assigned a weight factor to each expected task, and finally the participation would be the sum of number of hits for each task, multiplied by its weight factor. The values of each weight factor were defined because harder tasks and those having a larger impact in the participation within the community will score higher than easier or rather passive tasks. Table 1 presents the list of initial tasks to be performed and its initial weight factors during the experience: TABLE I.

TABLE II.

Factor Context

TASKS AND WEIGHT FACTORS

Task

Target

Weight Factor

Log into the community

0.5

Reply to a message from community manager

1.0

Post a new discussion topic

2.0

Reply to a message from a member

1.0

Rate a discussion topic

1.0

Privacy and Legal Issues Community Governance Participation and Motivation Platform Development Feedback Loops

We aim to quantify the participation of community members and encourage them to improve their quantity and quality of contributions by linking a visible tag to their

689

DESIGN ISSUES OF THE DEVELOPED COMMUNITY

Feature Support for discussions under an educational environment Business and Management students (20 – 24 years old) Published under a specific section of the Website. Account validation (via email) when signing up. Community Manager Rankings Peer Rating Open-Source Apache / PHP / Myself Environment Profile and Activity Notifications within the Social Graph of a Specific Member

explained because we sent a feedback message to all the students telling them about the existence of the community and the possibility they have to discuss about the topics treated during each lecture. On day 6 we introduced peer rating as a new feature. We can also consider that the wordof-mouth and the increasing activity on the community through time motivated users to create their accounts.

C. Methodology We carried out this experience in three consecutive stages. These corresponded to the steps carried out when launching the community, as well as intentionally affecting the participation strategy, and measuring its impact in the community and members’ lifecycles. We started from the very beginning by inviting users to take part in the community, until affecting and causing its demise by the action of the community manager. The stages that describe the methodology of this proof of concept are as follows: 1) Stage 1 (Invitation): The first stage was meant to invite the whole group of students enrolled in the course to take part in the community. We achieved this by telling students at the end of one of their lectures, that an online community was put on service in order to act as a support for discussion and for helping them with their study. 2) Stage 2 (Creation and Validation of Accounts): In the second stage, students created and validated their accounts, and set up their profiles. It is worth pointing out they were not forced to do so, but if they decided to take part of the community, they had to respect the initial commitment of posting six contributions within a specific week. 3) Stage 3 (Measure Community Activity): In this stage, we measured the proposed motivation and participation strategy, measuring quantity and quality of contributions. We also tracked data about user behavior and activity throughout the PVC lifecycle.

Figure 2. Evolution through time of created validated accounts

We divided the evaluation period according to the intervention of the community manager. The first time, we studied the effect of ranking (from day 2) and peer ratings (from day 6). On day 8 we stopped publishing new discussion topics and we took back the control on day 13 by posting new messages. Table 3 shows the result of the mean number of contributions during each one of these periods, measured as an average of the number of contributions over the length in days of each one, considering the role (active or passive) that the community manager took. The active role is defined as the community manager publishing daily discussion topics in order to enable the participation of members.

D. Research Questions In order to test the pertinence of designing a motivation and participation strategy, we carried out a first field proof of concept. These questions will help us to understand how participation is related to the design of a strong PVC. We state the following set of research questions: • (RQ 1) Does the number of users log into the community directly influence member participation? • (RQ 2) Is member participation affected by the ranking the community gave to them? • (RQ 3) Is the activity level of a community related to the number of active contributors that it has? • (RQ 4) Does the peer rating system affect the participation and content creation throughout the community? • (RQ 5) Does the community stop its activity when the community manager stops proposing new discussion topics?

TABLE III.

PARTICIPATION AND ROLE OF THE COMMUNITY MANAGER

Overall Participation

Mean Participation

201

28.7

Passive (days 8 to 12)

6

1.2

Active (days 13 to 15)

0

0.0

Role Active (days 1 to 7)

We will focus at first on understanding the participation strategy during the first active stage. Figure 3 shows the average number of contributions made by all the users that had created and validated their accounts from the beginning of the observation until the end of the passive period. At the end of the first active period of the community manager, we registered a total of 201 contributions made by the whole group of community members. Since 30 of them contributed at least once into the platform, each member made on average 0.95 contributions per day. Besides, 41 contributions of this total (in the form of 8 new discussion topics with their respective comments) were created and maintained by members of the community. During this period, this correlates to 20.39% of all the contributions made to the site.

E. Results All the students enrolled in the course (43) created their accounts over a time-span of 7 days. 30 of those students, corresponding to the 69.77% contributed at least once in the community. Nine of the students, who did not contribute, created their accounts but never logged back again. Figure 2 shows the evolution through time of created validated accounts. We can see that there are two local peaks on the number of validated accounts. The first one (day 2) is

690

Figure 3. Average number of contributions

Figure 5. Average number of log-ins

We can conclude that participation has local peaks when social features were added to the community and when community managers make contributions. However, this effect does not last long in time, since participation tends to decrease over the following days. Also, the higher peak on participation corresponds to the introduction of the peer rating system. It is worth noting that from day 8, when the community manager stopped publishing new discussion topics, participation dropped dramatically and stayed below a certain level of critical mass that finally led to the disintegration of the community. Figure 4 shows the distribution of member categories along time. We took into account only data from the first week of observation, since information from day 8 onwards is not relevant enough due to a lack of critical mass of contributions. Information is only available from day 2, since we needed an initial starting point for performing the analysis.

We note that after participation dropped, as well as did the log-ins into the system, the community entered into an irreversible process leading to its demise. This is due to not having reached the critical mass of log-ins and contributions needed in the community in order to ensure and sustain participation. Moreover, community members did not react as a whole group to feedback messages that were sent to them from day 13 onwards, indicating that the community manager created and reactivated the discussion topics. Only a few contributions from a selected group of users (those with the HIGH label) appeared and contributed in a quite small number. F. Analysis of Research Questions According to the results of this proof of concept, we may accept in this particular context that participation tends to rise when users log in more often into the community (RQ 1). As we can see on Fig. 4, users identified with LOW participation tend to decrease in time. This is due to several factors, such as new social features added to the community, as well as a response to their participation tag. We can also state peer pressure as a factor affecting the level of participation of this particular group. However, we need to measure participation in a larger gap of time in order to properly validate (RQ 2) and (RQ 3). As we can observe in Fig. 3, there is a local peak of participation when the peer rating system was introduced in the community. This partially explains and validates (RQ 4), since participation may also have improved because of other factors such as the spread by word-of-mouth about the community and the cumulative effect in time of the most active users’ participation. Finally, we may validate in this context (RQ 5), since the community clearly stopped its participation when the activity of the community manager ceased. This can be seen under Fig. 3 and more clearly on Fig. 5, showing that community members logged in only a couple of times, as compared to the activity during previous days.

Figure 4. Distribution of member categories

Among all the group of users, we identified a group of 10-20% of them with the label of “high participation” each day. These users also revealed themselves as a kind of “leader” by promoting discussion among the other members of the community. Their comments were the most rated ones and they maintained a steady participation during the whole observation process. They correspond to a 9.5% of the members of the community, and contributed with 28.9% of the overall participation during the initial week of observation. Figure 5 shows the average number of logs per day, consisting of all the created and validated accounts of each day.

V.

CONCLUSIONS

In this paper we proposed to tackle the problem of designing successful PVCs. We proposed a conceptual model and tested through field research a participation strategy based on intrinsic motivation. We analyzed the interaction among users, as well as the contribution patterns

691

they had over time, as well as analyzed the role of the community manager. We believe this study will bring ideas to designers on how to develop strong and sustainable PVCs, as well as how community managers may maintain and evolve communities through time. We consider that one of the fundamental issues to take into consideration when designing PVCs is to define a strong participation strategy, adapting it to the current stages of the community lifecycle and to how members are reacting towards it. Therefore, we developed, tracked and analyzed throughout their lifecycle, participation patterns in which members could reply or post their own discussion topics. The idea of working with these variables came from the need to quantify the relative impact of a followed participation strategy within the validity of the proposed model. We initiated two social functionalities in the community that seemed to work well, since they triggered participation with visible local peaks. Feedback proved also to be a plausible way to improve participation at certain periods of time. Highly active members acted as leaders who encouraged the whole group to participate, as well as proposed new discussion topics, which were highly praised by the community. The community decreased its level of participation when the community manager stopped to proposing new discussion topics (passive stage). In other words, participation during these dropped, reaching a level that could not be sustainable for the community. Consequently, this led to a reduced number of logs into the system, and the community finally entered into an irreversible process causing its death. This is due to not reaching the critical mass of log-ins and contributions to the community to ensure participation. When we introduced the role of community manager proposing new discussion topics again, members did not react, even if we also sent them a feedback message telling them about these new messages. Only highly committed students kept logging and posting messages. As future work, we will study community governance strategies and how they impact the global design of the community. We believe that when there is a strong governance structure, communities will regulate themselves and they will be successful over time without external influence. Also, we are interested in studying the applicability of this model in different kinds of contexts, as well as exploring different technical environments, such as mobile-based virtual communities, exploiting specific hardware capabilities and how they will eventually influence participation.

[3]

[4]

[5]

[6] [7]

[8]

[9]

[10]

[11]

[12] [13]

[14] [15]

[16]

[17] [18] [19] [20] [21]

[22]

[23]

ACKNOWLEDGEMENTS This work has also been partially supported by Fondecyt Project (Chile), grant: 1120207.

[24]

REFERENCES [1] [2]

[25]

http://www.facebook.com/press/info.php?statistics (last consulted on November 12, 2011) http://www.google.com/adplanner/static/top1000/ (last consulted on November 2, 2011)

692

P.B. Brandtzaeg, J. Heim, “User loyalty and online communities: why members of online communities are not faithful,” Proc. 2nd Int. Intelligent Technologies for Interactive Entertainment, Jan. 2008. P.B. Brandtzaeg, J. Heim, “Why people use social networking sites,” in Online Communities LNCS 5621, Springer-Verlag Berlin Heidelberg, 2009, pp. 143-152. C.M.L. Chan, M. Bhandar, K.-B. Oh, H.-C. Chan, “Recognition and participation in a virtual community,” Proc. of the 37th Hawaii Int. Conference on System Sciences, IEEE Press, Jan. 2004. T. Connolly, B. K. Thorn, Discretionary Databases: Theory, Data, and Implications. Newbury Park, CA: Sage Publications, 1990. R. Cullen, S. Morse, “Who’s contributing: do personality traits influence the level and type of participation in online communities?,” Proc. 44th Hawaii Int. Conf. on System Sciences , Jan. 2011. L.C. Giles, G.F. Glonek, M.A. Luszcz, G.R. Andrews, “Effect of social networks on 10 year survival in very old Australians: the Australian longitudinal study of aging,” Journal of Epidemiology and Community Health 59(7), pp. 574-579, 2005. F. Gutierrez, N. Baloian, G. Zurita, “Boosting participation in virtual communities,” Proc. of CRIWG 2011 LNCS vol. 6969, SpringerVerlag Berlin Heidelberg, 2011, pp. 14-29. J.J. Han, R.J. Zheng, Y. Xu, “The effect of individual needs, trust and identification in explaining participation intentions in virtual communities,” Proc. of the HICSS’07, IEEE Press, Jan. 2007. M.G. Hunter, R. Stockdale, “Taxonomy of online communities: ownership and value propositions,” Proc. of the 42nd Hawaii International Conference on System Sciences. IEEE Press, Jan. 2009. A. Iriberri, G. Leroy, “A life-cycle perspective on online community success,” ACM Computing Surveys, vol. 41, no. 2, 2009. A.N. Joinson, “Looking up’ or ‘Keeping up with’ people? Motives and Uses of Facebook,” CHI’08: Proc. of the 26th annual SIGCHI’08. ACM Press, pp. 1027–1036. New York, USA, 2008. A.J. Kim, Community building on the Web. Peachpit Press, 2000. .J. Koh, Y.-G. Kim, B. Butler, G.-W. Bock, “Encouraging participation in virtual communities,” Comm. of the ACM, vol. 50, no. 2, Feb. 2007. H.-F. Lin, “Determinants of successful virtual communities: contributions from system characteristics and social factors,” Information & Management, vol. 45, Oct. 2008, pp. 522-527. E. Mousavidin, L. Goel, “A life cycle model of virtual communities,” Proc. of HICSS’09, IEEE Press, Jan. 2009. B. Nonnecke, J. Preece, "Why lurkers lurk," Proc. of the 7th Americas Conf. on Inf. Systems, Boston, USA. 2001. R. Plant, “Online communities,” Technology in Society, vol. 26, 2004, pp. 51-65. J. Porter, Designing for the social Web. New Riders: Berkeley, 2008. J. Preece, “Sociability and usability in online communities: determining and measuring success,” Behavior and Information Technology Journal, vol. 20, issue 5, Jul. 2001, pp. 347-356. J. Preece, B. Nonnecke, D. Andrews, “The top five reasons for lurking: improving community experiences for everyone,” Computers in Human Behavior, vol. 20, 2004, pp. 201-223. J. Preece, B. Shneiderman, “The reader-to-leader framework: motivating technology-mediated social participation,” AIS Transactions on Human-Computer Interaction, vol. 1, issue 1. Mar. 2009. BS.J.J. Tedjamulia, D.L. Dean, D.R. Olsen, C.C. Albrecht, “Motivating content contributions to online communities: toward a more comprehensive theory,” Proc. of the 38th Hawaii International Conference on System Sciences, IEEE Press, Jan. 2005. M. Westerlund, R. Rajala, K. Nykänen, Timo Järvensivu, “Trust and commitment in social networking – Lessons learned from two empirical studies,” Proc. of the 25th IMP Conference, Marseille, France, 2009.