Decision Support Systems 50 (2011) 529–538

Contents lists available at ScienceDirect

Decision Support Systems

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

Information exchange in virtual communities under extreme disaster conditions

Yong Lu a,⁎, Dan Yang b

a Information Sciences & Technology, Pennsylvania State University, 76 University Dr., Hazleton, PA 18202, USAb School of Business Administration, Southwestern University of Finance and Economics, Chengdu, 642150, China

⁎ Corresponding author. Tel.: +1 570 245 3024; fax:E-mail addresses: yul14@psu.edu (Y. Lu), yangd@sw

0167-9236/$ – see front matter © 2010 Elsevier B.V. Aldoi:10.1016/j.dss.2010.11.011

a b s t r a c t

a r t i c l e i n f o

Article history:Received 17 January 2010Received in revised form 17 July 2010Accepted 6 November 2010Available online 11 November 2010

Keywords:Social capitalInformation exchangeVirtual communitiesDisaster communicationWenchuan earthquake

This study proposes a model to examine the mechanism by which social capital contributes to informationexchange in virtual communities. A hierarchical model is employed to analyze data from survey and onlinenetworks in the context of a major natural disaster.Results suggest that structural capital has a significant effect on cognitive capital, but it has no effect onrelational capital. Cognitive capital shows a significant effect on relational capital. This research also finds thatstructural capital increases information quantity, whereas relational capital and cognitive capital increaseinformation quality.This study reveals the characteristics of online social networks and then proposes socio-technical designprinciples to address the communication challenges under uncertain emergency.

+1 570 450 3182.ufe.edu.cn (D. Yang).

l rights reserved.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

Recent advances in Internet and communication technologies haveled to the rapid growth of social networking sites. Popular forms ofthese virtual social networks include Twitter, MySpace, Facebook,wikis, blog sites, and discussion forums. Participants utilize thesesocial networks to connect to people with similar social, economic,and political interests, forming various virtual communities incyberspace. There were more than 250 million active users ofFacebook [22] and 100 million users in MySpace [41].

Virtual communities (VCs) are self-organizing, voluntary, andopen participation systems that are created and sustained throughcomputer-mediated communications [57]. VCs are grassroots mediafor daily communication and entertainment. The availability of socialnetwork tools in the hand of ordinary people makes informationexchange potent and dynamic. However, when it comes to disastermanagement and communication, the role of VCs is unclear and notwell-studied. Few studies examine people's online behavior andcommunication pattern in the aftermath of disasters [18].

To improve existing research methods and to create new insightson the dynamic properties of virtual communities, this researchcaptures communication data from an online discussion forum andempirically investigates the relations between social capital andinformation exchange after a major earthquake.

A massive 8.0-magnitude earthquake struck Wenchuan, Sichuanprovince in China on May 12, 2008. This catastrophic disaster killed

nearly 70 000 people and displaced up to 10 million. People tried toreach their families and friends via many communication tools. Manyindividuals also sought support and exchanged information throughonline communities, such as mailing lists, chat rooms, and discussionforums. They shared their personal experiences, uploaded on-sitepictures and videos, and updated rescue progress. Online communi-ties and discussion forums constituted a useful proxy for theunderlying communication networks after the earthquake.

The empathy and shared reflection that brought people togethervia online virtual communities across barriers of time, distance, andculture, was revitalizing during this disaster [62]. Communication anda spirit of compassion can help strengthen any community, online andoffline. The glue that holds communities and other social networkstogether is called social capital [46]. Social capital is a resource thathelps sustain a community [38]. Putnam [47] asserts that social capitalencourages collaboration and cooperation between members ofgroups for their mutual benefits.

There are several theories to examine community response andinformation coordination in disaster environments, such as complexadaptive systems theory [16], sense-making theory [59], andorganizational learning theory [14]. However, these theories do notfully cover the social network functions for information exchangeduring a disaster. Social capital theory can capture the essentialcontent of social support and collaboration in the context of a disaster.

Therefore, we employ the theory of social capital to examine themechanism by which social capital contributes to informationexchange in VCs. Specifically, this research examines the followingtwo sets of relationships: (1) How the three dimensions of socialcapital interplay among themselves; and (2) How each dimension ofsocial capital influences information exchange in a VC.

530 Y. Lu, D. Yang / Decision Support Systems 50 (2011) 529–538

Such inquiry makes two important contributions. First, thisresearch empirically examines the mechanism of which social capitalimpacts information exchange in extreme disaster conditions. Wereveal the communication characteristics and impediments of onlinesocial networks for disaster communication, and then propose socio-technical design strategies to address the communication challengesunder uncertain emergency. Second, few studies examine the internalmechanism of social capital. This study implements a hierarchicalmodel to investigate the interplay relationships among three dimen-sions of social capital.

This article is organized as follows. First, the concepts of socialcapital, VCs, and information exchange are reviewed. Then, a model isdeveloped to examine how social capital influences informationexchange in a VC. This model is tested using survey and Web forumdata. The article concludes by discussing how the empirical findings ofthis study contribute to theory development and improve systemdesign for new social networking sites.

2. Theoretical background

2.1. Social capital

Social capital is defined as “resources embedded in a socialstructure that are accessed and/or mobilized in purposive action”(p. 29) [33]. The essence of social capital is quality social relations,which affects the capacity of people to come together to collectivelyresolve problems they face in common and achieve outcomes ofmutual benefit [35]. Thus, social capital can be understood as aresource for collective action, which may lead to a broad range ofindividual and group outcomes. For individuals, social capital includesaccess to the reciprocal, trusting social connections that contribute tothe processes of getting by or getting ahead. For communities, socialcapital reflects the ability of community members to participate,cooperate, organize, and interact [10].

2.2. Virtual communities

Virtual communities are online social networks in which peopleinteract to share information and knowledge, and engage in socialinteractions. Fernback and Thompson [23] define VCs as a set ofsocial relationships forged in cyberspace through repeated connec-tions within a specified boundary. It is the nature of socialinteractions and the resources embedded within the network thatsustains VCs [12].

Preece [45] suggests that a VC has four components: people, ashared purpose, policies, and computer systems. Balasubramanianand Mahajan [4] list the following five characteristics for a VC: anaggregation of people, rational members, interaction in cyberspacewithout physical collocation, a process of social exchange, and anobjective, identity, or interest shared by members. Several scholarshave studied VCs through the lens of social capital. Wasko andFaraj [56] investigate how individual motivations and social capitalinfluence knowledge contribution in an electronic network ofpractice. Chiu et al. [12] examine the motivations underlyingindividual's knowledge sharing in a professional VC. Robert et al.[48] examine the impact of social capital on knowledge integrationand performance in digitally enabled teams. Wang and Chiang [55]investigate how interactions among participants contribute to thecreation and advancement of social capital in an online auctionsite.

Prior studies primarily examined individual motivations behindknowledge sharing in VCs, and attempted to address the questionwhy people share knowledge and information online. Few studiesempirically examined the internal mechanism of social capital.However, when it comes to disaster management, the role of socialcapital is unclear and not well-studied. Very few studies investigate

information exchange for online social networks under extremedisaster conditions.

2.3. Attributes of information exchange

This study focuses on one specific type of VCs, web discussionforums, where individuals discuss interesting topics and build socialties into networks of relationships. Individuals may choose to act assilent viewers or active contributors to these discussions. In adiscussion forum, individuals are able to engage in knowledgesharing, information inquiry, and learning through posting andresponding to messages.

We chose to study the discussion forums immediately after anatural disaster because the need for information and informationexchange is intensified. One important observation about disasteremergencies is that they increase need for information and also affectcommunication pattern [21]. Sharing quality information duringemergencies is both vital and challenging. When there is a suddenchange of situation or reality in a human community, like a disaster,both the individuals and communities need new and accurateinformation to orient their actions and responses. Social networkscan serve as one of the primary channels whereby individuals andsocial units can acquire reliable, relevant information for planningtheir actions and responses, as well as sharing the information to therest of the community.

Furthermore, valid and timely information exchange is critical inemergency response operations. In the aftermath of any extremeevent or disaster, the required rate of information sharing dramati-cally increases and the quality of the information becomes even morecritical. Base on prior studies, information exchange can be measuredby information quality and information quantity [12,56].

Information quality refers to the quality of content of informa-tion exchanged in a VC. McKinney et al. [37] identify fivedimensions for information quality: relevance, timeliness, reliabil-ity, scope, and perceived usefulness. Chiu et al. [12] measureinformation quality by six items: relevance, ease of understanding,accuracy, completeness, reliability, and timeliness. In this study,we measures information quality by focusing on four dimensions:reliability, accuracy, timeliness, and relevance. These four dimen-sions are most important aspects for information quality in avirtual community.

Information quantity represents the total amount of informa-tion exchanged in one VC. It may also be represented by themessages sent and messages replied to by an individual in a Webforum over a specific time period [12,56]. In this study, we use twotypes of message numbers to represent information quantity: totalnumber of messages initiated by an individual (seed messages)and total number of messages replied by an individual (responsemessages) during the study's period. These two numbers reflectthe quantity of information exchange activities for an individual inthe VC.

3. Research model and hypotheses development

To understand how VCs support emergency and disastercommunication, we will examine each of the three dimensions ofsocial capital that influences the quality and quantity of informationexchange. Based on the literature review, we develop a researchmodel and propose hypotheses (Fig. 1). We implement a hierarchicalmodel for this research according to Wetzels et al. [60]. As shown inFig. 1, structural capital is a second-order construct including onefirst-order construct: social interaction ties. Relational capital is asecond-order construct including two first-order constructs: trustand reciprocity. Cognitive capital is a second order construct

Information

quality

Structural capital

Relational capital

Cognitive capital

Information quantity

H4a

H5a

H5b

H6b

Social interaction

ties

Trust

Reciprocity

Shared language

Shared vision

H2

H3

H4b

H6a

Second-orderFirst-order

H1

Fig. 1. Research model.

531Y. Lu, D. Yang / Decision Support Systems 50 (2011) 529–538

including two first-order constructs: shared language and sharedvision.

3.1. Associations among the three dimensions of social capital

Social capital is a multi-dimensional concept comprising structuralcapital, relational capital, and cognitive capital [38]. Previous studieshave suggested the existence of close interrelationships among thesethree dimensions [38,53].

Structural capital reflects the overall pattern of interactions amongindividuals [38,52]. It is characterized by the centrality, connectivity,and hierarchy of relationships among individuals [15]. The location ofan actor's contacts in a social network provides certain advantages forthe actor. People can use their social contacts to get jobs, to obtaininformation, or to access specific resources [8].

Prior research suggests that one way to measure structural capitalis to determine the number of social interaction ties an individual haswith others in the network [2] Tsai and Ghoshal [26] consider socialinteraction ties as channels for information and resource flows. Amember may gain access to another's resources through their socialinteractions.

The relational dimension of social capital is conceptualized as thenature and quality of the relationships among individuals and howthose relationships affect their behaviors [38]. Relational capitalinvolves social actors trusting other actors within the group and beingwilling to reciprocate favors or other social resources in thecommunity. Trust is a set of specific beliefs related to benevolence,integrity, and reliability with respect to another party [38]. Reciproc-ity refers to the expectation of participants that their informationcontributed will lead to their future requests for information beingmet [29].

Cognitive capital comprises values attitudes, beliefs, andperceptions of support, which enables shared interpretations andmeanings within a group [38]. Cognitive capital is embodied inattributes like a shared language or a shared vision that facilitatesa common understanding of collective goals and proper waysof acting in a social system. Shared language goes beyond thelanguage itself. It also addresses “the acronyms, subtleties, andunderlying assumptions that are the staples of day-to-day inter-

actions” (p. 836) [31]. Shared vision embodies the collective goalsof an organization and is regarded as “a bonding mechanism thathelps different parts of an organization to integrate or to combineresources” (p. 467) [26].

3.1.1. Linking structural capital and relational capitalThe structural dimension of social capital, manifesting as social

interaction ties, may stimulate trust and reciprocity, which representthe relational dimension of social capital.

Previous studies have suggested that trust results from socialinteractions [27]. As two participants interact over time, theirtrusting relationship will become more concrete, and the actors aremore likely to perceive each other as trustworthy. Therefore, aparticipant occupying a central position in a social network is likelyto be perceived as trustworthy by other participants in thenetwork.

Frequent social interaction leads to participants sharing moreinformation with others, and therefore creating more reciprocityrelationship. We expect a member with high structural capital in a VCto be likely to develop and enhance relational capital like trust andreciprocity in other members.

H1. Structural capital is positively associated with relational capital in avirtual community.

3.1.2. Linking structural capital and cognitive capitalStructural capital may also encourage and develop shared

language and shared vision, which represent the cognitive dimensionof social capital. The association between structural capital andcognitive capital relies on the premise that social interaction plays acritical role in sharing and shaping a common set of goals and valuesamong an organization's members.

Previous studies on organizational socialization [54] have high-lighted the importance of social interactions in helping individuals tolearn organizational cultures and values. Through the process of socialinteraction, individuals realize and adopt their organization's lan-guage, codes, beliefs, and visions. Wang and Chiang [55] find anexistence of a positive and significant link between social interactionties and shared vision in an online auction community. In a VC, an

532 Y. Lu, D. Yang / Decision Support Systems 50 (2011) 529–538

individual would be likely to share language and vision with othermembers through their social interactions. Hence,

H2. Structural capital is positively associated with cognitive capital in avirtual community.

3.1.3. Linking cognitive capital and relational capitalShared language and shared value, the major manifestations of the

cognitive dimension of social capital, may also encourage thedevelopment of trusting relationships. A trusting relationship be-tween two parties implies that “common goals and values havebrought and kept them together” [5]. Sitkin and Roth [49] state thattrusting relationships are rooted in “value congruence” – thecompatibility of an individual's value with an organization's values.With a shared language and value, organization members are inclinedto trust one another, as they can expect that they all work for commongoals and will not be hurt by any other member's pursuit of self-interest. Within a community, any member that shares the commu-nity's shared value or language is likely to be perceived as trustworthyby other members in the community.

Shared language and value may encourage the development of thereciprocity relationships. As Ouchi [40] noted, “Common values andbeliefs provide the harmony of interests that erase the possibility ofopportunistic behavior” (p. 138). Language is the way by whichpeople discuss and exchange information. Shared language facilitatespeople to ask questions and conduct business together. Shared valuecan bind the members of a community where people are likely toreturn the benefits they receive from others. With shared languageand value, members tend to respect each other and havemore mutualreciprocity. Therefore, we expect that the following relationshipbetween cognitive capital and relational capital:

H3. Cognitive capital is positively associated with relational capital in avirtual community.

3.2. Social capital and information exchange

3.2.1. Structural capital and information exchangeStructural capital reflects the overall pattern of interactions among

individuals. Social interaction ties are channels for information andresource flow [38]. A participant may gain access to another'sinformation and resources via social interaction ties. Larson [30]found that the more social interactions undertaken by members, thegreater intensity, frequency, and breadth of information exchange.Structural capital and network ties influence “both access to partiesfor combining and exchanging knowledge and anticipation of valuethrough such exchange” (p. 252) [38].

The conversations that occur in VCs are similar to the socialinteractions in face-to-face networks. Posting and responding tomessages create social ties between individuals. Social interaction tiesrepresent the strength of the relationships, the amount of time spent,and communication frequency amongmembers in a VC [12]. Previousstudies have provided empirical support for the positive impact ofsocial interaction ties on knowledge exchange [26] and resourceexchange [52]. Thus, an individual's social interaction ties influencehis or her information exchange in a VC.

H4a. Structural capital is positively related to information quality in avirtual community.

H4b. Structural capital is positively related to information quantity in avirtual community.

3.2.2. Relational capital and information exchangeTwo elements of relational capital, trust and reciprocity, may

influence information exchange. Trust has been viewed as a key factorthat provides a context for cooperation [26] and effective information

and knowledge exchange [1,28]. Trust influences information ex-change in two ways. First, trust allows individuals to rationalize theirdecisions to contribute to a conversation and enables the exchange ofmore useful information. Second, trust enables individuals to freelyexchange information. In particular, higher levels of trust improveinformation contribution by increasing the amount [19] and types ofinformation exchanged [3].

Reciprocity has been highlighted as a benefit for individuals toengage in social exchange [6]. The social exchange theory suggeststhat participants expect mutual reciprocity and that, in turn, justifiestheir expense in terms of time and effort in sharing their knowledge[50]. Therefore, relational capital impacts the quality and quantity ofinformation exchange in a VC. Thus:

H5a. Relational capital is positively related to information quality in avirtual community.

H5b. Relational capital is positively related to information quantity in avirtual community.

3.2.3. Cognitive capital and information exchangeEngaging in a meaningful information exchange requires some

degree of shared understanding among members, such as sharedlanguage and shared vision [38].

Shared language is essential to information exchange in acommunity. It provides an avenue in which individual membersunderstand each other and build a common vocabulary in theirdomains. Therefore, shared language will help motivate the partici-pants to be actively involved in information exchange activities andenhance the quality of information.

Shared values and goals bind the members of a community, makecooperative action possible, and eventually benefit the organization[13]. As a result, members who have a common vision will be morelikely to become partners to exchange their information andknowledge. The common value and vision that members share willincrease the quality and quantity of their exchanged information.Therefore:

H6a. Cognitive capital is positively related to information quality in avirtual community.

H6b. Cognitive capital is positively related to information quantity in avirtual community.

4. Research method and data collection

4.1. Operationalization of constructs

A survey instrument was developed by identifying appropriatemeasurements from a comprehensive literature review. Measure-ment items were adapted from prior articles wherever possible. Someminor modifications were made to the existing scales to make thosemore suitable to this research context. All of the items were measuredon a seven-point Likert scale, ranging from “strongly disagree” to“strongly agree.” Appendix A summarizes the constructs used in thisstudy and their operationalization.

Three items for measuring social interaction ties focus on closerelationships, time interaction, and frequent communication,similar to those applied by Chiu et al. [12]. Trust was adaptedfrom prior studies [12,29], with three items to measure anindividual's beliefs in other member's truthfulness and onlinebehaviors. Reciprocity was also adapted from prior studies [12,29],with three items to measure the fairness of knowledge, helpfulness,and sharing. Shared language was measured with a two-item scaleadapted from Chiu et al. [12]. It focuses on message understand-ability and meaningful communication pattern. Shared vision wasassessed with a three-item scale adapted from Chiu et al. [12]. It

Table 1Loadings of the indicator variables (CR) (AVE) (Cronbach's Alpha).

Construct Indicator Loading CR AVE Cronbach's Alpha

Social interaction ties (SIT) SIT1 0.93 0.95 0.85 0.81SIT2 0.93SIT3 0.91

Trust (TR) TR1 0.82 0.87 0.70 0.78TR2 0.82TR3 0.86

Reciprocity (RE) RE1 0.83 0.88 0.70 0.79RE2 0.87RE3 0.82

Shared language (SL) SL1 0.93 0.94 0.88 0.87SL2 0.95

Shared vision (SV) SV1 0.90 0.91 0.76 0.84SV2 0.87SV3 0.86

Information quality(QUAL)

Qual1 0. 87 0.87 0.63 0.79Qual2 0.85Qual3 0.72Qual4 0.72

Information quantity(QUAN)

Quan1 0.70 0.89 0.81 0.84Quan2 0.72

533Y. Lu, D. Yang / Decision Support Systems 50 (2011) 529–538

refers to same goal and value of information contribution amongmembers. Information quality was assessed with items adaptedfrom Lin [34] and McKinney et al. [37]. These items measured fourattributes of information quality: reliability, accuracy, timeliness,and relevancy. Information quantity was measured with thenumbers of seed messages and response messages for eachindividual.

4.2. Data collection

The context of the data collected for this study is the web forumactivities in the immediate aftermath of a massive earthquake thatstruck Wenchuan, Sichuan province of China on May 12, 2008.Online communities and discussion forums constituted a usefulproxy for the underlying communication networks after theearthquake [62].

The research was conducted using students enrolled in a majoruniversity in Chengdu, which is the capital city of Sichuan Provinceand only 50 mi away from the earthquake's epicenter. Many of thestudents participated in the web forum have families and friendswhose lives were directly impacted by the earthquake. There aretwo types of data collected for this study. One type of data wascollected from a web-based survey conducted for the members ofthis web forum community. The other type of data was collecteddirectly from the server database of the web forum. Two types ofmessages were extracted for each member: total number ofmessages initiated by an individual (seed messages) and totalnumber of messages replied by an individual (response messages)during the study's period.

4.2.1. Survey administrationAfter the survey instrument was developed, we organized a

panel of researchers to examine the questionnaire and assess thecontent validity, such as ease of understanding, sequence ofitems, logical consistence, and contextual relevance. After thereview, we recruited 20 students from the focal university for apilot study to finalize the questionnaire and listened to theircomments on the items. Their comments and suggestions led toseveral minor modifications of items wording and structure ofthe instrument. We reworded these items: two items for sharedlanguage (SL1 and SL2), one items for trust (TR2), and one itemsfor information quality (QUAL4).

This survey asked students to recall their web forum experience inthe first month after the earthquake. By the time this survey wasconcluded, 513 questionnaires were collected, of which 475 weredeemed complete and valid responses for data analysis. The numberof students who visit the web forums was 2500 to 3000 during thestudy period. Thus the estimated response ratio of the web survey isabout 20%.

4.2.2. Forum dataThis web forum for students at the university was chosen for this

study because web forums are widely used by college students tocommunicate about their campus life, social issues, classes, news, andso on. Web forums are one of the important parts of their college life,and they emerged as one of the primary sources of information forstudents in the aftermath of the earthquake.

Communications through the web forum increased dramaticallyafter the earthquake, and lasted for about one month (fromMay 12 toJune 11). After June 11, the discussions about the earthquakedecreased because it was near the end of the semester and studentswere leaving the campus. We collected the forum activity data fromMay 12 to June 11 (one month after the earthquake) from the serverdatabase. There were total 101,838 messages posted that month. Thedaily average number of messages and participants were 3285 and

1173, respectively. This means each participant sent 2.8 messagesdaily in the research period.

5. Results and analysis

5.1. Measurement validation

Partial least square (PLS) is used for measurement validation andtesting the structural model. Unlike a covariance-based structuralequation modeling method such as LIRREL and AMOS, PLS employs acomponent-based approach for estimation, and can handle bothformative and reflective constructs [11]. In general, PLS is better suitedfor explaining complex relationships be avoiding inadmissiblesolutions and factor indeterminacy [24,32]. Hence, we chose PLS toaccommodate the presence of a large number of variables, relation-ships and moderating effects.

The measurement model comprises the research constructs andtheir associated indicators (measures). The quality of the constructsand indicators could be evaluated by assessing the reliabilities and theconvergent and discriminant validities of the research constructs[17,24].

Convergent validity tests the relationships among indicators in thesame construct. It is assessed by examining the correlation (loading)between the indicator and its construct. An indicator is considered tobe reliable if its loading is more than 0.70 [11]. As shown in Table 1,the loadings for all indicators of this study are above 0.70, suggestingadequate convergent validity.

Construct reliability is calculated using composite reliability scoresprovided by PLS. Construct reliability is considered satisfactory if thecomposite reliability scores are higher than 0.80 [39]. All of ourconstructs are considered having adequate reliability, as shown inTable 1.

Cronbach's Alpha is another indicator for construct reliability,often referred to as internal consistency reliability of the constructmeasures. A construct is considered to have adequate internalconsistency reliability if the Cronbach's Alpha is greater than 0.70[25]. As shown in Table 1, all of our constructs demonstrate goodconstruct reliability.

A commonly accepted rule for assessing discriminant validityrequires that the square root of average variance extracted (AVE) islarger than the correlations between the construct and any otherconstruct in the model [11]. As shown in Table 2, all constructs meetthis requirement. Thus, all the constructs in the model displayadequate discriminant validity.

Table 2Correlations of latent variables.(Square roots of the AVE are the bolded diagonal value).

Mean SD SIT TR RE SL SV Quality Quantity

SIT 2.32 1.60 0.92TR 5.22 1.07 0.15 0.84RE 4.41 1.25 −0.05 0.35 0.84SL 5.18 1.04 0.08 0.50 0.37 0.94SV 4.76 1.22 0.11 0.65 0.52 0.44 0.87Quality 4.96 1.11 0.13 0.61 0.48 0.45 0.65 0.79Quantity 2.79 2.09 0.44 0.08 −0.04 0.08 0.01 0.08 0.90

534 Y. Lu, D. Yang / Decision Support Systems 50 (2011) 529–538

5.2. Common method variance

For self-reported data, there is a potential for common methodbiases from multiple sources such as consistency motif and socialdesirability [44]. Steps to safeguard against common method biasinclude the use of different types of measures across constructs anddifferent scale types for key construct measures. We attempted toenforce a procedural remedy by collecting data from two sources: onefrom survey and another from online discussion forum. In addition,we performed statistical analyses to assess the common method bias.

First, a Harmon one-factor test [43] was conducted on the sevenconstructs in our theoretical model including SIT, TR, RE, SL, SV,Quality, and Quantity. Results from this test showed that seven factorsare present and the most covariance explained by one factor is 24.31percent, indicating that common method biases are not a likelycontaminant of our results.

Second, based on Podsakoff et al. [44] and Williams et al. [61], weincluded in the PLS model a common method factor whose indicatorsincluded all the principal constructs' indicators and calculated eachindicator's variances substantively explained by the principal con-struct and by the method. As shown in Appendix B, the resultsdemonstrate that the average substantively explained variance of theindicators is .58, while the averagemethod based variance is .011. Theratio of substantive variance to method variance is about 53:1. Inaddition, most method factor loadings are not significant. Given thesmall magnitude and insignificance of method variance, we contendthat the method is unlikely to be a serious concern for this study.

Structural capital

Relational capital

Cognitive capital

R2=.21

R2=.56

H4a: .04

H

H

H6b: .03

Social interaction

ties

Trust

Reciprocity

Shared language

Shared vision

.84**

.80**

.77**

.91**

H2: .12**

H1: .01

H3: .75**

H4b: .44**

H6a: .37*

Fig. 2. Analysis of results. (Solid lines indicate significant paths. Da

5.3. Hypothesis testing

Fig. 2 shows the results of hypotheses tests. First, we examine theinternal relationships between three dimensions of social capital.Structural capital has a significant, positive effect on cognitive capital,but it has no effect on relational capital. In addition, cognitive capitalshows a significant, positive effect on relational capital.

Second, we examine how each of the three dimensions of socialcapital influences information exchange. Structural capital has apositive and significant influence on information quantity, but it hasno influence on information quality. Relational capital and cognitivecapital have significant and positive impacts on information quality,but they have no significant impacts on information quantity.

In addition to the direct effects, we can also derive the indirecteffects of the four first-order constructs on information exchange.According to Edwards [20], the indirect effect is calculated as theproduct of each first-order construct load and the correspondent pathcoefficient. For example, the loading of trust on relational capital is0.84 and the coefficient of relational capital on information quality is0.38. The indirect effect between trust and information quality is 0.84* 0.38=0.32. Table 3 shows the indirect effects for the four first-orderconstructs: trust, reciprocity, shared language, and shared vision. Allthe four constructs have significant, positive effects on informationquality, but have no effects on information quantity.

To make the results be more robust, we further check two indirectimpacts. First, the indirect impact of structural capital on relationalcapital through cognitive capital is insignificant (β=0.09, pN0.05).Second, indirect effect of structural capital on information qualitythrough relational capital and cognitive capital is insignificant(β=0.08, pN0.05).

6. Discussion

6.1. Summary of results

The aim of this study is to study how VCs support information flowand communication in the context of disasters. We investigated thecomplex mechanism in which social capital influences informationexchange in a VC. First, we found that cognitive capital has asignificant, positive effect on relational capital, but structural capital

Information quality

R2=.51

Information quantity

5a: .38**

5b: .01

*

.03

R2=.20

shed lines indicate non-significant paths.) *pb0.05; **pb0.01.

Table 3Assessing the hierarchical model.

Hierarchical model

Second-order model

Relational capital Cognitive capital

TR 0.84 [0.80, 0.89]a

RE 0.80 [0.75, 0.85]SL 0.77 [0.72, 0.81]SV 0.91 [0.88, 0.95]

Structural model

Information quality Information quantity

TR 0.32⁎⁎ [0.30, 0.35 ] 0.01 [0.00, 0.01]RE 0.30⁎⁎ [0.27, 0.34] 0.01 [0.00, 0.01]SL 0.28⁎⁎ [0.25, 0.33] 0.02 [0.00, 0.03]SV 0.34⁎⁎ [0.30, 0.37] 0.03 [0.00, 0.04]

*pb0.05; **pb0.01.a Percentile estimate of 95% confidence interval.

535Y. Lu, D. Yang / Decision Support Systems 50 (2011) 529–538

has no influence on relational capital. This implies that structuralcapital is not a predicator of relational capital, and senses of trust andreciprocity cannot be derived directly from social interaction ties in aVC. These results suggest that achieving high trust and reciprocityrequires establishing a shared vision and language to ensurealignment between individuals and community values.

Second, a surprising finding is that while structural capital has asignificant and positive influence on information quantity, it hasalmost no influence on information quality. These results are contraryto prior studies on traditional interpersonal communications, where itis shown that individuals who are connected to a large number ofothers are more likely to sustain contributions and provide highquality information [7]. One possible explanation is that VCs haveunique characteristics, such as open environment and anonymousmembership. VCs are voluntary and open participation systems [57].Participants are typically strangers and they usually do not haveestablished personal relationships. Therefore, VCs create weakelectronic links between participants. This characteristic of openparticipation between strangers in VCs is different from interpersonalcommunities.

Additional analysis was performed to link information quantity toinformation quality. Information quantity has no impact on informa-tion quality (β=.03). The result indicates that an individual who,while sending out many messages in a VC, does not necessarilyprovide high quality of information. We submit these results toinformation overload in online communities. The chief challenge forthe emergency response organization is not the scarcity of informa-tion, but the information overload: too many resources and too muchinformation strains the capacity of the management system as well asthe communication system [36]. Due to their open and anonymousnature, online social networking could also be the sources ofintentional or unintentional misinformation, which, especially in theaftermath of a disaster emergency, could jeopardize public safety andthe recovery effort.

Third, relational capital and cognitive capital have significant andpositive impacts on information quality, but have no significantimpacts on information quantity. These results are contrary to priorstudies in face-to-face settings, where results show positive relation-ships between social capital and information quantity, and peoplecommunicate more when they trust each other and have a sharedvision. These differences can be explained by the public goodscharacteristics of VCs. Wasko et al. [56] explain that participation inVCs is a form of collective action, which is exhibited through theinteractive posting and responding of messages. This interactionproduces and maintains the public good of knowledge and informa-tion that is stored and is available to anyone in the community.

In traditional interpersonal communications, messages have to besent to each individual, increasing thequantity of information.Messagesposted in a VC are saved and available to all participants in thecommunity. There is no need to repeat messages with similar topics.Therefore, relational and cognitive capital increases the quality ofinformation but not necessarily increase the quantity of information.

6.2. Implications for research

The components and dimensions of social capital are welldiscussed in the literature of social sciences and IS. What is lessunderstood is how the core elements of social capital interact witheach other. Three key aspects of this study signify our contribution tothe theory of social capital and VCs.

First, this study investigates the interplay relationships among thethree dimensions of social capital. Prior studies focus on an individual'smotivations behind information sharing in VCs, and little researchexamines the internal relationships of social capital. This studyexamines the three relationships: structural capital has a significantimpact on cognitive capital, but not on relational capital; cognitivecapital has a significant impact on relational capital. This study furthersour understanding of the internal attributes of social capital.

Second, this study extends the prior work on the unique character-istics of VCs. Chiu et al. [12] found that structural capital increasedknowledge quantity, but not knowledge quality in a professional VC.Wehave the similar findings in an online discussion forum. Our study alsofound that relational capital and cognitive capital had impacts oninformation quality, but no impacts on information quantity. Wasko etal. [57] explained that VCs have unique characteristics, such asanonymous membership and voluntary participation. The participationin VCs is a form of collective action and has the characteristics of publicgoods. Considering the dramatic growth and popularity of virtual socialnetworks, it is pressing for future research to understand what differentcommunication patterns and relationships between VCs and traditionalinterpersonal communities.

6.3. Implications for practice

Valid and timely information sharing is critical in emergencyresponse operations. Although social networking sites can improveinformation flow and respond quickly to dynamic communicationneeds during disasters [42], they are subject to several impediments,such as information overload and misinformation.

Information overload is a term coinedbyAlvin Toffler [51] that refersto the presence of bewildering amounts of Information, more than canbe effectively absorbed or processed by an individual. Electronic socialnetworks have created a new way of feeding information to people.Because of the existence of having so much information available,people either cannot assimilate it all or feels too overwhelmed [9].

Moreover, in extremedisaster situations, people face the ingestion ofinformation, almost instantaneously, without knowing the validity ofthe content and the risk of misinformation. In our study, there weremore than three thousandsmessages posted on the electronic network.It is really information overload and explosion for each participant.

Therefore, this study calls for socio-technical solutions thataddress the challenges of interoperability, authenticity, usability,and organizational applicability of citizen generated informationunder uncertain emergency. The remainder of this section discusses indetail a set of strategies and design principles.

• System design to overcome information overloadThis study found that one of the drawbacks of social networks is theinformation overload. People face with information explosion in theaftermath of the disaster and need a reliable source to confirminformation and clarify facts. Strategies need to be developed toprovide reliable and relevant messages. One way to accomplish this is

Independentvariables

Item

Structural capitalSocial interactionties (SIT)

SIT1: I maintain close social relationships withsome members in the forum.

SIT2: I spend a lot of time interacting with somemembers in the forum.

SIT3: I have frequent communication with somemembers in the forum.

Relational capitalTrust (TR) TR1: I believe that people in the forum do not use

unauthorized knowledgeTR2: I believe that people in the forum use other's

knowledge appropriately.TR3: Members in the forum are truthful in dealing

with one another.Reciprocity (RE) RE1: I know that other members in the forum will

help me, so it's only fair to help other members.RE2: I believe that members in the forum would

help me if I need it.RE3: When I share my knowledge through the

forum, I believe that my queries for knowledgewill be answered in future.

Cognitive capitalShared language (SL) SL1: Members in the forum use understandable

communication patterns during the discussion.SL2: Members in the forum use understandable

narrative forms to post messages or articles.Shared vision (SV) SV1: Members in the forum share the vision of

helping others solve their problems.SV2: Members in the forum share the same

goal of learning from each other.SV3: Members in the forum share the same

value that helping others is pleasant.Dependent variables

Information quality(Qual)

Qual1: The information exchanged by membersin the forum is reliable.

Qual2: The information exchanged by members inthe forum is accurate.

Qual3: The information exchanged by members inthe forum is timely.

Qual4: The information exchanged by members inthe forum is relevant to the discussion topics.

Informationquantity1 (Quan)

Quan1: Number of messages initiated by an individualQuan2: Number of messages replied by an individual

1The two items for information quantity were transformed using log transformation tofit the normality assumption of PLS analysis. The similar method is used in the studies ofRefs. [12,56].A seven-point scale was used: 1=less than one message; 2=1 to 2 messages; 3=3 to5 messages; 4=5 to 10 messages; 5=10 to 20 messages; 6=20 to 40 messages;7=more than 40 messages.

536 Y. Lu, D. Yang / Decision Support Systems 50 (2011) 529–538

to use moderators who play a proactive role in ensuring quality whilecontrolling quantity. During amajor event,moderators should identifyvaluable and reliable messages from thousands postings and list themon top in order to disseminate important information quickly.

• New systems can manage information accuracy and addressmisinformationDue to their open and anonymous nature, online social networkscould be the sources intentional or unintentional misinformation,which, especially in the aftermath of a natural disaster, could causesignificant harm to the members and the collectives. We also needto be concerned about problematic rumors. New systems need tointegrate multiple disaster databases for validation, require sendersprovide more details for confirmation, and ask moderators on dutyto identify unreliable messages.

• System design to establish trust and reciprocityThis study suggests improving information quality through trustand reciprocity during disasters.Administrators and community leaders should develop strategies tofoster trust and boost reciprocity among its members. Systemdesign should have features to help new members and encouragethem respond to postings of others.

• Establish shared vision and languageEmergency responseoperations arealsomoreeffectivewhenmembershave shared vision and language. Pre-disaster communication andinteraction is a key aspect to establish shared vision and language.Design features could identify leaders for online communities andpromote their influence and articulate their vision and leadership.

6.4. Limitations

Although the findings are encouraging and useful, this researchhas several limitations that require further examination and addi-tional research.

First, structural capital was assessed by each individual's socialinteraction ties in the VC. The research results show social interactionties have a positive influence on information quantity while having noinfluence on information quality. Although other studies [12,26] alsoused social interaction ties as the indicator for structural capital, socialinteraction ties may not be the best indicator for structural capital inVCs. Further research could attempt to explore other indicators ormultiple measures for structural capital, such as degree centrality orbetweenness centrality [58].

In addition, there is room to add other variables to explore how toeffectively use information for disaster management. Future studiescould investigate other factors which are predicted by informationquality and information quantity, such as disaster rescue progress andsocial well-beings.

Third, this study measures information quality in terms ofreliability, accuracy, timeliness, and relevance. Further research mayinclude other dimensions for information quality, such as complete-ness and ease of understanding.

7. Conclusion

VCs can play an important and significant role in distributingmessages, coordinating community activities, and deliveringperson-to-person relief during a disaster emergency. Many VCsare open social networks, which make them attractive forindividuals to seek information and ask for help during anemergency period. However, our study shows that VCs also haveimpediments in information overload and misinformation. There-fore, we call for socio-technical design solutions to address thechallenges of interoperability, authenticity, usability, reliability ofcitizen generated information under conditions of uncertainty.

This study has significant implications for researchers and practi-tioners. For researchers, this study extends the social network research

by investigating the characteristics of VCs in the context of naturaldisaster. One promising research area is to study the communicationpatterns and behaviors for VCs under emergency conditions. Forpractitioners, Effective information systems that provide timely accessto comprehensive, relevant, and reliable information are critical fordisaster management. This study identifies design strategies andprinciples to support disaster rescue and relief, thus sheds valuablelights on how VCs could be used effectively for disaster recovery andmanagement. Only a comprehensive approach to address three majorissues – technological, sociological, and organizational – can provide areliable communication system in extreme disaster situations.

Acknowledgement

We are grateful for financial support from the China NaturalSciences Foundation Project (70672112,71072167) and the LeadingAcademic Program, 211 Project of Southwestern University of Financeand Economics.

Appendix A. Construct definition and operationalization

537Y. Lu, D. Yang / Decision Support Systems 50 (2011) 529–538

Appendix B. Common method variance analysis

Construct Indicator Substantive factorloading (R1)

R12 Method factorloading (R2)

R22

Socialinteractionties (SIT)

SIT1 0.760⁎ 0.578 0.130 0.017SIT2 0.723⁎⁎ 0.523 0.201⁎ 0.040SIT3 0.712⁎⁎ 0.507 −0.015 0.000

Trust (TR) TR1 0.851⁎⁎ 0.724 0.026 0.001TR2 0.908⁎⁎ 0.824 0.038 0.001TR3 0.822⁎⁎ 0.676 0.072 0.005

Reciprocity (RE) RE1 0.703⁎⁎ 0.494 −0.163⁎ 0.027RE2 0.832⁎⁎ 0.692 0.086 0.007RE3 0.768⁎⁎ 0.590 0.074 0.005

Shared language(SL)

SL1 0.711⁎⁎ 0.506 0.088 0.008SL2 0.682⁎⁎ 0.465 0.073 0.005

Shared vision (SV) SV1 0.665⁎⁎ 0.442 0.097 0.009SV2 0.722⁎⁎ 0.521 0.150 0.023SV3 0.719⁎⁎ 0.520 0.012 0.000

Informationquality (Qual)

Qual1 0.733⁎⁎ 0.537 0.053 0.003Qual2 0.831⁎⁎ 0.691 0.057 0.003Qual3 0.855⁎⁎ 0.731 −0.195⁎ 0.038Qual4 0.716⁎⁎ 0.513 0.111 0.012

InformationQuantity (Quan)

Quan1 0.674⁎⁎ 0.454 0.098 0.010Quan2 0.782⁎⁎ 0.612 0.060 0.004

Average 0.758 0.580 0.053 0.011

*pb0.05; **pb0.01.

References

[1] P.S. Adler, S.W. Kwon, Social capital: Prospects for a New Concept, Academy ofManagement Review 27 (1) (2002) 17–40.

[2] M.K. Ahuja, D.F. Galletta, K.M. Carley, Individual centrality and performance invirtual R&D groups: An empirical study, Management Science 49 (1) (2003) 21.

[3] K. Andrews, B.L. Delahaye, Influences on knowledge processes in organizationallearning: The psychosocial filter, Journal of Management Studies 37 (6) (2000)797–810.

[4] S. Balasubramanian, V. Mahajan, The economic leverage of the virtual community,International Journal of Electronic Commerce 5 (3) (2001) 103–138.

[5] B. Barber, The Logic and Limits of Trust, Rutgers University Press, New Brunswick,NJ, 1983.

[6] P.M. Blau, Exchange and Power in Social Life, Wiley, New York, 1964.[7] R. Burt, Structural Holes: The Social Structure of Competition, Harvard University

Press, Cambridge, 1992.[8] R.S. Burt, The Network Structure of Social Capital, Research in Organizational

Behavior 22 (2000) 345.[9] L. Carver, M. Turoff, Human–Computer interaction: The human and computer as a

team in emergency management information systems, Communications of TheACM 50 (3) (2007) 33–38.

[10] J. Cavaye, Social Capital: A Commentary on Issues, Understanding and Measure-ment, (Observatory PASCAL—Place Management, Social Capital and LearningRegions, Australia, 2004).

[11] W.W. Chin, The partial least approach to structural equation modeling, in: G.A.Marcoulides (Ed.), Modern Methods for Business Research, Lawrence ErlbaumAssociates, Mahwah, NJ, 1998, pp. 295–336.

[12] C.-M. Chiu, M.-H. Hsu, E.T.G. Wang, Understanding knowledge sharing in virtualcommunities: An integration of social capital and social cognitive theories,Decision Support Systems 42 (2006) 1872–1888.

[13] D. Cohen, L. Prusak, In Good Company: How Social Capital Makes OrganizationsWork, Harvard Business School Press, Boston, MA, 2001.

[14] M.D. Cohen, L.S. Sproull, Organizational Learning, Sage, Thousand Oaks, CA, 1996.[15] J.S. Coleman, Social capital in the creation of human capital, American Journal of

Sociology 94 (1988) 95–120.[16] L. Comfort, Shared Risk: Complex Systems in Seismic Response, Prgamon Press,

New York, NY, 1999.[17] T.D. Cook, D.T. Campbell, Quasi Experimentation: Design and Analytical Issues for

Field Settings, Rand McNally, Chicago, 1979.[18] J.M. Day, I. Junglas, L. Silva, Information flow impediments in disaster relief supply

chains, Journal of the Association for Information Systems 10 (8) (2009) 637–660.[19] K.T. Dirks, D.L. Ferrin, Trust in leadership: Meta-analytic findings and implications

for research and practice, Journal of Applied Psychology 87 (4) (2002) 611–628.[20] J.R. Edwards, Multidimensional constructs in organizatinal behavior research: An

integrative analytical framework, Organizational Research Methods 4 (2) (2001)144–192.

[21] K.T. Erikson, A New Species of Trouble: The Human Experience of ModernDisasters, W.W. Norton and Co., New York, 1994.

[22] Facebook Statistics, in, (Facebook, 2009).[23] J. Fernback, B. Thompson, Computer-mediated communication and the American

collectivity: The dimensions of communitywithin cyberspace, Annual Convention ofthe International Communication Association (Albuquerque, New Mexico), 1995.

[24] C. Fornell, D.F. Larcker, Evaluating structural equation models with unobservedvariables andmeasurement error, Journal ofMarketing Research18 (1) (1981) 39–50.

[25] D. Gefen, D. Straub, M.C. Boudreau, Structural equation modeling and regression:Guidelines for research practice, Communications of the AIS 4 (7) (2000).

[26] S. Ghoshal, W. Tsai, Social capital and value creation: The role of intrafirmnetworks, Academy of Management Journal 41 (1998) 464–476.

[27] M. Granovetter, Economic action and social structure: The problem of embedd-edness, American Journal of Sociology 91 (3) (1985) 481–510.

[28] Q. Huang, R. Davison, H. Liu, J. Gu, The impact of leadership style on knowledge-sharing intentions in china, Journal of Global Information Management 16 (4)(2008) 67.

[29] A. Kankanhalli, B.C.Y. Tan, K.K. Wei, Contributing knowledge to electronicknowledge repositories: An empirical investigation, MIS Quarterly 29 (1)(2005) 113–143.

[30] A. Larson, Network dyads in entrepreneurial settings: A study of the governance ofexchange relationships, Administrative Science Quarterly 37 (1) (1992) 76–104.

[31] E.L. Lesser, J. Storck, Communities of practice and organizational performance,IBM Systems Journal 40 (4) (2001) 831–841.

[32] H. Liang, N. Saraf, Q. Hu, Y. Xue, Assimilation of enterprise Systems: The effect ofinstitutional pressures and the mediating role of top, MIS Quarterly 31 (1) (2007)59–87.

[33] N. Lin, Social Capital: A Theory of Social Structure and Action, CambridgeUniversity Press, Cambridge, 2001.

[34] H.-F. Lin, Determinants of successful virtual communities: Contributions fromsystem characteristics and social factors, Information and Management 45 (8)(2008) 522–527.

[35] K. Lochner, I. Kawachi, B. Kennedy, Social Capital: A Guide to Its Measurement,Health Place 5 (1999) 259–270.

[36] B.S. Manoj, A.H. Baker, Communication challenges in emergency response,Communications of the ACM 50 (3) (2007) 51–53.

[37] V. McKinney, K. Yoon, F.M. Zahedi, The measurement of web-customersatisfaction: An expectation and disconfirmation approach, Information SystemsResearch 13 (3) (2002) 296–315.

[38] J. Nahapiet, S. Ghoshal, Social capital, intellectual capital, and the organizationaladvantage, The Academy of Management Review 23 (2) (1998) 242–266.

[39] J.C. Nunnally, Psychometric Theory, McGraw-Hill, New York, 1978.[40] W.G. Ouchi, Markets, bureaucracies, and clans, Administrative Science Quarterly

25 (1980) 129–141.[41] J. Owyang, A Collection of Social Network Stats for 2009, 2009.[42] L. Palen, S.R. Hiltz, S.B. Liu, Online forums supporting grassroots participation in

emergency preparedness and response, Communications of the ACM 50 (3)(2007) 54–58.

[43] P. Podsakoff, D. Organ, Self-reports in organizational research: Problems andprospects, Journal of Management 12 (4) (1986) 531–544.

[44] P. Podsakoff, S. MacKenzie, J. Lee, N. Podsakoff, Common method biases inbehavioral research: A critical review of the literature and recommendedremedies, Journal of Applied Psychology 88 (5) (2003) 879–903.

[45] J. Preece, Online communities: Designing usability, supporting sociability, Wiley,New York, 2001.

[46] J. Preece, Supporting community and building social capital, Communications ofthe ACM 45 (4) (2002) 37–39.

[47] R.D. Putnam, Bowling Alone: The Collapse and Revival of American Community,Simon and Schuster, New York, NY, 2000.

[48] L.P. Robert Jr., A.R. Dennis, M.K. Ahuja, Social capital and knowledge integration indigitally enabled teams, Information Systems Research 19 (3) (2008) 314–334.

[49] S.B. Sitkin, N.L. Roth, Explaining the limited effectiveness of legalistic “remedies”for trust/distrust, Organization Science 4 (1993) 367–392.

[50] J.W. Thibaut, H.H. Kelly, The Social Psychology of Groups, John Wiley & Sons, NeyYork, 1959.

[51] A. Toffler, The Third Wave, Bantam, New York, 1980.[52] W.P. Tsai, Social structure of “coopetition” within a multiunit organization:

Coordination, competition, and intraorganizational knowledge sharing, Organi-zation Science 13 (2) (2002) 179–190.

[53] W. Tsai, S. Ghoshal, Social capital and value creation: The role of intra-firmnetworks, Academy of Management Journal 41 (1998) 464–478.

[54] J. Van Maanen, E.H. Schein, Toward a theory of organizational socialization, in: B.M.Staw (Ed.), Research in Organizational Behavior, JAI Press, Greenwich, CT, 1979,pp. 209–264.

[55] J.C. Wang, M.J. Chiang, Social Interaction and Continuance Intention in OnlineAuctions: A SocialCapital Perspective,DecisionSupport System, forthcoming(2009).

[56] M.M. Wasko, S. Faraj, Why should I share: Examining social capital and knowledgecontribution in electronic networks of practice, MIS Quarterly 29 (1) (2005) 35–57.

[57] M.M. Wasko, R. Teigland, S. Faraj, The provision of online public goods: Examiningsocial structure in anelectronic network of practice,Decision Support Systems 47 (3)(2009) 254–265.

[58] S. Wasserman, K. Faust, Social Network Analysis: Methods and Applications,Cambridge University Press, Cambridge, 1994.

[59] K.E. Weick, Sensemaking in Organizations, Sage, Thousand Oaks, CA, 1995.[60] M. Wetzels, G. Odekerken-Schröder, C.v. Oppen, Using PLS path modeling for

assessing hierarchical construct models: Guidelines and empirical illustration,MIS Quarterly 33 (1) (2009) 177–195.

[61] L.J. Williams, J.R. Edwards, R.J. Vandenberg, Recent advances in causal modelingmethods for organizational andmanagement research, Journal of Management 29(6) (2003) 903–936.

[62] X. Zhou,Wenchuan earthquake as reflected in the Chinese Internet, in, (China.org.cn,2008).

538 Y. Lu, D. Yang / Decision Support Systems 50 (2011) 529–538

Yong Lu is an assistant professor of information sciences and technology atPennsylvania State University, Hazleton campus. His research interests center atvirtual communities, social capital, and information security. His research articles haveappeared in Computers and Education, Journal of Computer Information Systems,Communications of the Association for Information Systems, International Conference onInformation Systems, and others.

Dan Yang is a professor of Business Administration in the School of Business Adminis-tration, SouthwesternUniversity of Finance and Economics in China.His research interestsinclude reform of state-owned enterprises, corporate governance, behavioral accountingand IPO pricing. His work has been published in leading academic journals includingJournal of Accounting Organizations and Society, Journal of Economic Research, Journal ofInvestment Research, Journal of Accounting Research, and Journal of Financial Research.