The impacts of information quality and system quality on users' continuance intention in information-exchange virtual communities: An empirical investigation

Decision Support Systems xxx (2012) xxx–xxx

DECSUP-12226; No of Pages 12

Contents lists available at SciVerse ScienceDirect

Decision Support Systems

j ourna l homepage: www.e lsev ie r .com/ locate /dss

The impacts of information quality and system quality on users' continuance intentionin information-exchange virtual communities: An empirical investigation

YiMing Zheng, Kexin Zhao 1, Antonis Stylianou ⁎Department of Business Information Systems and Operations Management, Belk College of Business, University of North Carolina at Charlotte, 9201 University City Blvd, Charlotte,NC 28223, USA

⁎ Corresponding author. Tel.: +1 704 687 7605.E-mail addresses: [email protected] (Y. Zheng), kz

[email protected] (A. Stylianou).1 Tel.: +1 704 687 7637.2 http://www.eweek.com/c/a/Search-Engines/Yaho

Products-775120/. Retrieved on 2/28/2012.

0167-9236/$ – see front matter © 2012 Published by Elhttp://dx.doi.org/10.1016/j.dss.2012.11.008

Please cite this article as: Y. Zheng, et al., Theexchange virtual co..., Decision Support Syste

a b s t r a c t
a r t i c l e i n f o
Available online xxxx

Keywords:Virtual communitiesIS post-adoptionIS Success modelInformation qualitySystem qualityIndividual benefits

An information-exchange virtual community (VC) is an IT-supported virtual space that is composed of a group ofpeople for accessing, sharing and disseminating topic-related experiences and knowledge through communicationand social interaction [36,43].With the increasing number of VCs and low switching cost, it is challenging to retainexisting users and encourage their continued participation. By integrating the IS post-adoption research and ISSuccess model, we propose a research framework to investigate VC users' continuance intention from a qualityperspective. Based on a field survey, we find that information and system quality directly affect perceived indi-vidual benefits and user satisfaction, which ultimately determine user continuance intention to consume and toprovide information. Furthermore, by modeling information quality and system quality as multifaceted con-structs, our results reveal key quality concerns in information-exchange VCs. Implications for VC design andmanagement are also discussed.

© 2012 Published by Elsevier B.V.

1. Introduction

An information-exchange virtual community (VC) is an IT-supportedvirtual space that is composed of a group of people for accessing, sharingand disseminating topic-related experiences and knowledge throughcommunication and social interaction [36,43]. Examples include onlineforums, message boards, news groups, etc.

Due to the prevalent Internet access and increasingly advancedWeb2.0 applications, the growth of information-exchange VCs has beenphenomenal and millions of people have become VC users in recentyears. For example, Yahoo Groups has 115 million users and formed10 million groups2 (eWeek.com). IMDB (Internet Movie Database)has attracted over 4 million users with more than 13 million posts asof January 2012 (www.big-boards.com).

A user decides his participation based on his individual needs andexperiences of using a VC. Should he find that a VC does not satisfyhis needs, he can stop using the VC or switch to another VC of thesame type, if available. However, retaining existing users is criticalfor a VC's long-term development [35,71]. According to marketing re-search, existing customers may exhibit voluntary citizenship behaviors(e.g., helpful, constructive behaviors) that are valued or appreciated bythe organization, in addition to consumption of pre-paid products or

[email protected] (K. Zhao),

o-Refreshes-Upgrades-Some-

sevier B.V.

impacts of information qualims (2012), http://dx.doi.org/

services [31]. It is found that continued membership positively in-creases members' identification with the organization [8] and reducesthe likelihood of lapsing [7]. Furthermore, according to the theory of net-work externalities [38], if a VC canmaintain a large pool of existing users,it will attract more new users. Individuals are more likely to join largerVCs than smaller ones, as larger VC are assumed to have more informa-tion sources [32]. The presence of network externalities also enablesVCs to leverage economies of scale to operate and grow in a cost-effective way and provide more benefits to users [29]. Therefore, it isimportant to understand what factors drive the continuance intention toparticipate in VCs.

Although a number of information systems (IS) studies have exam-ined user participation behaviors in VCs [13,36], a limited number ofstudies have paid special attention to user retention and continuedparticipation [15,71]. It is a challenging issue given that user participa-tion is voluntary. Ma and Agarwal [48] reported that not many VCswere successful in retaining users and motivating their continuedusage, which ended up with membership loss.

To address this issue, we investigate users' continuance intention toparticipate in VCs by examining the role of information quality andsystem quality. Butler [13] argued that the amount of information onits own is not enough to retain users, unless it is transferred to benefitsfor users, leading to a sustainable VC. Gu et al. [32] found that the valueof a VC increases with the number of high-quality postings, whichhelps users achieve individual benefits and meet their needs. Users aremore likely to adopt high-quality information as it provides judgment-relevant content [72]. High-quality information also enhances the repu-tation of a VC and user loyalty, and can serve as a competitiveweapon toattract and retain members [45]. Furthermore, user participation is

ty and system quality on users' continuance intention in information-10.1016/j.dss.2012.11.008

http://www.big-boards.com

http://dx.doi.org/10.1016/j.dss.2012.11.008

mailto:[email protected]



http://www.eweek.com/c/a/Search-Engines/Yahoo-Refreshes-Upgrades-Some-Products-775120/

http://www.eweek.com/c/a/Search-Engines/Yahoo-Refreshes-Upgrades-Some-Products-775120/

Unlabelled image


Unlabelled image

http://www.sciencedirect.com/science/journal/01679236


2 Y. Zheng et al. / Decision Support Systems xxx (2012) xxx–xxx

largely facilitated through an effective technical infrastructure [50]. Dueto the limited information processing capability of human beings, an ex-cess increase in information volume is likely to cause informationoverload, discourage active participation and, ultimately, lead to mem-bership loss [13,36]. VCs need to implement good filtering or organiza-tion mechanisms to minimize users' efforts on processing information.When a VC provides easy and flexible ways for users to browse/postinformation and interact with others, it will be more likely to encourageusers' involvement, making participation interactive and enjoyable.In these regards, we argue that the quality of both the information itselfand the technical infrastructure matter. Ultimately, quality can be usedas a strategy to “lock-in” users [32].

As the focus of this study is the impact of quality on continuanceintention, the selection of information-exchange VCs helps to mini-mize confounding factors salient in other types of VCs. For instance,in social networking sites, users' continuance intention to participatecould be driven by not only information and system quality but alsoby personal relationships developed offline, etc.

Recently, a number of researchers have investigated quality in VCs.Lin and Lee [46] argued that information quality influences user satis-faction and intention to participate, which will determine their loyaltyto a VC. Cheung et al. [16] found that the relevance and comprehensive-ness of electronicword-of-mouth influence information usefulness and,subsequently, the users' information adoption decisions in a food VC. Intheir study of a Korean stock message board, Park et al. [54] indicatedthat higher perceived quality leads to higher perceived usefulness ofthe community, which impacts users' seeking and sharing behaviors.Lin [45] found that higher information quality and system qualityincreases user satisfaction, leading to more member loyalty. Zhangand Watts [72] showed that both information and system qualityhave a positive relationshipwith information adoption from a cognitiveperspective in two online forums. In a study of social networking sites,Zhang [71] found that information quality plays a substantial role indeveloping sense of community, while system quality does not.

While the IS studies reviewed above have examined IS quality inVCs, the majority of them treated quality as an abstract and aggregatedconcept. Prior research has widely agreed that IS quality is a multiface-ted concept with a variety of dimensions [52,63]. These quality dimen-sions have been extensively examined in the context of organizationalIS and E-commerce websites. Unfortunately, they have not been fullyunderstood in the VC context. Although Zhang [71] investigated bothinformation and system quality from multi-dimensional perspectives,this study differs from Zhang's paper as follows. First, we explicitlydifferentiate consumption intention from provision intention. Second,we focus on continuance intention, not frequency of usage analyzedby Zhang [71]. Third, this study examines information-exchange VCsinstead of social networking sites studied by Zhang [71].

We draw on two research streams to develop our model. The ISpost-adoption literature helps us understand what factors directlyaffect users' intention to continue using a system. The IS Successmodel provides us a guideline to investigate system usage from aquality perspective. The integration of the two streams enables us tobetter understand the impacts of quality on users' future participationin VCs. Based on a field survey, we find that information and systemquality directly affect perceived individual benefits and user satis-faction, which ultimately determine user continuance intention toconsume and to provide information. Furthermore, by modeling in-formation quality and system quality as multifaceted constructs, ourresults reveal key quality concerns in information-exchange VCs.

The remainder of this paper is organized as follows. In Section 2, wepresent the theoretical background. Section 3 highlights the special fea-tures of VCs that call for extension of prior studies. The research modeland hypotheses are proposed in Section 4. Section 5 discusses method-ology and data collection, followed by analysis and results in Section 6.Discussion, contributions, limitations and future research are presentedin Section 7, with concluding remarks in Section 8.

Please cite this article as: Y. Zheng, et al., The impacts of information qualiexchange virtual co..., Decision Support Systems (2012), http://dx.doi.org

2. Theoretical background

2.1. IS post-adoption research

The IS post-adoption literature [9,37,55] extends the research on in-dividual technology acceptance research [22,62] to examine user beliefsand attitudes after initial IS use. Research on initial IS adoption is wellstudied under the framework of the Technology Acceptance Model(TAMand TAM2) [22], theUnited Theory of Acceptance andUse of Tech-nology (UTAUT) [62] and other related models. There are a number ofkey factors that determine initial adoption such as, perceived usefulness,perceived ease of use, subjective norms/social factors/image, perceivedbehavioral control, etc. This research stream has studied how and whyindividuals adopt new IS [62]. In contrast, research on IS post-adoptionattempts to understand how and why individuals continue using ISafter the initial adoption. IS post-adoption studies emphasizes that con-tinuing IS usage is driven by conscious decisions from past experience[9,39], mainly based on two aspects: perceptions of usefulness and affec-tive or emotional responses to the use of IS. Perceived usefulness refers toa user's ex-post expectations and beliefs about system effectiveness andthe net benefits of systemuse frompast experience [9]. Affective or emo-tional responses to the use of technology, called user satisfaction, is auser's emotional or psychological state following IT use experience [22].

A number of studies have shown that initial IS adoption and post-adoption are different, as perceptions and beliefs about IS usage changeover time and individuals get used to the system. Bhattacherjee [9]developed the IS continuance model in which perceived usefulnessand user satisfaction are the two direct predictors of IS continuance in-tention. Following the same logic, Kim and Son [39] found that perceivedusefulness and satisfaction serve asmechanisms of commitment leadingto continued use of the same online service. Karahanna et al. [37] foundthat behavioral beliefs about using an IS (perceived usefulness) determinethe intention to continue using it, whereas subjective norms fromtop management, supervisor and peers only influence an individual'sdecision on initial adoption, NOT continued usage in the later stage.Perceived ease of usehas been excluded from the IS-post adoption studiesdiscussed above. The core premise is that perceived ease of use maybecome insignificant in predicting future use as an individual gains ex-periences by keeping using the system [9,10]. Hence, we do not includeperceived ease of use in the model.

2.2. The IS success model

The IS Success model, initially developed by DeLone and McLean[23], provides a clear taxonomy for conceptualizing andoperationalizingIS success. The model includes six dimensions of IS success: informationquality, system quality, use, user satisfaction, individual impact, and or-ganizational impact. In the updated version of the IS Successmodel [24],individual impact and organizational impact are collapsed with otherimpacts (e.g., consumer impact, societal impact) into one category,called net benefits. Service quality is added as another dimension toreflect the effectiveness of the service provider, such as the IS depart-ment of an organization or customer service of an e-commerce website.

2.3. The integration of the IS post-adoption research and IS success model

The primary reason for integrating the IS post-adoption literatureand IS Successmodel is to deepen our understanding of the role of qual-ity in the IS post-adoption stage. IS post-adoption research has widelyagreed that continuance intention is directly determined by affectivecharacteristics of the system such as perceived usefulness and usersatisfaction based on past experiences [9,37,39]. However, it does notspecifically discuss the role of quality. As DeLone and McLean [24]suggested, net benefits “cannot be analyzed and understood without‘system quality’ and ‘information quality’ measurements” (pp. 25). Hence,

ty and system quality on users' continuance intention in information-/10.1016/j.dss.2012.11.008


3Y. Zheng et al. / Decision Support Systems xxx (2012) xxx–xxx

we refer to the IS Success model that pinpoints the primary role of qual-ity and incorporate quality into the post-adoption usage in our model.

DeLone and McLean [24] also emphasized that the IS success modelis a process model in which the linkages among success dimensions arenot a causal sense. The causal relationships among selected successdimensions should be contingent on the objectives and context of theempirical investigation, as shown in Fig. 2 in the updated IS Successmodel [24]. In our case, we are interested in the impact of quality.

Although the IS Successmodel can provide us a guideline for systemuse from a quality perspective, it focuses mostly on intention to use oruse. Intention to use refers to the first-time adoption intention of anindividual who has not used the system before. Use refers to the actualusage in terms of duration, frequency and depth of use. The IS Successmodel does not explicitly explain why existing users continue usingan IS. Furthermore, system use is considered as one success measurein the original model, and it is embedded in a feedback loop: initialuse results in net benefits, which influence subsequent use. To specifi-cally model continuance intention as the dependent variable, we referto IS post-adoption studies in order to build the causal associationbetween net benefits, satisfaction, and continuance intention. In theseregards, the two streams are complementary, serving as the theoreticalfoundations for our research model.

3. VCs as a special type of information systems

In addition to combine insights from two theoretical streams, weextend prior research by incorporating the special features of VCs,which differentiate them from traditional IS.

3.1. User participation in VCs

Users can participate in VCs in two ways: information consumptionand information provision [54]. VC users can consume information bybrowsing or posting questions for help. They can also contribute infor-mation by replying to questions, and initiating or participating in topicdiscussions in VCs. Researchers have agreed that information consump-tion and provision are indispensable parts of VCs [3,54], as they are bothdesirable social behaviors [58]. According to marketing research, tomaintain membership activities, users are expected to coproduce aswell as consume information contents [31]. Thus, it is critical to under-stand these two different types of participation behavior in VCs [11].

There are a number of VC studies that examine continued participa-tion. Chen [15] found that contextual factors (social interaction ties)and technological factors (knowledge quality and system quality) posi-tively influence users' continuance intention to exchange knowledge inthe future in a professional VC for knowledge workers. In their study ofFacebook,Wang et al. [64] illustrated that computer self-efficacy affectsFacebook users' continued usage through cognitive and affective deci-sion processes. Similarly, in his study of social networking sites, Zhang[71] argued that sense of community is a significant factor inmotivatingusers to use the services frequently. However, none of these studieshave considered information consumption and provision separately.

3.2. Individual benefits of participating in VCs

Individuals can participate in VCs for a variety of reasons. Accordingto the Social Exchange Theory [26], people engage in social interactionin the hope that they can get some rewards back from the interaction.This suggests that individuals are motivated to participate in a VC byvarious benefits. Butler et al. [14] summarized four types of individualbenefits: information benefits, social benefits, visibility benefits, andaltruistic benefits. Information benefits refer to access to useful informa-tion [21,27,66]. Social benefits are social and spiritual support that userscan get through information exchange with others [40,65]. Visibilitybenefits are achieved when a personal reputation is enhanced as a resultof contributing knowledge or being socially active in aVC [66]. Compared

Please cite this article as: Y. Zheng, et al., The impacts of information qualiexchange virtual co..., Decision Support Systems (2012), http://dx.doi.org/

with the previous three types of benefits, which are about self-interests,altruistic benefits refer to the psychological happiness achieved throughpro-social behaviors of helping others [21]. Sometimes, users may feelobligated to give back to the VC that has given much to them [12].

The IS post-adoption research uses perceived usefulness and the ISSuccess model uses net benefits to capture user evaluation of thesystem performance in improving task performance and effectivenessin the organizational and E-commerce contexts. In the VC context, themeaning of perceived usefulness and net benefits should be broadenedas VC users can obtain different types of individual benefits. There-fore, we newly develop the perceived individual benefits construct.

3.3. Information quality and system quality in VCs

In VCs, the majority of information contents are user-generated andthere is limited control over whowill post what information, and wheninformation will be posted. As a result, user perception of informationquality in VCs could be different from that in traditional IS. For example,information reliability may be a big concern to VC users because infor-mation comes from strangers whom would be difficult for users totrust and rely on. In contrast, when employees use an IS at work, infor-mation reliability can be greatly ensured because information enteredinto the system is typically processed and managed ex ante. Objectivityof the contents could also be perceived differently in VCs. As users arefree to express their opinions in VCs, those opinions are subjective inthe absence of clear-cut answers. However, information in organiza-tional IS and E-commerce websites is more factual and transaction-oriented. In addition, in VCs the breadth and depth of information areimportant, as users expect to find useful information from differentperspectives and with detailed information that can enhance theirunderstanding and decision-making.

System quality dimensions may also have different relative impor-tance in VCs. In VCs, especially the larger ones, huge amounts of informa-tion are being posted every day and information overload is more likelyto occur. Thus, users need an effective navigation tool to facilitate theirsearch process and minimize information processing cost [36]. Since VCusers do not interact with each other face-to-face and are strangers inthe real world, they may be concerned about their personal informationbeing disclosed to the public. It is crucial for a VC to protect individualusers' privacy andmake users feel safe and comfortablewhen participat-ing in the VC. Therefore, security is an important system quality dimen-sion in VCs. Relatively speaking, in the organizational context, whereonly internal employees can use a particular system, security can be han-dled and guaranteed more easily than in VCs. Another salient systemquality dimension is interactivity. It captures the degree towhich the sys-tem can facilitate direct user interaction in a VC. It includes interactingwith other users by variousmethods, and allowing users to access others'profiles etc. By contrast, traditional IS users tend to have less direct onlineinteraction with others as they simply obtain information from thesystem within the organization. Ma and Agarwal [48] argued that inter-activity should be greatly encouraged and expected in VCs because activeuser interaction is critical to information exchange in VCs.

4. Hypotheses development

Based on the above discussion,we examine two types of continuanceintention: continuance intention to consume and continuance intention toprovide. The former represents the extent towhich a user intends to con-tinue browsing or seeking information in a VC, while the latter refersto the extent to which a user intends to continue contributing hisknowledge.

Based on the IS post-adoption literature, we propose that perceivedindividual benefits and user satisfaction are the two direct antecedentsthat will influence existing users' continuance intention to consume andprovide information in a VC. Grounded on the IS Success model, weargue that perceived information quality and perceived system quality




are theprimary factors that impact individual benefits and user attitudestoward the system, ultimately leading to continuance intention in theVC. Although the IS Success model does not build direct links betweeninformation quality/system quality and perceived individual benefits,the relationships were validated by a couple of organizational studies[1,20]. Below is the researchmodel (Fig. 1), in which the arrows indicatea positive impact from one construct to another (e.g., perceived indi-vidual benefits positively influences continuance intention to consume).

In this study, we do not consider service quality, which captures thequality of typical service-related activities provided by IS departmentswithin organizations (e.g., IS training and support) or online shoppingwebsites (e.g., order tracking and return). The reason is that the qualityof most services provided by a VC (e.g., information search and sharing)can be encompassed in either information or system quality. In addi-tion, active involvement from VC managers is typically limited sinceinformation exchange is mostly driven by individual users.

4.1. Determinants of continuance intention in VCs

According to the IS post-adoption research, perceived individualbenefits are a major determinant of continuance intention to use thesystem. We define it as the extent to which an individual's perceptionof the benefits of participating in a VC based on his experiences[10,34,39]. It captures four types of benefits in VCs: information, social,visibility and altruistic benefits.

In an information-exchange VC, a user's primary goal is to accessinformation either by purely browsing information or by participatingin a discussion with others. During this process, users can receive notonly information, but also social support and recognition within thecommunity [12]. Moreover, a voluntary behavior of helping otherswill occur when users enjoy contributing to the VC to benefit others.So, information exchange could be more than a learning process.When users can benefit from information sharing, social support andvoluntary contribution, their sense of community will be enhanced[12]. As a result, this will greatlymotivate users' involvement and activeparticipation in the VC in the future.

In addition, according to the Social Exchange Theory [26], informa-tion exchange in VCs is a reciprocal process. When users can obtainuseful information and social support from others in a VC, they may bemotivated to help others as a return of what they have received. Onthe other hand, users who provide information may also expect to re-ceive help and access to superior resources from others. It is foundthat active users who provide information or participate in discussionwill feel discouraged if they cannot get useful information and responsesfrom the community [45]. Thus, it is expected that individual benefitswill motivate two-way participation.

When users get benefits from participating in a VC, they will formconcrete expectations and their beliefs about the VC's performancewill be updated [10]. For instance, higher perceived benefits lead toincreased users' expectation, which may encourage them to continueto use the VC. Hence, we hypothesize:

H1a. Perceived individual benefits are positively related to users'continuance intention to consume in the VC.

H1b. Perceived individual benefits are positively related to users'continuance intention to provide in the VC.

Fig. 1. The resea


User satisfaction is an individual's emotional or psychologicalstate following VC usage experiences [9,22]. Based on both the ISpost-adoption and IS Success research, user satisfaction is a salientattitude factor that impacts the intention to continue using a system.Maintaining user satisfaction is an important aspect of VC sustain-ability [11]. If a user is not satisfied with a VC, he can freely decreasehis participation or terminate his membership and switch to anotherVC that can provide better resources [32]. Hence, we hypothesize:

H2a. User satisfaction with the VC is positively related to users' con-tinuance intention to consume in the VC.

H2b. User satisfaction with the VC is positively related to users' con-tinuance intention to provide in the VC.

According to the IS post-adoption research, individual benefitswill positively influence their post-adoption attitudes toward the sys-tem. In fact, positive perceptions about individual benefits enhanceuser satisfaction and motivate them to continue participating [58].Therefore, we posit:

H3. Perceived individual benefits are positively related to user satis-faction with the VC.

4.2. The role of information quality and system quality

Perceived information quality is an individual's evaluation of thesystem's performance in providing information based on his experienceof using the system [51,52]. Gu et al. [32] pointed out that low-qualityinformation is distracting because it increases users' search andinformation-processing costs. Users could waste their time and effortson reading useless posts. Outdated posts make it more difficult forusers to find valuable information. Information from an unreliable orcommercial-based source may be biased and may mislead a discussion.In fact, userswill benefit fromVC participation onlywhen a VC providesinformation valued by them [13]. High-quality posts and discussionwillhelp users have a better understanding of the topic, feel support fromothers and make a better decision [72]. High-quality information bene-fits not only userswhowant to obtain useful information and get adviceon a particular topic, but also users who provide information. For exam-ple, with valuable information, a user can help more people who needinformation and increase his reputation and personal image in the com-munity [14]. Thus, we believe that information quality plays an impor-tant role in creating various benefits to users. As user valuation of aninformation-exchange VC depends largely on the quality of the posts,user satisfaction is expected to increase [32]. To this end, we posit:

H4a. Perceived information quality of the VC is positively related toperceived individual benefits.

H4b. Perceived information quality of the VC is positively related touser satisfaction with the VC.

Perceived system quality represents an individual's evaluation of theperformance of system features based on his experience of using thesystem [51,52]. According to the IS Success model, system quality isan important factor in evaluating IS success. Hong et al. [33] suggestedthat technology is at the core of post-adoption intentions and behaviors.Markus [50] argued that technological features are critical in supporting

rch model.


image of Fig.�1


Table 1Respondent demographics.

Site A (N=281)

Gender Tenure Education Hours/week

Male 40.9% First time 0.4% High school 8.5% b5 h 53.7%Female 59.1% b6 months 10.0% Vocational 5.7% 5–10 h 28.1%Age b1 year 6.4% Some college 24.2% 10–20 h 12.5%18–30 10.0% 1–2 years 21.4% Bachelor 33.5% >20 h 5.7%31–40 39.5% 2–5 years 53.4% Master or PhD 24.9%41–60 26.3% >5 years 8.5% Others 3.2%>60 14.2%

Table 2Proposed dimensions of information quality in VCs.

Dimension Definition References

Reliability The extent to which information is regarded as true,believable and credible

[50,64,68]

Objectivity The extent to which information is unbiased,unprejudiced, and impartial

[42]

Value-added The extent to which information is beneficial andprovides advantages from its use

[41,51]

Timeliness The extent to which information is sufficientlyup-to-date for the task at hand

[41,42]

Richness The extent to which information is enough for fulfilling [42,51]


user interaction online, and further the success of a VC. Indeed, a VCis expected to design the system so that users are able to efficiently and -effectively access information and participate in group discussion [11,48].To minimize information overload, a VC should provide clear and well-organized navigation and search tools that allow users to easily locateinformation or post messages [32]. Filtering techniques can also beimplemented, giving usersflexibility of blocking junk information andfil-tering out irrelevant information [36]. A VC may also implement rewardmechanisms to increase users' visibility and acknowledge their contribu-tions, motivating them to interact and contribute [35,48]. When thesystemmakes participation easy, enjoyable and appreciated by the com-munity, users are more likely to maximize their benefits of participationand be satisfied with a VC. Thus, we posit:

H5a. Perceived systemquality of theVC is positively related to perceivedindividual benefits.

H5b. Perceived system quality of the VC is positively related to usersatisfaction with the VC.

4.3. Control variables

Consistent with prior research, a number of control variables suchas age, gender, education, tenure, and average hours spent per weekwere included to control for user heterogeneity [44,48,60,61,62].

5. Research methodology

5.1. Data collection

Weused aweb-based survey to validate themodel. Four VCmanagersand several VC researchers reviewed the initial survey prior to datacollection. A pilot study was conducted in a sports VC and a graduate-student VC. Based on 54 responses, we revised the survey to ensurethat the items could be correctly understood.

After finalizing the survey based on expert comments and thepilot study, we collected data from existing users in a large travelVC (Site A3). Site A is one of the largest VCs for travelers to shareworld-wide travel-related information or experiences in its travelforums. Users are free to participate without any incentives fromthe community (e.g., financial). Based on our agreement with them,a manager who was responsible for daily operations posted the sur-vey as a sticky link at the top of each discussion forum for two daysin 2009. 284 users participated in the survey with 3 responses ex-cluded from the analysis due to missing data. The total sample sizeis 281. Table 1 presents the demographic profile of the respondents.

5.2. Instrument development

Whenever possible, survey items were adopted from prior re-search using a 5-point Likert scale (Appendix A). Several items werenewly developed based on relevant literature.

As prior empirical studies examined information quality as a multi-faceted construct in a variety of contexts [1,51,52], perceived informationquality is modeled as a formative construct that includes six dimensionsto be examined in the VC context (Table 2) [56]. As a formativeconstruct, its indicators form or cause the creation or change in the con-struct and a change in an indicator does not imply a similar directionalchange for the other indicators [17]. In our study, as each quality dimen-sion forms a portion of the overall quality, it is appropriate to modelquality as formative. All the quality dimensions are first-order as theyhave measurable indicators. Perceived information quality is a second-order construct as it is formed by first-order quality dimensions and

3 We cannot reveal the name of Site A, due to a confidentiality agreement with thecommunity managers.


does not have direct measurable indicators. All of the dimensionsexcept richness are reflective. The richness dimension is formative cap-turing information volume, diversity and depth. These three featuresof richness are complementary and do not mean the same thing.

System quality is another quality dimension, which represents howusers evaluate their interaction with IT features of the system [52].Similar with information quality, system quality is multifaceted innature [69,70]. Based on the literature on system and website quality[48,51], we identified 5 dimensions that form the second-order con-struct perceived system quality (Table 3). All of the dimensions exceptinteractivity aremodeled as reflective first-order constructs. Interactivitywas newly developed to capture various interactive IT features in VCs.The use of one feature does not imply the use of another one. Therefore,the features jointly determine interactivity of a VC, which is modeled asa formative construct.

Perceived individual benefits are proposed as a formative construct,which is composed of four benefits: information, social, visibility, and al-truistic benefits. There are a number of studies that have explored theben-efits of participating in VCs [14,48,65,66]. However, studies empiricallyvalidating and testing those benefits are limited. To ensure face validityof the perceived individual benefits construct, we conducted an extensiveliterature review and consulted with senior VC researchers and man-agers. Continuance intention to consume and continuance intention toprovide are both reflective constructs. Respondents were asked to indi-cate their intention to continue consuming information through be-haviors such as browsing, seeking information and posting questions.Similarly, respondents were asked to indicate their intention to continueproviding information such as posting messages, initiating or partici-pating in topic discussions. User satisfaction is a reflective construct mea-sured by validated items [9,22]. All control variables are measured bysingle-item questions.

6. Data analysis and results

We used SPSS to prescreen the dataset and did not find univariatenormality, linearity ormulticollinearity problems. As a second generationof multivariate analysis, structural equationmodeling (SEM) provides uswith the flexibility to model multiple predictors, construct unobserved

a specific needFormat The extent to which information is presented in a way

that is easy to understand[42]



Table 3Proposed dimensions of system quality in VCs.

Dimension Definition References

Navigation The degree to which a user can easily go back and forthwithin the VC

[51,70]

Accessibility The degree to which the system and the information itcontains can be accessed with relatively low effort

[42,51,52]

Appearance The degree to which the system is visually attractiveand appropriate

[51]

Security The extent to which a user's information is protectedand access to personal information is restricted

[70]

Interactivity The degree to which the system can facilitate directuser interaction in an VC

[48,51,53]


latent variables, model measurement errors for observed variables andstatistically test theoretical assumptions against empirical data [19].The Partial Least Square (PLS)-based SEM techniquewas used to validatethemodel as themodel contains both formative and reflective constructsand violates the assumption of multivariate normality [17].

PLS requires a sample size with at least 10 times the largest numberof indicators of the construct in the model [69]. For the hypothesizedmodel, both perceived information quality and perceived systemqualityhad the most indicators, with 17 respectively. Thus, the minimumsample size was 170. The sample size for this model was 281, whichexceeded the minimum requirement.

We tested non-response bias. Similar to Ma and Agarwal [48], thefirst analysis was to compare means for all major and control variablesfor early and late respondents, and the difference was not statisticallysignificant based on the t-test. Furthermore, to ensure the representa-tiveness of the sample, we randomly selected 281 forum users fromthe VC. Based on the publicly available user profile data, we found nosignificant differences (at the 1% level) in tenure between respondentsand randomly-selected VC users.

6.1. Measurement model

We ran a confirmatory factory analysis in SmartPLS 2.0 andassessed reliability, convergent validity and discriminant validity forthe reflective constructs [59]. We also followed the procedure usedby Petter et al. [56] to validate the formative constructs: richness,interactivity and perceived individual benefits.

Reliability is evaluated by computing AVE (Average VarianceExtracted), CR (Composite Reliability) and Cronbach's alpha [5]. Thegeneral acceptable cut-off values are 0.50 for AVE, and 0.70 for bothCR and Cronbach's alpha [4,28]. In Table 4, although two of them haveCronbach's Alpha slightly lower than 0.7, all the AVEs and CRs areabove the cut-off values, indicating that all of the reflective items are re-liable. Convergent validity reflects the extent to which the items foreach construct are measuring the same construct. We use 0.60 as thecut-off value for convergent validity [17,39]. Table 4 shows that all theitems meet this requirement and are significant at the 1% level.

Discriminant validity reflects the extent to which constructs are sig-nificantly different from each other. It is assessed by examining if thecorrelation between a pair of constructs is less than the square root ofAVE of each construct [17,28]. Table 5 shows that all of the squareroots of AVEs on the main diagonal are greater than the pairwise corre-lations between constructs on the off diagonal, implying that all con-structs are distinct. In addition, we checked item cross-loadings basedon the SmartPLS results. Each item loads higher on its designatedconstruct than other constructs and the cross-loading differences arehigher than the suggested threshold of 0.10 [30,34]. As a result, thereare no severe cross-loading problems in the items.

Furthermore, there are three formative constructs in our model:richness, interactivity and perceived individual benefits. We validatedthem by following the guidelines by Petter et al. [56]. Reliability ischecked by item correlations and variance inflation factor (VIF). Unlike


reflective constructs, we do not expect high correlations among items ofa formative construct. We ran two correlations: one among the forma-tive items for each construct; the other between these formativeitems with other reflective items. Table 6 shows that the correlationamong each pair of items is low for each construct (0.3–0.6). Thehighest correlations between Rich1, Rich2, Rich3, Inte1, Inte2, Inte3,PIB1, PIB2, PIB3, PIB4 and other itemswere around 0.3–0.6 respectively.Since each item accounts for a unique part of the construct, we do notexpect multicollinearity of a formative construct. From Table 6, thereis no multicollinearity problem as all the VIFs are less than 3.3 [25]. Tocheck the construct validity, we examined the item weights. Table 4shows that the itemweights for richness, interactivity and perceived indi-vidual benefits are significant at the 5% level. Therefore, all the formativeconstructs are reliable and valid.

6.2. Structural model

SmartPLS 2.0 was used to test the structural model and hypothe-ses [59]. A bootstrapping procedure with 500 iterations was performedto examine the statistical significance of the weights of sub-constructsand the path coefficients [17]. As PLS does not generate overall good-ness of fit indices, the R2 is the primary way to evaluate the explanatorypower of the model [66].

The results in Table 7 indicate that all dimensions of the foursecond-order constructs are significant at the 1% level. Based on theweights, information reliability and format are the two most impor-tant dimensions of information quality perceived by users. In termsof system quality, users are more concerned about site navigationand security of personal information.

Fig. 2 below summarizes the results of the path analysis. With allthe paths significant at the 1% level, the hypothesized relationshipsare supported. User satisfaction and perceived individual benefitsjointly explain 61.0% of the variance in the continuance intention toconsume information, and 41.7% of the variance in the continuanceintention to provide information. As hypothesized, information qual-ity and system quality are the two direct factors that affect users' per-ceived individual benefits (R2=44.5%) obtained from participating inSite A. User satisfaction (R2=69.6%) depends largely on perceptionsof information quality, system quality and individual benefits. Forthe control variables, continuance intention to consume is positivelyaffected by tenure. Interestingly, continuance intention to provide isnegatively related to age, gender and education while it is positivelyrelated to average hours spent per week.

In regard to model validity, Chin [18] suggested that we use R2

values of 0.67, 0.33, or 0.19 to describe the endogenous latent variablesas substantial, moderate or weak respectively. Accordingly, user satis-faction (R2=0.696) is described as strong, while perceived individualbenefits (R2=0.445), continuance intention to consume (R2=0.610)and continuance intention to provide (R2=0.417) as moderate.

6.3. Common method variance

Common method variance (CMV) refers to “variance that is attribut-able to the measurement method rather than to the construct of interest”[57: pp. 879]. CMVmay exist due to the single surveymethod used to col-lect responses. We addressed its potential threat by following Podsakoffet al.'s [57] guidelines. At the design stage of the study, we invited fourVC managers to review the survey and revised the items based on theircomments. The survey items were presented in a random order and acouple of reverse-coded items were incorporated to reduce the potentialimpact of CMV [47]. At the data analysis stage of the study, we appliedthree statistical techniques to control CMV. The Harman's one-factor testindicated that there wasmore than one factor that accounted for thema-jority of covariance [57]. In the partial correlation procedures, the structuralmodel was shown not to be affected greatly with a slightly increased inR2 after a general factor was added into the model [57,67]. Finally, the



Table 4Reliability, internal consistency and convergent validity.

Second-order construct First-orderconstruct

AVE (average varianceextracted)

CR (compositereliability)

Cronbach'salpha

Items Parameterestimatea

T value

Perceived information quality Reliability 0.57 0.84 0.75 Reli1 0.78⁎⁎⁎ 19.92Reli2 0.78⁎⁎⁎ 21.71Reli3 0.64⁎⁎⁎ 13.48Reli4 0.81⁎⁎⁎ 27.51

Objectivity 0.77 0.87 0.70 Obje1 0.87⁎⁎⁎ 39.13Obje2 0.88⁎⁎⁎ 36.59

Value-added 0.79 0.88 0.73 Usef1 0.89⁎⁎⁎ 46.59Usef2 0.89⁎⁎⁎ 27.78

Timeliness 0.71 0.83 0.59 Time1 0.82⁎⁎⁎ 25.18Time2 0.86⁎⁎⁎ 31.79

Richnessb N.A. N.A. N.A. Rich1 0.46⁎⁎⁎ 5.04Rich2 0.48⁎⁎⁎ 5.86Rich3 0.50⁎⁎⁎ 7.03

Format 0.57 0.84 0.74 Form1 0.82⁎⁎⁎ 38.65Form2 0.67⁎⁎⁎ 9.16Form3 0.70⁎⁎⁎ 14.44Form4 0.81⁎⁎⁎ 30.76

Perceived system quality Navigation 0.66 0.91 0.87 Navi1 0.86⁎⁎⁎ 43.21Navi2 0.77⁎⁎⁎ 22.68Navi3 0.87⁎⁎⁎ 45.22Navi4 0.84⁎⁎⁎ 44.40Navi5 0.71⁎⁎⁎ 17.31

Accessibility 0.56 0.79 0.61 Acce1 0.81⁎⁎⁎ 32.35Acce2 0.74⁎⁎⁎ 18.77Acce3 0.70⁎⁎⁎ 15.96

Appearance 0.62 0.83 0.70 Appe1 0.82⁎⁎⁎ 30.90Appe2 0.72⁎⁎⁎ 28.81Appe3 0.81⁎⁎⁎ 15.14

Security 0.63 0.84 0.71 Secu1 0.81⁎⁎⁎ 25.45Secu2 0.83⁎⁎⁎ 33.52Secu3 0.74⁎⁎⁎ 17.18

Interactivity N.A. N.A. N.A. Inte1 0.41⁎⁎⁎ 6.59Inte2 0.32⁎⁎⁎ 5.19Inte3 0.62⁎⁎⁎ 9.41

Perceived individual benefits N.A. N.A. N.A. NB1 0.20⁎⁎⁎ 7.11NB2 0.56⁎⁎⁎ 4.46NB3 0.33⁎⁎ 2.35NB4 0.17⁎⁎⁎ 2.71

User satisfaction 0.90 0.97 0.95 Satis1 0.96⁎⁎⁎ 147.73Satis2 0.95⁎⁎⁎ 100.42Satis3 0.94⁎⁎⁎ 98.25

Continuance intention to consume 0.85 0.94 0.91 InfoCon1 0.94⁎⁎⁎ 74.77InfoCon2 0.95⁎⁎⁎ 99.83InfoCon3 0.87⁎⁎⁎ 38.36

Continuance intention to provide 0.83 0.94 0.90 InfoPro1 0.91⁎⁎⁎ 61.06InfoPro2 0.89⁎⁎⁎ 52.25InfoPro3 0.94⁎⁎⁎ 83.37

a We report factor loadings for reflective constructs and weights for each item of the formative constructs: Richness, Interactivity, and Perceived Individual Benefits.b Richness, Interactivity and Perceived Individual Benefits are formative constructs. According to the literature on the formative construct, internal consistency and reliability check

is not important and necessary (Petter et al. [55]).⁎⁎⁎ Significant at the 1% level of significance.⁎⁎ Significant at the 5% level of significance.


marker-variable technique indicated the low percentage (1/147=0.68%)of the significance change to variable correlations when adjusted forCMV [49]. The analysis showed that CMV is not a big concern in this study.

6.4. Mediating effects

We followed Baron and Kenny's procedures [6] to test two mediat-ing effects: 1. the mediating role of perceived net benefits and user sat-isfaction and 2. the mediating role of user satisfaction. In the firstanalysis, we removed perceived individual benefits and user satisfac-tion from themodel and tested the direct impacts of the two quality di-mensions on two continuance intentions. The second analysis includedboth direct and mediated paths. Table 8 indicates that there are signifi-cant direct relationships between the two quality dimensions and thetwo intentions. However, the relationships were fully mediated whenperceived individual benefits and user satisfaction were added to themodel, as path coefficients in themediatedmodel became insignificant.


Our results indicate that IS quality plays an indirect role in thepost-adoption via affective characteristics (e.g., individual benefits anduser satisfaction). Similarly, the results of the mediating effect 2 showthat user satisfaction partially mediates the relationship between netbenefits and the two continuance intentions.

7. Discussion

Our study proposes a model to understand VC users' continuedparticipation from a quality perspective. We offer several insights onpost-adoption, system design and management in VCs.

7.1. Users' continuance intention to participate in VCs for informationexchange

We captured users' intention to participate in a VC by both informa-tion consumption and provision. There are similarities and differences



Table 5Discriminant validity of the key constructs.

Mean S.D. AVE Re Ob Va Ti Ri Fo Na Ac Ap Se In PIB US CICI

CIPI

Reliability 3.41 0.57 0.57 0.751

Objectivity 2.98 0.83 0.77 0.54 0.88Value-added 4.15 0.58 0.79 0.49 0.32 0.89Timeliness 3.90 0.65 0.71 0.50 0.27 0.52 0.84Richness 3.86 0.57 NA 0.52 0.28 0.53 0.53 NAFormat 3.64 0.63 0.74 0.59 0.50 0.33 0.42 0.55 0.86Navigation 3.62 0.70 0.66 0.36 0.24 0.29 0.37 0.45 0.47 0.81Accessibility 3.65 0.60 0.56 0.35 0.23 0.34 0.37 0.44 0.45 0.71 0.75Appearance 3.53 0.76 0.62 0.33 0.24 0.28 0.39 0.41 0.46 0.71 0.63 0.79Security 3.74 0.58 0.63 0.48 0.33 0.32 0.35 0.47 0.39 0.48 0.48 0.39 0.79Interactivity 3.72 0.61 NA 0.32 0.28 0.35 0.39 0.51 0.46 0.61 0.53 0.57 0.49 NAPerceived individual benefits 4.07 0.64 NA 0.50 0.31 0.62 0.43 0.53 0.40 0.44 0.46 0.38 0.55 0.48 NAUser satisfaction 4.11 0.77 0.90 0.58 0.37 0.59 0.48 0.57 0.52 0.53 0.53 0.50 0.59 0.50 0.76 0.95Continuance intention to consume 4.41 0.67 0.84 0.45 0.28 0.52 0.36 0.47 0.39 0.37 0.38 0.32 0.52 0.45 0.73 0.72 0.92Continuance intention to provide 4.06 0.84 0.83 0.35 0.20 0.35 0.27 0.35 0.32 0.26 0.32 0.24 0.44 0.30 0.54 0.58 0.68 0.91

1. The square root of AVE of every multi-item construct (first-order and second-order) is shown on the main diagonal.


between factors motivating those two types of participation behaviors.We find that both continuance intentions are directly determined byperceived individual benefits and user satisfaction. When users receivevarious benefits from a VC, their attachment to the community could bestronger, whichwill encourage their continued involvement and partic-ipation. For users who consume information, they will be more likely tocontinue seeking information from the VC when their personal needsare met. In addition, the results indicated that users are willing toshare and contribute information to help others in return for receivingbenefits from the community. So, individual benefits and user satisfac-tion facilitate two-way participation.

However, we found that user characteristics affect information con-sumption and provision differently. The longer a user is part of a VC, themore likely he will continue using the site for information seeking. Itcould be that the user is used to the system and the site becomes hisfirst choice for information seeking. The more hours a user spends perweek in a VC, themore likely he tends to provide information, implyingthat providing information takes time. Younger users, perhaps due totheir familiarity with Internet technologies and social media, are morewilling to contribute than older users. Males participate more activelyin topic discussion than females, which may indicate that males aremore interested in interacting with others about travel-related topics.Interestingly, better-educated users are less likely to provide informa-tion than less-educated ones. One possible explanation is that they aremore knowledgeable and aware of online security and privacy, sothey are more cautious about sharing personal experience online.

Moreover, our study suggests that the functional robustness (infor-mation and system quality) of the system needs to be taken into ac-count in the post-adoption stage. However, perceived information

Table 6Reliability of the formative constructs.

Richness Interactivity

Correlation VIF Correlation VIF

Rich1 Rich2 Rich3 Inte1 Inte2 Inte3

Rich1 1.00 1.078 Inte1 1.00 1.135Rich2 0.20 1.00 1.098 Inte2 0.30 1.00 1.195Rich3 0.23 0.26 1.00 1.111 Inte4 0.26 0.34 1.00 1.162

Perceived individual benefits

Correlation VIF

PIB1 PIB2 PIB3 PIB4

PIB1 1.00 1.746PIB2 0.63 1.00 1.811PIB3 0.46 0.48 1.00 1.580PIB4 0.30 0.36 0.48 1.00 1.339


benefits and user satisfaction fully mediate the impacts of informationand system quality on continuance intention, suggesting that infor-mation quality and system quality play a fundamental role in post-adoption stage through individual benefits and user attitudes, NOT adirect role. This finding helps justify the validity of integrating the ISpost-adoption research and IS Success model.

7.2. The importance of information quality

Our findings indicate that a VC with high-quality user-generatedcontents enables users to access superior resources, facilitates infor-mation communication and improves the effectiveness of interaction.Information that is sufficient, relevant, valuable, detailed and fromcredible sources enables users to obtain various kinds of benefits.

In terms of dimensions of information quality, our results are similarwith McKinney et al.'s study [51]. Information reliability and format arethe twomost important dimensions. Information reliability becomes theprimary concern because informationmainly comes from strangerswhowould be difficult for users to trust and rely on. In a travel-relatedVC likeSite A, users would like to see comments from real travelers, instead of aperson who posts comments simply for commercial purposes. One re-spondent mentioned that information will be more reliable and trust-worthy when the VC eliminates commercials. So, VCs need to monitorthe source of posts to ensure that posts are created by ordinary users.Format is the second most important dimension. With a huge amountof information available online, information should be formatted in away that is easy to read, so that users can quickly understand and pro-cess information for their own purposes. It is interesting to find that ob-jectivity is the least important dimension. This could be attributed to thefact that users are free to express their own opinions based on personalexperiences in aVC. It is understandable that their comments are subjec-tive to their own judgments and experiences. As a result, users care less

Table 7Weights of the first-order constructs on the designated second-order constructs.

Second-order construct First-order construct Weight T value

Perceived information quality Reliability 0.315⁎⁎⁎ 16.88Objectivity 0.148⁎⁎⁎ 11.67Value-added 0.215⁎⁎⁎ 10.79Timeliness 0.164⁎⁎⁎ 9.63Richness 0.193⁎⁎⁎ 12.26Format 0.287⁎⁎⁎ 11.24

Perceived system quality Navigation 0.417⁎⁎⁎ 22.09Accessibility 0.199⁎⁎⁎ 17.10Appearance 0.204⁎⁎⁎ 14.18Security 0.220⁎⁎⁎ 12.77Interactivity 0.180⁎⁎⁎ 15.12

⁎⁎⁎ Significant at the 1% level of significance.



Fig. 2. PLS structural results.


about objectivity of the comments. This dimension reflects the specialfeatures of VCs.

7.3. The importance of system quality

System quality matters in terms of how the system can be effectivelydesigned andmanaged to enhance user experienceswith the VC. In termsof dimensions of systemquality, navigation is rated as themost importantdimension. This result provides empirical support for Jones et al.'s study[36]. In order to alleviate information overload, effective navigationtools are necessary to help them easily identify information. The resultsconfirm prior research on system navigation as a critical indicator ofwebsite success [51]. Security is the second most important dimensionof system quality. Although a VC provides a free platform for strangersto communicate, it should implement controlmechanisms to protect per-sonal information, which makes users feel comfortable when interactingwith others. For example, a VC may authorize user access to the profilesof others based on user type (e.g., regular vs. premium member). Sinceonline security is directly related to trust, it is critical for a VC to addressusers' concerns about privacy in order to encourage their active andcontinued participation [70]. Interestingly, our analysis shows that inter-activity is the least important dimension. On one hand, the significanceof the interactivity dimension indicates that users would like to know

Table 8Path coefficients of mediating effects.

Mediating effect 1

Non-mediated model Mediated model (perceivedindividual benefits and usersatisfaction as mediators)

Continuanceintention toconsume

Continuanceintention toprovide



Perceivedinformationquality

0.409⁎⁎⁎ 0.305⁎⁎⁎ 0.047 −0.028

Perceivedsystemquality

0.248⁎⁎⁎ 0.198⁎⁎ −0.023 −0.045

Mediating effect 2

Non-mediated model Mediated model (usersatisfaction as the mediator)





Perceivedindividualbenefits

0.739⁎⁎⁎ 0.642⁎⁎⁎ 0.449⁎⁎⁎ 0.448⁎⁎⁎

⁎⁎⁎ Significant at the 1% level of significance.⁎⁎ Significant at the 5% level of significance.


each other (e.g., user profile and past contributions), as a better under-standing of users in the discussion group may facilitate the effectivenessof information sharing. On the other hand, it has the lowest weightamong all the system dimensions. Users join information-exchange VCsmainly to consume and provide information via reading and writingposts, thus their personal online interaction might be limited.

Our model suggests that information quality and system quality arethe two complementary quality dimensions determining users' contin-uance intention. Based on the results, information quality isweighted asslightly more influential than system quality. Although this finding isconsistent with Agarwal and Venkatesh's work [2], we need to takeboth factors into account when we evaluate the value of a VC and usersatisfaction with it.

7.4. Theoretical contributions

This study adds to the current work with a particular focus on thequality of both user-generated contents and the system itself ininformation-exchange VCs. We found that IS quality matters in the ISpost-adoption stage. In other words, higher quality facilitates the trans-formation of VC resources to individual benefits and increases users'positive attitudes toward the community [13], leading to users' inten-tion to continue using the VC. Second, our model indicates that theimpacts of information and systemquality are fullymediated by benefitsanduser satisfaction. Comparedwith prior IS success studies thatmainlyfocused on the direct relationships among success dimensions, ourstudy actually examined the mediating effects among success dimen-sions and highlighted how quality plays a role in the IS post-adoptionframework. Furthermore, modeling quality as a multi-dimensional con-struct allows us to pay attention to the specific quality concerns by users.

This study contributes to the literature by integrating two researchstreams as the theoretical lens: IS post-adoption and IS Success research.Traditional IS post-adoption research emphasizes the effects of affectivecharacteristics of the system (e.g., perceived usefulness and user satisfac-tion) in the post-adoption stage [9,37,39]. However, the IS post-adoptionresearch does not specifically discuss the role of quality. Hence, integrat-ing the IS Success model that pinpoints the primary role of quality intothe post-adoption research provides us a better theoretical foundationfor themodelwith a particular focus onquality. The integration highlightsthe functional characteristics of the VC (e.g., information and the system)as the fundamental factors in the post-adoption stage of system usage.

To the best of our knowledge, this is also one of the first IS studiesthat differentiate and investigate two types of user continuance in-tention in VCs for information exchange. Given that both consumingand providing are socially desirable behaviors in VCs, separating thesetwo behaviors allows us to think in depth from different perspectivesand provide better rationale for each behavioral intention.

Last but not least, this study proposes and empirically incorporatesfour types of individual benefits in VCs to capture users' evaluation ofthe system performance of a VC: information, social, visibility and


image of Fig.�2



altruistic benefits. In order to reflect VCs that exhibit characteristics be-yond those of a traditional IS, this study extends the generic perceivedusefulness to a more nuanced perceived individual benefits construct.

Perceived information quality• ReliabilityInformation is trustworthy.Information is not reliable. (R)1

Information is free of errors.Information comes from reputablesources.

• Richness (formative)Information is not sufficient enough formy needs. (R)Information contains a widerange of topics/subjects.Information is in-depth for a giventopic.

• ObjectivityInformation is biased. (R)Information presents an impartial view.

• FormatInformation is clear in meaning.Information is presented in a consistentform.Information is formatted concisely.

• RelevancyInformation is informative for my needs.Information is valuable for my needs.

• TimelinessInformation is current.Information is out-of-date. (R)

Perceived system quality• NavigationXXX provides tools for me to easilylocate information (e.g., table ofcontents, use of categories,and index).The descriptions for each link are clear.The navigation aids are effective.

• SecurityXXX protects information againstunauthorized access.XXX protects my personal information.We feel safe to participate in XXX.

7.5. Managerial implications

This study offers several practical implications for VC design andman-agement. First, VCs need to set up quality control mechanisms to ensurethe quality of contents. A VC should frequently monitor, filter or removeposts that are from unreliable or biased sources (e.g., commercial ads).The community could also ask users to rate the helpfulness of individualposts and highlight a number of most helpful posts. It could help otherusers to better understand the topic and lead the discussion on track. Sec-ond, VCs need tomake information search efficient by providing effectivenavigational tools, tracking users' past browsing behaviors and makingtopic recommendations or highlighting themost popular topics currentlybeing discussed. Recommendations and a highlight of popular topicsmaybring eye-catching effects to attract users to participate. Third, VCs couldprovide a predefined template for certain posts (e.g., pros and cons of atravel destination, with additional comments). By providing amore orga-nizedway to express opinions, the community allowsusers tomore easilyread,write and follow the discussion. Itmay also encouragemore users tocontribute. Fourth, VCs could also use IT artifacts to acknowledge usercontributions and increase the visibility of active users in the community.Examples include giving users badges (e.g., top contributor and goldmember) based on their contribution history and placing an announce-ment of active users to the whole community. Last but not least, sinceuser tenure andusage frequencymatter, the community needs to regular-ly track users' activities. For example, when users do not participate for along period, the community may send an email telling them what ishappening in the community. It may give users a gentle reminder of thecommunity activities and attract users back to discussion.

We can control how to accessinformation.We can control how fast to gothrough XXX.

• AccessibilityAll elements load quickly(text, graphics, etc.).All loadable elements are visible.The search engines are not powerful. (R)

• Interactivity (formative)XXX allows me to interact with otherusers by various methods(e.g., discussion board, email, blog).XXX allows me to know more aboutother users and their participation(e.g., a user profile, tenure on thegroup, number of postings, etc.).XXX allows me to get feedback from andgive feedback to others regarding thequality of messages (e.g., messagerating).

• AppearanceThe structure of informationpresentation is logical.XXXhas visually attractive screen layout.The layout acrossXXX is not uniform. (R)

Perceived individual benefits (formative) User satisfactionI find useful information that I need mostI enjoy the experience of learningfrom others.Posting messages makes me feelconfident in my expertise.Being able to sharemakesme feel happy.

I'm very satisfied/pleased/delighted withthe overall experience of using XXX.

Continuance intention to consumeI intend to continue: browsing information in XXX/seeking for information/postingquestions in order to seek for help.

Continuance intention to provideI intend to continue: posting messages in reply to other messages/initiating topicdiscussions/participating in discussions.

Control variables

7.6. Limitations and future research

We used continuance intention as our dependent variables, whichis a well-established approach in the IS post-adoption research [9].However, as the survey was anonymous, we could not contact thoserespondents again and track their actual use behaviors. Future re-search could examine actual continued usage in a longitudinal study.

As we validated the model in a large travel VC, some of the resultsmay not be applicable to small VCs. For example, users may be closerwith each other in small VCs and user interactivity might play a moreimportant role in determining the system quality. In addition, thefindings from an information-exchange VC may not be generalizedto other types of VCs, such as social networking sites (e.g., Facebook).Future research could examine how the proposed research model canbe applied and extended to small VCs or different types of VCs inorder to validate our results in a broader context.

Our research is one of the first IS studies to examine individualbenefits of participation in VCs from multiple perspectives. Moreefforts are needed to validate the construct theoretically and method-ologically. Similarly, we developed new scales for interactivity tocapture direct user interaction supported by VCs. It is necessary tofurther refine this system quality dimension in future research.

• Gender A. Male B. Female• Age A. 18–30 B. 31–40 C. 41–60 D. >60• What is the highest level of education you have completed?A. High school or equivalent B. Vocational/technical school (2 year) C. Some collegeD. Bachelor's degree E. Master's or higher degree F. Other (please specify): ______• How many months have you been using XXX (either as a reader or poster)?A. This is my first time. B. b6 months C. b1 yr D. 1–2 yrs E. 2–5 yrs F. >5 yrs• On average, howmany hours per week do you spend on XXX (reading or posting)?A. b5 h B. 5–10 h C. 10–20 h D. >20 h

1R: Reverse-coded item.

8. Concluding remarks

Our study suggests that information and system quality are criticalin retaining existing VC users. We also highlight various quality dimen-sions and individual benefits in the VC context. We hope this study canopen up more research ideas on quality control of user-generated con-tents, system design and individual benefits in VCs.


Acknowledgments

The authors acknowledge valuable suggestions and support fromthe managers of three virtual communities. The authors would alsolike to thank the senior editor, associate editor and anonymous re-viewers for their constructive comments and suggestions throughoutthe review process.

AppendixA. Survey items (5-point Likert scale: 1— strongly disagreeand 5 — strongly agree)




References

[1] R. Agarwal, J. Prasad, The role of innovation characteristics andperceived voluntarinessin the acceptance of information technologies, Decision Sciences 28 (1997) 557–580.

[2] R. Agarwal, V. Venkatesh, Assessing a firm's web presence: a heuristic evaluationprocedure for the measurement of usability, Information Systems Research 13(2002) 168–186.

[3] M.K. Ahuja, J.E. Galvin, Socialization in virtual groups, Journal of Management 29(2003) 161–185.

[4] R.P. Bagozzi, U.M. Dholakia, Intentional social action in virtual communities, Journalof Interactive Marketing 16 (2002) 2–21.

[5] R.P. Bagozzi, Y. Yi, On the evaluation of structural equation models, Journal of theAcademy of Marketing Science 16 (1988) 74–94.

[6] R.M. Baron, D.A. Kenny, The moderator–mediator variable distinction in socialpsychological research: conceptual, strategic and statistical considerations, Journalof Personality and Social Psychology 51 (1986) 1173–1182.

[7] C.B. Bhattacharya, When customers are members: customer retention in paidmembership contexts, Journal of the Academy of Marketing Science 26 (1998)31–44.

[8] C.B. Bhattacharya, R. Hayagreeva, M.A. Glynn, Understanding the bond of identi-fication: an investigation of its correlates among art museum members, Journalof Marketing 59 (1995) 46–57.

[9] A. Bhattacherjee, Understanding information systems continuance: an expectation–confirmation model, MIS Quarterly 25 (2001) 351–370.

[10] A. Bhattacherjee, G. Premkumar, Understanding changes in belief and attitude to-ward information technology usage: a theoretical model and longitudinal test, MISQuarterly 28 (2004) 229–254.

[11] A.L. Blanchard, Definition, Antecedents, and Outcomes of Successful Virtual Com-munities, Encycl. of E-collab, IGI Global, 2007.

[12] A.L. Blanchard, M.L. Markus, The experiences sense of a virtual community: char-acteristics and processes, The Database for Advances in Information Systems 35(2004) 65–79.

[13] B.S. Butler, Membership size, communication activity, and sustainability: a resource-based model of online social structures, Information Systems Research 12 (2001)346–362.

[14] B.S. Butler, L. Sproull, S. Kiesler, R. Kraut, Community effort in online groups: whodoes the work and why? in: S. Weisband (Ed.), Leadership at a Distance, LawrenceErlbaum Associates, New Jersey, 2002.

[15] Y.L. Chen, The factors influencing members' continuance intention in professionalvirtual communities: a longitudinal study, Journal of Information Science 33(2007) 451–467.

[16] C.M.K. Cheung, M.K.O. Lee, N. Rabjohn, The impact of electronic word-of-mouth: theadoption of online opinions in online customer communities, Internet Research 18(2008) 229–247.

[17] W.W. Chin, Issues and opinions on structural equation modeling, MIS Quarterly22 (1998) 7–16.

[18] W.W. Chin, The partial least squares approach to structural equation modeling,in: G.A. Marcoulides (Ed.), Modern Methods for Business Research, LawrenceErlbaum Associates, New Jersey, 1998, pp. 295–358.

[19] W.W. Chin, B.L.Marcolin, P.R.Newsted, Apartial least squares latent variablemodelingapproach for measuring interaction effects: results from a Monte Carlo simulationstudy and an electronic-mail emotion/adoption study, Information Systems Research14 (2003) 189–217.

[20] E.K. Clemons, M.C. Row, Limits to interfirm coordination through informationtechnology: results of a field study in consumer goods packaging distribution,Journal of Management Information Systems 10 (1993) 73–95.

[21] D. Constant, L. Sproull, S. Kiesler, The kindness of strangers: the usefulness ofelectronic weak ties for technical advice, Organization Science 7 (1996) 119–135.

[22] F.D. Davis, R.P. Bagozzi, P.R. Warshaw, User acceptance of computer technology: acomparison of two theoretical models, Management Science 35 (1989) 982–1003.

[23] W.H. DeLone, E.R. McLean, Information systems success: the quest for the depen-dent variable, Information Systems Research 3 (1992) 60–95.

[24] W.H. DeLone, E.R. McLean, The DeLone and McLean model of information systemssuccess: a ten-year update, Journal of Management Information Systems 19 (2003)9–30.

[25] A. Diamantopoulos, H.M.Winklhofer, Index constructionwith formative indicators: analternative to scale development, Journal of Marketing Research 38 (2001) 259–277.

[26] R.M. Emerson, Social exchange theory, Annual Review of Sociology 2 (1976)335–362.

[27] T. Finholt, L. Sproull, Electronic groups atwork, Organization Science 1 (1990) 41–64.[28] C. Fornell, D.F. Larcker, Evaluating structural equation models with unobservable

variables and measurement error, Journal of Marketing Research 18 (1981) 39–50.[29] J. Gallaugher, Challenging the new conventionalwisdomof net commerce strategies,

Communications of the ACM 42 (1999) 22–29.[30] D. Gefen, D. Straub, A practical guide to factorial validity using PLS-Graph: tutorial

and annotated example, Communications of the Association for InformationSystems 16 (2005) 91–109.

[31] T.W. Gruen, J.O. Summers, F. Acito, Relationship marketing activities, commitmentand membership behaviors in professional associations, Journal of Marketing 64(2000) 34–49.

[32] B. Gu, P. Konana, B. Rajagopalan, H.M. Chen, Competition among virtual communitiesand user valuation: the case of investing-related communities, Information SystemsResearch 18 (2007) 68–85.

[33] S. Hong, J.Y.L. Thong, K.Y. Tam, Understanding continued information technologyusage behavior: a comparison of three models in the context of mobile internet,Decision Support Systems 42 (2006) 1819–1834.


[34] J.J.P. Hsieh, A. Rai, M. Keil, Understanding digital inequality: comparing continueduse behavioral models of the socio-economically advantaged and disadvantaged,MIS Quarterly 32 (2008) 97–126.

[35] Q. Jones, Virtual-communities, virtual settlements & cyber-archaeology: a theoreticaloutline, Journal of Computer-Mediated Communication 3 (1997).

[36] Q. Jones, G. Ravid, S. Rafaeli, Information overload and the message dynamicsof online interaction spaces: a theoretical model and empirical exploration, Infor-mation Systems Research 15 (2004) 194–210.

[37] E. Karahanna, D. Straub, N.L. Chervany, Information technology adoption acrosstime: a cross-sectional comparison of pre-adoption and post-adoption beliefs,MIS Quarterly 23 (1999) 183–213.

[38] M. Katz, C. Shapiro, Product introduction with network externalities, The Journalof Industrial Economics 40 (1992) 55–83.

[39] S.S. Kim, J.-Y. Son, Out of dedication or constraint? A dual model of post-adoptionphenomena and its empirical test in the context of online services, MIS Quarterly33 (2009) 49–70.

[40] P. Konana, S. Balasubramanian, The social–economic–psychological model oftechnology adoption and usage: an application to online investing, DecisionSupport Systems 39 (2005) 505–524.

[41] Y.W. Lee, Crafting rules: context-reflective data quality problem solving, Journalof Management Information Systems 20 (2003–2004) 93–119.

[42] Y.W. Lee, D.M. Strong, B.K. Kahn, R.Y. Wang, AIMQ: a methodology for informationquality assessment, Information Management 40 (2002) 133–146.

[43] F.S.L. Lee, D. Vogel, M. Limayem, Virtual community informatics: a review andresearch agenda, Journal of Information Technology and Theory and Application5 (2003) 47–61.

[44] M. Limayem, S.G. Hirt, C.M.K. Cheung, How habit limits the predictive power ofintention: the case of information systems continuance, MIS Quarterly 31 (2007)705–737.

[45] H.F. Lin, Determinants of successful virtual communities: contributions from systemcharacteristics and social factors, Information Management 45 (2008) 522–527.

[46] H.F. Lin, G.G. Lee, Determinants of success for online communities: an empiricalstudy, Behavior and Information Technologies 25 (2006) 479–488.

[47] M.K. Lindell, D.J. Whitney, Accounting for common method variance in cross-sectional research designs, Journal of Applied Psychology 86 (2001) 114–121.

[48] M. Ma, R. Agarwal, Through a glass darkly: information technology design, identityverification and knowledge contribution in online communities, InformationSystems Research 18 (2007) 42–67.

[49] N.K. Malhotra, S.S. Kim, A. Patil, Common method variance in IS research: a com-parison of alternative approaches and a reanalysis of past research, ManagementScience 52 (2006) 1865–1883.

[50] M.L. Markus, Technology-shaping effects of e-collaboration technologies: bugsand features, International Journal of e-Collaborative 1 (2005) 1–23.

[51] V. McKinney, K. Yooh, F.M. Zahedi, The measurement of web-customer satisfaction:an expectation and disconfirmation approach, Information Systems Research 13(2002) 296–315.

[52] R.R. Nelson, P.A. Todd, H.W. Barbara, Antecedents of information and systemquality:an empirical examination within the context of data warehousing, Journal ofManagement Information Systems 21 (2005) 199–235.

[53] J.W. Palmer, Web site usability, design, and performance metrics, InformationSystems Research 13 (2002) 151–167.

[54] J.H. Park, P. Konana, B. Gu, A.C.M. Leung, An investigation of information sharingand seeking behaviors in virtual communities, in: Proceedings of Thirty FirstInternational Conference on Information Systems, St. Louis, MO, 2010.

[55] M. Parthasarathy, A. Bhattacherjee, Understanding post-adoption behavior in thecontext of online services, Information Systems Research 9 (1998) 362–379.

[56] S. Petter, D. Straub, A. Rai, Specifying formative constructs in information systemsresearch, MIS Quarterly 31 (2007) 623–656.

[57] P.M. Podsakoff, S.B. MacKenzie, J.-Y. Lee, N.P. Podsakoff, Common method biasesin behavioral research: a critical review of the literature and recommendedremedies, Journal of Applied Psychology 88 (2003) 879–903.

[58] C. Ridings, D. Gefen, B. Arinze, Psychological barriers: lurker and poster motiva-tion and behavior in online communities, Communications of the Associationfor Information Systems 18 (2006) 329–354.

[59] C.M. Ringle, S. Wende, A. Will, SmartPLS 2.0 (beta), www.smartpls.de2005.[60] M. Srite, E. Karahanna, The role of espoused national cultural values in technology

acceptance, MIS Quarterly 30 (2006) 679–704.[61] V. Venkatesh, M.G. Morris, Why don't men ever stop and ask for directions?

Gender, social influence, and their role in technology acceptance and usagebehavior, MIS Quarterly 24 (2000) 115–140.

[62] V. Venkatesh, M.G. Morris, G.B. Davis, F.D. Davis, User acceptance of informationtechnology: toward a unified view, MIS Quarterly 27 (2003) 425–478.

[63] R.Y. Wang, D.M. Strong, Beyond accuracy: what data quality means to dataconsumers, Journal of Management Information Systems 12 (1996) 5–34.

[64] D. Wang, L. Xu, H.C. Chan, Understanding users' continuance of Facebook: the roleof general and specific computer self-efficacy, in: Proceedings of Twenty NinthInternational Conference on Information Systems, Paris France, 2008.

[65] M.M.Wasko, S. Faraj, It is what one does: why people participate and help others inelectronic communities of practice, The Journal of Strategic Information Systems 9(2000) 155–173.

[66] M.M. Wasko, S. Faraj, Why should we share? Examining social capital and knowl-edge contribution in electronic networks of practice, MIS Quarterly 29 (2005)35–57.

[67] L.J. Williams, J.A. Cote, M.R. Buckley, Lack of method variance in self-reportedaffect and perceptions of work: reality or artifact? Journal of Applied Psychology74 (1989) 462–468.


http://www.smartpls.de


12 Y. Zheng et al. / Decision Support

[68] B.H. Wixom, P.A. Todd, A theoretical integration of user satisfaction and technologyacceptance, Information Systems Research 16 (2005) 85–102.

[69] B.H. Wixom, H.J. Watson, An empirical investigation of the factors affecting datawarehousing success, MIS Quarterly 25 (2001) 17–41.

[70] Z. Yang, S. Cai, Z. Zhou, N. Zhou, Development and validation of an instrument tomeasure user perceived service quality of information presenting web portals,Information Management 42 (2005) 575–589.

[71] Z. Zhang, Feeling the sense of community in social networking usage, IEEE Trans-actions on Engineering Management 57 (2010) 225–239.

[72] W. Zhang, S.A. Watts, Capitalizing on content: information adoption in two onlinecommunities, Journal of the Association for Information Systems 9 (2008) 73–94.

YiMing Zheng is a doctoral student in the Department of Business Information Sys-tems and Operations Management at the University of North Carolina at Charlotte,US. Her research interests are Web 2.0, social media, E-commerce and IT auditing.Her papers have been published and presented in conferences such as InternationalConference on Information Systems (ICIS), American Conference on Information Systems(AMCIS), INFORMS annual meeting, The Workshop on e-Business (WeB), and ThePre-ICIS Workshop on Human–Computer Interaction.


Dr. Kexin Zhao is an Assistant Professor in the Department of Business InformationSystems and Operations Management at the University of North Carolina at Charlotte,US. She received her PhD degree from the University of Illinois at Urbana-Champaign,US. Her research interests are IT standardization, electronic commerce, digital piracy,and virtual communities. Her papers have been published in journals such as DecisionSupport Systems, Electronic Markets, Industrial and Corporate Change, InternationalJournal of Electronic Commerce, and Journal of Management Information Systems.

Dr. Antonis C. Stylianou has over 25 years of experience in computer information sys-tems. Currently, he is professor of management information systems and a member ofthe graduate faculty at the University of North Carolina at Charlotte, US. His industryexperience includes an appointment in the information management department atDuke Energy. Dr. Stylianou has published numerous research articles in ManagementScience, European Journal of Information Systems, Decision Sciences, Information &Management, International Journal of Electronic Commerce, Communications of theACM, and other journals. He is a frequent presenter on the management of informationsystems, and serves as a consultant to organizations. He currently serves as a senioreditor for the Database for Advances in Information Systems journal.

Systems xxx (2012) xxx–xxx



Documents

The impacts of information quality and system quality on users' continuance intention in information-exchange virtual communities: An empirical investigation