To Share or Hide? A Social Network Approach to Understanding Knowledge Sharing and Hiding in Organizational Work Teams

Knowledge Sharing and Knowledge Hiding

Knowledge sharing in organizational settings refers to the provision of work-related information, know-how, and feedback regarding how to accomplish work tasks (Hansen, 2002). Knowledge sharing can be voluntary or based on solicitation. Voluntary knowledge sharing is the act of spontaneously providing knowledge in the absence of knowledge seeking (i.e., someone in need of certain knowledge has made specific requests for such knowledge) (Cleveland & Ellis, 2015). On the other hand, knowledge sharing based on solicitation is the act of providing knowledge in response to a specific request made by the knowledge seeker. When two organizational members mutually engage in knowledge sharing, whether voluntarily or based on solicitation, knowledge exchange occurs such that these members reciprocate the provision of knowledge in the same or different knowledge domains and collaboratively use each other’s knowledge to achieve organizational goals (Sergeeva & Andreeva, 2015).

An abundance of research has studied why people share their knowledge with others. Most of these research efforts suggested that individual knowledge sharing was driven by both internal and external motivating factors (Arazy et al., 2016). Some of the internal motivating traits include individuals’ positive attitudes toward knowledge sharing (Witherspoon et al., 2013), their high-openness personalities (Cabrera et al., 2006), and their knowledge self-efficacy (Kim & Yun, 2015).

External motivators of knowledge sharing include organizational incentives such as rewards (Cabrera et al., 2006; Witherspoon et al., 2013), cognitive job demands and autonomy (Gagné et al., 2019), as well as an organizational culture that supports knowledge sharing (Witherspoon et al., 2013). In addition, past research has found that organizational members’ perceived time pressure had a negative influence on their knowledge sharing because such perceptions made them feel “too busy” to share what they knew with their coworkers (Connelly et al., 2014).

Although knowledge sharing is almost ubiquitously promoted in today’s organizations, knowledge hiding is still prevalent (Connelly et al., 2019). Knowledge hiding is defined as intentional withholding knowledge from others who specifically request such knowledge (Connelly et al., 2012). In other words, knowledge hiding is a deliberate, non-compliant response to knowledge seeking. A commonly confused concept with knowledge hiding is knowledge hoarding, which refers to secretly preserving and carefully guarding the knowledge that one has acquired (Connelly et al., 2012). An important distinction between the two concepts is that knowledge hiders intentionally withhold or conceal knowledge from knowledge seekers, whereas knowledge hoarders indulge in accumulating and protecting their knowledge (Webster et al., 2008).

To date, little research has sought to uncover the reasons why organizational members hide or withhold their work-related knowledge from their coworkers. Connelly et al. (2012) may be the first empirical study that specifically examines how and why knowledge hiding takes place in real-world organizations. Their research found that knowledge hiding was positively influenced by interpersonal distrust and knowledge complexity, but negatively affected by task-relatedness of the knowledge and a pro-knowledge sharing organizational climate (Connelly et al., 2012). Later research showed that knowledge hiding was associated with psychological traits such as Machiavellianism (Pan et al., 2018), perceived organizational politics (Malik et al., 2019), and distrust and competitiveness within the organization (Hernaus et al., 2019).

In sum, far more research exists on organizational knowledge sharing than on knowledge hiding. An implicit but questionable assumption of these studies is that knowledge sharing and hiding are mutually exclusive (Connelly et al., 2012). In other words, when the factors promoting knowledge sharing are present, knowledge hiding is demoted accordingly. Further, almost no research has examined both knowledge sharing and knowledge hiding within the same theoretical and analytical framework (cf. Gagné et al., 2019). It is yet unclear how knowledge sharing is related, if at all, to knowledge hiding. In addition, while the majority of the research on knowledge sharing and hiding was conducted in Western organizational contexts, an increasing number of studies have begun to pay attention to non-Western regions such as East Asia (Jiang et al., 2019) and South Asia (Malik et al., 2019). However, in order to expand the generalizability of the theoretical and empirical implications of these studies, more research is needed to study knowledge sharing and hiding in a broader span of national and cultural contexts.

Lastly, previous research predominantly attributed organizational knowledge sharing and hiding to individual traits and organizational properties without considering the effects of interpersonal relationships that connect individuals within the organization (Gagné et al., 2019). After all, knowledge sharing and hiding are inherently interpersonal relationships (Connelly & Zweig, 2015) because they are defined by the relational roles played by the people involved in such behaviors (i.e., knowledge sharing would exist only if there is someone providing the knowledge and someone receiving the shared knowledge ¹ ). Thus, knowledge sharing and hiding will not be fully understood without taking into consideration those interpersonal relationships that help cultivate and constrain the knowledge sharing and hiding relationships among organizational team members.

Therefore, to bridge these research gaps, this study will employ and extend transactive memory systems (TMS) theory (Wegner, 1986) and the dialectical tension model (Gibbs et al., 2013) to explore how knowledge sharing is related to knowledge hiding, as well as how they are associated with various work and social relationships in organizational work teams. This study will empirically test these relationships with data collected from a multitude of organizational work teams from both the U.S. and China.

Transactive Memory Systems Theory

For over three decades, transactive memory systems (TMS) research has established a multi-level theoretical framework to understand how knowledge is distributed, recognized, shared, and used at the dyadic, group, organizational, and inter-organizational levels (Wegner, 1986). Central to the development of TMS in a work team is the recognition of team members’ expertise (Moreland, 1999). After the experts are recognized in the team, non-experts in need of the expertise can retrieve knowledge from the experts without becoming experts themselves. Non-experts may also allocate expertise-related information to the experts because such information can be optimally processed and stored by the experts. Through an integrated process of expertise recognition, knowledge retrieval, and knowledge allocation, TMS can be developed within a work team where team members become increasingly specialized in their expertise domains and reduce their redundant efforts to learn and manage knowledge outside their own specialty areas (Lewis, 2004).

Existing TMS research has highlighted several signature interpersonal relationships that are essential for team TMS development and relevant to intra-team knowledge sharing and hiding. First, an important premise of TMS development is the extent to which individuals depend on each other’s expertise to successfully accomplish their work tasks (Hollingshead, 2001). In other words, without task interdependence, team members may be capable of completing their jobs entirely on their own and lack the motivation to share knowledge with or seek knowledge from their coworkers. Therefore, task interdependence in a work team should increase knowledge sharing among team members given their reliance on each other’s expertise for task completion. Likewise, it is unlikely for task interdependent team members to hide knowledge from one another because such behavior would impede their knowledge transfer to successfully accomplish their work tasks (Fong et al., 2018).

Second, perception of team members’ knowledge (i.e., expertise recognition) is also an integral component and mechanism of TMS development. Without knowing the expertise possessed by other members, TMS may not be accurately established (Wegner, 1986). After all, it is based on “who knows what” that team members make a decision on “with whom to share the knowledge” and “from whom to seek knowledge” in TMS development. Thus TMS theory suggests that team members are likely to share knowledge with those whom they perceive as experts so that such knowledge can be optimally processed and stored by the experts in that particular knowledge domain. Moreover, when team members need to use knowledge outside their domains of expertise, they will request such knowledge from the perceived experts who are expected to share the requested knowledge with the knowledge seekers.

Although not explicitly addressed by TMS theory, perception of team members’ expertise may affect knowledge hiding in more complex ways. On the one hand, team members may hide knowledge from those coworkers whom they perceive as non-experts in a particular domain so that the non-experts would not be distracted or overwhelmed by the knowledge outside their expertise domains (Hollingshead et al., 2007). Further, team members who perceive themselves as non-experts in a particular knowledge domain may choose to hide their knowledge in that domain from other team members so as not to distribute flawed or low-quality knowledge (Lou et al., 2013). These knowledge hiding mechanisms may contribute to the efficiency and quality of TMS development. On the other hand, when team members hide knowledge from the perceived experts in a particular domain, or when the perceived experts hide knowledge from team members who request such knowledge, these knowledge hiding behaviors would impede TMS development and diminish effective knowledge management within the team (Shi & Weber, 2018).

Finally, there is an ongoing debate in the current TMS scholarship over the role of interpersonal communication in TMS development. After reviewing a collection of TMS studies, Hollingshead and Brandon (2003) asserted that intra-team communication had an overall positive influence on TMS development in the following ways: (1) it facilitated expertise recognition and assisted with the decisions on “who will learn what” within the team, (2) it supported knowledge retrieval by helping team members accurately and tactfully approach the experts for information, and (3) it enhanced team performance by giving team members an opportunity to develop a shared mental model of their task completion. However, based on a review of existing TMS literature, Pavitt (2006) argued that interpersonal communication was an imperfect conduit of expertise recognition and might not benefit TMS development. For example, it was found that the lack of communication did not worsen team performance for those teams whose members were trained apart (Moreland & Myaskovsky, 2000). Additionally, the positive effects of interpersonal communication on team performance would decrease as virtual teams developed a shared understanding of “who knows what” (Yoo & Kanawattanachai, 2001).

Despite the mixed findings and controversies regarding the role of interpersonal communication in TMS development, the effects of communication on knowledge sharing and hiding may be less equivocal. When team member A communicates frequently with team member B, member A is likely to share knowledge with member B because it should cost less time and effort given the existing communication pathway between the two members (Yuan et al., 2010). In return, such knowledge sharing may strengthen the communication ties between members A and B. Similarly, due to member A’s frequent communication with member B, member A is unlikely to hide knowledge from member B because such behavior may cause relational damage and impede member A’s communication with member B in the future (Connelly & Zweig, 2015). In this sense, interpersonal communication serves as a “lubricant” for knowledge sharing and “buffer” for knowledge hiding in organizational work teams.

Although TMS theory offers an instrumental perspective to explain how and why work-related knowledge can be shared and hidden among a collective of individuals, most TMS research examines knowledge sharing and hiding separately rather than within the same theoretical framework. A clear conceptual distinction between knowledge sharing and hiding is missing in the current TMS scholarship. To fully make sense of the increasing empirical support of the co-existence of knowledge sharing and hiding in today’s organizations, additional theoretical perspectives should be considered. Therefore, this study will apply the dialectical tension model to complement TMS theory in delineating the relationship between knowledge sharing and hiding.

The Dialectical Tension Model

Grounded in dialectical theory, the dialectical tension model (Gibbs et al., 2013) suggests that while employees are encouraged and empowered by their organizations to engage in open communication and knowledge sharing with their coworkers, they also demonstrate the propensity and capability of doing exactly the opposite: withholding their knowledge, making themselves less accessible to knowledge seekers, and evading knowledge requests. Therefore, the relationship between knowledge sharing and hiding can be characterized by a dialectical tension, which implies that knowledge sharing and hiding are two opposing but co-existing trends in organizational work teams.

Existing literature has shed light on several interpersonal relationships that may affect the intensity of the dialectical tension between knowledge sharing and hiding (Gagné et al., 2019). First, competition within the work team may intensify the dialectical tension such that knowledge hiding may suppress knowledge sharing (Hernaus et al., 2019). For example, when team member A perceives member B to be a strong competitor in the work team, member A is less likely to share knowledge with, but more likely to hide knowledge from, member B. This is because personal knowledge is increasingly considered and leveraged as intellectual capital in the workplace (Massaro et al., 2014). Thus, team members who are in competition with each other may strive to preserve their intellectual capital and maintain their competitive advantage by hiding rather than sharing their knowledge.

By contrast, interpersonal trust may contribute to the alleviation of the dialectical tension between knowledge sharing and hiding (Hernaus et al., 2019). After all, knowledge sharing entails a significant degree of uncertainty and risks. The knowledge shared by the knowledge provider may be abused by the knowledge seeker and even used against the interest of the knowledge provider. Thus, when the knowledge provider lacks trust in the knowledge seeker, the dialectical tension between knowledge sharing and hiding may be intensified and tilted toward knowledge hiding. However, when there is sufficient trust between the knowledge provider and seeker, the dialectical tension may be alleviated to the extent that knowledge sharing has an equal, if not higher, chance to occur. Indeed, a content analysis of 103 knowledge management studies (Cleveland & Ellis, 2015) found lack of trust to be the primary reason why organizational members withheld their knowledge in the workplaces, whereas the presence of interpersonal trust was shown to be a driving force of knowledge sharing (Holste & Fields, 2010).

Finally, the dialectical tension between knowledge sharing and hiding can be eased by positive interpersonal relationship variables such as affection and interpersonal justice. It is expected that when team member A likes member B personally, member A will be socially motivated to share knowledge with member B and will also gain a greater level of relational gratification from knowledge sharing (Connelly et al., 2012). Meanwhile, member A is unlikely to hide knowledge from member B because such behavior will be dissonant with member A’s personal affection for member B. Furthermore, when team member A feels that s/he has been treated with justice by member B, member A is likely to reciprocate such a benevolent relationship by sharing knowledge rather than hiding knowledge (Webster et al., 2008). Taken together, interpersonal affection and justice will mitigate the dialectical tension between knowledge sharing and hiding such that knowledge sharing may very likely suppress knowledge hiding when those interpersonal relationship variables are present in a work team.

Social Network Approach and Research Questions

To properly study how knowledge sharing is related to knowledge hiding, as well as how the aforementioned interpersonal relationships are related to knowledge sharing and hiding, this research employs a social network approach. A social network is commonly composed of a collection of individuals (defined as nodes) who are connected by a specific type of interpersonal relationship (defined as ties) (Scott, 2000). When studying an organizational work team as a social network, the nodes refer to individual team members, and the network ties can include “communication, influence, workflow, activation, or any other relationship of interest between two nodes in the network” (Monge & Contractor, 2003, p. 105). While traditional non-network approaches to studying teams primarily focus on how individual and team behaviors can be influenced by personal and collective attributes within the team, the social network approach focuses on how such behaviors can be affected by the relationships and interactions that connect part or all of the team members.

The social network approach has unique advantages when applying the dialectical tension model to examine knowledge sharing and hiding in the same conceptual and analytical framework. The dialectical tension model (Gibbs et al., 2013) suggests that while knowledge sharing and hiding seem to be two opposing forces, they do not necessarily supersede each other but can co-exist in organizational settings. Thus, two networks can emerge concurrently in a given organizational work team. The knowledge sharing network depicts how team members are connected through the knowledge sharing relational ties among them. Likewise, the knowledge hiding network represents how team members are connected through the knowledge hiding relational ties among them. By studying the relationship between the knowledge sharing and knowledge hiding networks, this research will further uncover whether and how the dialectical tension is manifested among members of a work team. Therefore, the first question this research seeks to address is the following:

The social network approach is also instrumental in examining how the knowledge sharing and hiding networks are related to other co-evolving networks in a given organizational work team. As previous research suggests, knowledge sharing and hiding may be driven and constrained by a multitude of interpersonal relationships between the knowledge provider and the knowledge seeker (Connelly et al., 2012; Gagné et al., 2019). For example, team member A may share knowledge with member B because member A perceives member B to be an expert within the team. Meanwhile, member A may hide knowledge from member C because member A does not trust member C. Thus, simply inquiring about a team member’s general tendencies of knowledge sharing and hiding will mask the relational characteristics that influence this member’s knowledge sharing and hiding relationships with a specific coworker. In comparison, the social network approach taps into an individual’s relational ties with each and every other team member and offers greater insights into the dynamics and nuances of interpersonal relationships including, but not limited to, knowledge sharing and hiding.

Based on the aforementioned discussion of TMS theory and the dialectical tension model, two dimensions of interpersonal relationships have emerged to be strongly related to organizational knowledge sharing and hiding. The first dimension is work relationships that are created and sustained through professional and job-related interactions in the workplaces. Each of these work relationships can be conceived as a network in a given organizational work team: the task interdependence network (who relies on whom to complete job tasks), the perception of expertise network (who considers whom to be the expert in the work domains), the perceived competition network (who competes with whom in the team), and the work-related communication network (who communicates with whom for work purposes). As discussed earlier, many of these networks represent signature interpersonal relationships that are especially critical in team TMS development processes.

The second dimension is social relationships, which refer to those interpersonal relationships that are established based on personal and social connections rather than work-related interactions. Again, each of these social relationships can be conceived as a network in a given organizational work team: the trust network (who trusts whom), the affection network (who likes whom), the interpersonal justice network (who has treated whom with justice), and the social communication network (who communicates with whom for social purposes). As previously discussed, these socially generated or oriented relationships are particularly helpful in alleviating the dialectical tension between knowledge sharing and hiding that co-exist in a work team.

Therefore, using and extending TMS theory and the dialectical tension model, this study proposes the following research questions regarding how networks of work and social relationships within an organizational work team are related to the knowledge sharing and hiding network respectively.

Sample and Procedures

This research conducted online surveys in 31 organizational work teams in the U.S. and China during 2016 to 2018. A total of 200 employees completed the survey, including 95 members from 16 teams in the U.S. and 105 members from 15 teams in China. The U.S. and China are the world’s two largest economies, but represent dramatically different political, cultural, and social systems. Research on organizational work teams in both countries overcomes a major limitation in prior knowledge sharing and hiding research that primarily focuses on Western organizations (Connelly et al., 2019), and helps broaden the organizational, cultural, and national contexts of studying TMS development and the dialectical tension model.

Within each country, work teams of a wide array of organizational backgrounds were selected, ranging from health care, finance, advertising, to education. Each team included a small group of employees who had been working closely with each other in the organization (e.g., employees working in the same department; employees from different departments collaborating with each other on a shared project). Upon organizational approval, the leader of each team was asked to provide detailed background information about the team, including the names and email addresses of all members, their job titles, and the primary work tasks performed by the team members. Based on such information, an online survey was customized and emailed to each team member to be completed individually. Table 1 provides details of the size, organizational background, response rate, and team members’ demographic information of all 31 work teams who participated in this research.

OPEN IN VIEWER

Table 1.

Organizational and Demographic Information of Work Teams Across the U.S. and China.

ID	Industry	Size a (original/responded)	Response rate	Age (mean)	Gender (M/F)	Ethnicity (highest %)
US1	Accounting	10/6	60%	43	5/1	100% White
US2	Estimating services	13/7	54%	46	2/5	43% Latino
US3	Expediting services	6/3	50%	40	2/1	67% Latino
US4	Inventory management	10/8	80%	40	6/2	63% Latino
US5	Sales	10/5	50%	28	0/5	100% White
US6	Web design	11/11	100%	29	11/0	82% White
US7	Aerospace engineering	6/6	100%	32	6/0	67% White
US8	Medical research	7/7	100%	55	0/7	100% White
US9	Non-profit	7/5	71%	38	4/1	80% Asian
US10	Real estate	7/5	71%	41	1/4	40% White
US11	Pharmaceutical	5/5	100%	28	2/3	40% Latino
US12	Legal services	6/6	100%	47	2/4	50% White
US13	Marketing	6/6	100%	45	4/2	50% White
US14	Aerospace	6/5	83%	56	5/0	60% White
US15	Medical research	6/6	100%	32	3/3	67% Asian
US16	Pharmaceutical	7/4	57%	40	1/3	50% White
U.S. total		123/95	77% (average)
China1	Advertising	9/9	100%	32	4/5	100% Han
China2	Marketing	5/5	100%	31	1/4	100% Han
China3	Advertising	6/6	100%	36	3/3	100% Han
China4	Education	8/8	100%	40	0/8	100% Han
China5	IT R&D	9/7	78%	41	6/1	100% Han
China6	Electronics	5/5	100%	29	5/0	100% Han
China7	Hardware R&D	14/14	100%	32	11/3	100% Han
China8	Software R&D	7/7	100%	27	3/4	57% Han
China9	Engineering	5/5	100%	40	4/1	100% Han
China10	Software R&D	5/5	100%	32	1/4	100% Han
China11	Management	6/5	83%	43	3/2	100% Han
China12	Medical research	5/4	80%	32	2/2	100% Han
China13	Product design	9/9	100%	30	3/6	100% Han
China14	IT R&D	8/8	100%	30	6/2	75% Han
China15	Finance	11/8	73%	34	6/2	88% Han
China total		112/105	94% (average)

a

The first size refers to the total number of team members to whom the survey was distributed, and the second size refers to the total number of team members who completed the survey. As this study focuses on analyzing whole-network data (i.e., every team member provides data about their relational ties with everyone else in the same team), it is essential for the network analysis to be performed on complete network data rather than network data with missing responses. Therefore, for those teams whose response rate was less than 100%, this study adopted the “complete-case” approach as suggested in the previous research on the treatment of missing network data (Huang et al., 2019; Huisman & Steglich, 2008). The “complete-case” approach removes the non-respondents’ relational data, and analyzes a smaller network that includes only those members who completed the survey.

This research collected whole-network data from each team as the survey asked each team member to report her/his relationships with each and every other member in the same team regarding a specific type of interpersonal relationship proposed in the research questions. Whole-network data differ from egocentric network data, which primarily concentrate on the focal (or “ego”) member’s network connections (Marsden, 2005). Some personal and demographic data were also collected in the survey.

Measures

Analysis

Since social network data violate the sample independence assumption of traditional statistical analysis, this research employs exponential random graph modeling (ERGM) analysis to address the proposed research questions. ERGM analysis has been developed to test the significance and strength of interdependencies between networks (Robins et al., 2007). Although scholars have emphasized the importance and viability of using ERGM techniques to analyze interpersonal relationships and communicative behaviors (Shumate & Palazzolo, 2010), almost no research has applied this method to analyze organizational knowledge sharing and hiding (Connelly et al., 2019).

Specifically, this research used XPNet to perform the ERGM analyses. XPNet is a computer program developed for bivariate network analysis using the Monte Carlo Markov Chain (MCMC) maximum likelihood estimation (Wang et al., 2006). To answer each of the research questions proposed, the multiplexity parameter was estimated in the ERGM analysis to test the extent to which network X was related to network Y. A goodness-of-fit test was also performed to assess the robustness of each estimation. A positive, significant, and robust parameter estimate indicates that network X is more likely to be positively related to network Y than to occur by random chance. A larger (when significant and robust) estimate value implies a greater chance for such a positive relationship between network X and network Y to occur in the observed networks.

Given that a total of 31 organizational work teams participated in the surveys, this research generated 31 networks for each variable under study. Instead of conducting the ERGM analysis on each of these 31 networks separately, this research performed a “meta-level” ERGM analysis on the entire 31 networks simultaneously. Following a procedure employed in previous research (Zhu et al., 2013), this study combined all 31 networks into a “meta-level” network in which each of the 31 networks was entered into the diagonal of a “meta-level” matrix and the rest of the matrices were filled with structural zeros. Thus, the structural zeros represented the relational ties that did not exist among members of different teams. When running ERGM analyses on the XPNet, the effects of these structural zeros were controlled for, which generated a single set of estimation and simulation results for each research question.

To answer RQ1, the multiplexity parameter was estimated to test the extent to which the knowledge sharing network was related to the knowledge hiding network. To answer RQ2a, the multiplexity parameter was estimated to test how each of the four work relationship networks (i.e., the task interdependence network, the perception of expertise network, the perceived competition network, and the work communication network) was related to the knowledge sharing network respectively. To answer RQ2b, the multiplexity parameter was estimated to test how each of the four social relationship networks (i.e., the trust network, the affection network, the interpersonal justice network, and the social communication network) was related to the knowledge sharing network respectively. To answer RQ3a, the multiplexity parameter was estimated to test how each of the four work relationship networks was related to the knowledge hiding network respectively. Finally, to answer RQ3b, the multiplexity parameter was estimated to test how each of the four social relationship networks was related to the knowledge hiding network respectively.

Co-existent Relationship Between Knowledge Sharing and Hiding

One of the primary findings of this research is that knowledge sharing is not significantly related to knowledge hiding. In other words, knowledge sharing and hiding do not replace each other, but can and do co-exist in organizational work teams. This finding reveals a co-existent relationship between knowledge sharing and hiding, both of which are associated with two different dimensions (work and social) of interpersonal relationships respectively. On the one hand, team members share what they know primarily due to existing work relationships that connect the knowledge sharer and sharee through professional or job-related interactions. On the other hand, team members hide what they know mostly due to the absence of positive relationships that connect the knowledge sharer and sharee socially. In this sense, this research not only validates the existence of the dialectical tension between knowledge sharing and hiding (Gibbs et al., 2013), but also uncovers the two primary sources of the dialectical tension (i.e., the work and social relationships among team members). This research lends additional empirical support to the proposition of a co-existent relationship between knowledge sharing and knowledge hiding in the knowledge management literature (Connelly et al., 2012; Gagné et al., 2019). The specific relationship between knowledge sharing and hiding discovered in this research can be illustrated in a matrix shown in Figure 1.

OPEN IN VIEWER

Figure 1.

Illustration of the co-existent relationship between knowledge sharing and hiding.

As the top right quadrant in Figure 1 (labeled as S-NH) demonstrates, when both pro-sharing work relationships and con-hiding social relationships are present, team members are sharing and not hiding their work-related knowledge, which may be the optimal state of intra-team knowledge transfer. For example, in a web design team, the graphic designer is sharing her/his knowledge with the HTML coder because they have been frequently communicating about the design of the logo on the homepage. At the same time, the graphic designer is not hiding her/his knowledge from the usability tester within the team because the usability tester has treated her/him with justice.

In the top left quadrant (labeled as S-H), the pro-sharing work relationships are present but the con-hiding social relationships are absent. In this scenario, team members are sharing knowledge but are hiding knowledge as well. Again, in the example of the web design team, the graphic designer is obligated to share knowledge with the HTML coder for work reasons because they depend on each other to create all the graphics needed for web design. However, the graphic designer is hiding knowledge from the usability tester for social reasons such as lack of trust.

The bottom left quadrant (labeled as NS-H) refers to a scenario in which both the pro-sharing work relationships and con-hiding social relationships are absent. Thus team members are not sharing knowledge and are hiding knowledge at the same time. This may be the least desirable situation when knowledge transfer is vital to successful team performance. In the example of the web design team, the graphic designer is not sharing knowledge with the usability tester because they hardly have any job overlaps that require them to exchange ideas and information. On the other hand, the graphic designer is hiding knowledge from the HTML coder simply because they are not friends with each (i.e., lack of social communication).

Finally, in the bottom right quadrant (labeled as NS-NH), the pro-sharing work relationships are absent but the con-hiding social relationships are present. In this scenario, team members are not sharing knowledge but are not hiding knowledge either. Again, in the example of the web design team, the graphic designer is not sharing knowledge with the usability tester because they are not dependent on each other for task completion. However, the graphic designer is not hiding knowledge from the HTML coder because they are good friends and socialize with each other very frequently.

Communication Does Matter

This research lends further support to the critical role of interpersonal communication in influencing knowledge sharing and hiding and consequently, TMS development. In response to the ongoing debate over whether communication matters in TMS development (see Hollingshead & Brandon, 2003; Pavitt, 2006), this study affirms that interpersonal communication is significantly related to team members’ knowledge sharing and hiding, albeit in different directions. While work-related communication facilitates knowledge sharing, social communication hinders knowledge hiding. Taken together, regardless of whether team members communicate for work purposes or social reasons, intra-team communication promotes the sharing and exchange of knowledge while reducing the hiding and withholding of knowledge within the team.

This research supports previous research (Lewis & Herndon, 2011) in concluding that a more nuanced approach to studying the role of communication in TMS development is needed and beneficial. As demonstrated in this study, if team members are not engaged in social communication, there is a greater risk for them to hide knowledge from each other. Likewise, if team members are not communicating for work purposes in the workplace, there is a greater barrier for them to share knowledge with each other. Coupled with the previous finding that the lack of communication decreases the accuracy in identifying expertise within the work team (Su, 2012), it is evident that knowledge exchange and transfer will be greatly impaired in the absence of interpersonal communication. Consequently, there will be limited use of distributed expertise within the team, and a greater level of individual effort will be wasted in learning, seeking, and processing knowledge outside team members’ expertise domains. Eventually, TMS development will be delayed and even eroded in the work team. In this sense, intra-team communication, whether for work or social purposes, contributes to TMS development vis-à-vis promoting knowledge sharing and reducing knowledge hiding.

Support for and Extension of TMS Theory

This research empirically validates one of the central premises and foundations of TMS development, task interdependence (Hollingshead, 2001). According to TMS theory, when team members are interdependent on each other’s expertise, information, and resources to successfully complete their work tasks, they will become increasingly more specialized in their expertise domains, and more dependent on other experts to store, process, and provide knowledge in their non-mastered knowledge domains. This mechanism instigates the emergence of TMS and fuels sustainable growth of TMS.

Furthermore, this study reiterates the importance of interpersonal trust in TMS development. Trust is considered to be a foundation and facilitator of effective knowledge sharing and transfer (Choi & Cho, 2019). When members could potentially lose their competitive advantages by sharing critical knowledge (Hollingshead et al., 2007), they will certainly hide knowledge from someone whom they do not trust, due to concerns such as that the knowledge could be tampered with, abused, or used against their own interest. However, such knowledge hiding may trigger what researchers call a reciprocal distrust loop (Černe et al., 2014) within the team so that an increasing number of members will engage in knowledge hiding rather than sharing, which will have a perverse impact on TMS development and team performance. Indeed, Černe et al. (2014) found that knowledge hiding not only obstructed knowledge hidees’ information seeking and creativity, but also hampered the creativity of the knowledge hiders themselves.

This study identifies relational mechanisms that are not traditionally examined in TMS research but are shown to be instrumental in influencing TMS development. First, interpersonal justice was found to be a significant barrier for knowledge hiding in this research. In other words, if member A perceives her/himself to be treated without justice by member B, member A is more likely to hide work-related knowledge from member B. This retaliation may be better explained by social exchange theory (Roloff, 1981), which suggests that people are likely to exchange favors to maximize their benefits. Likewise, people have the tendency to return disfavor in order to minimize their costs or take revenge. If this mechanism continues to be reciprocated, it will impede knowledge transfer and ultimately disintegrate the TMS within the work team. Second, this research found that knowledge hiding could be positively influenced by one type of work relationship, perceived competition. A prior study showed that perceived competition was not directly related to knowledge sharing (Connelly et al., 2014), which is replicated in the current research. More importantly, the present research revealed that although perceived competition had no direct impact on knowledge sharing, it would promote knowledge hiding. In other words, when team members are in competition with each other in the workplace, they are more likely to hide or withhold their knowledge from each other.

Finally, this study demonstrates the benefits and viability of employing advanced social network analysis techniques (e.g., ERGM analysis) to examine the dynamics and interdependencies among various interpersonal relationships. In a recent review of current issues and future directions of TMS research, Lewis and Herndon (2011) call for more research effort to adopt a social network approach to TMS studies. Indeed, those mechanisms central to TMS development (e.g., task interdependence, expertise recognition, knowledge seeking and retrieval) are intrinsically relational, and should be further examined from a social network perspective. After all, team members do not work independently, but constantly create, maintain, and dissolve their relational ties with each other. A social network approach holds promise in uncovering the richness and complexity of the inter-connections among work team members.

Managerial Implications

In recent decades, organizations worldwide have taken drastic measures to promote knowledge sharing and information transfer in the workplaces (Janus, 2016) only to find that their employees are often reluctant to share what they know, and even hide their knowledge from their coworkers (Connelly et al., 2019). A primary implication of the present research is that management should consider promoting knowledge sharing and reducing knowledge hiding with different strategies and tactics. On the one hand, in order to encourage and facilitate knowledge sharing, organizations should strategically increase task interdependence among the employees by developing collaborative initiatives and team-based projects. In addition, management should maximize opportunities for employees to engage in work-related communication on and off the work sites via face-to-face or virtual meetings, brainstorming sessions, formal and informal feedback, computer-supported collaborative team work, and other collaborative activities.

On the other hand, in order to diminish knowledge hiding, organizations should be dedicated to cultivating positive social relationships among their employees. Specifically, management should help establish a collegial and trusting culture in which team members not only treat each other with respect and justice, but also develop a sense of mutual trust so that they feel safe and comfortable sharing their knowledge. Interpersonal injustice should be condemned culturally and penalized administratively. Further, the management should optimize workspace and time for employees to engage in social communication. For example, organizations can integrate social activities (e.g., cook-offs, games, races, and parties) into professional work. Finally, management should recognize that while too little competition would stifle employee creativity and productivity, too much competition could boost knowledge hiding and power struggles. Management should foster constructive competition within the organization such that the competing process is transparent and fair, and the competing outcomes are based on merit and ethics.

Limitations and Future Directions

This study has some limitations that call for additional research to address them. First, all relational data collected and analyzed in this research were survey data, which entailed self-reported biases. Previous research suggests that survey questions may trigger potential “cooperation bias” so that respondents tend to over-report knowledge sharing as it is generally considered a desirable behavior by the researchers or the organization (Witherspoon et al., 2013), but to under-report or even deny knowledge hiding because such behavior connotes a negative or anti-social meaning (Connelly et al., 2012). Indeed, this study found that knowledge sharing was reported to be much more frequent than knowledge hiding. ¹ A remedy implemented by this study was to decrease the cut-off threshold when dichotomizing the knowledge hiding network. In this way, any knowledge hiding frequency reported as “Rarely” rather than “Sometimes” was counted as the presence of the knowledge hiding relationship. The assumption is that in situations when members are likely to under-report their knowledge hiding, a self-reported “Rarely” may very likely suggest a more frequent knowledge hiding behavior. However, to achieve an even more accurate and objective measurement of knowledge sharing and hiding, future research should explore alternative data sources and data collection methods such as archival and observational data.

Second, although this research sought to maximize the external generalizability by analyzing data collected from work teams of various organizational backgrounds in the U.S. and China, this study does not specifically focus on how team, organizational, and national characteristics and differences could influence knowledge sharing and hiding. Multi-level analyses are needed to adequately examine the effects rendered by team-, organization-, and nation-specific properties. Moreover, while the U.S. and China are the world’s two largest economies, they comprise dramatically different political, cultural, and social systems. Thus, future research should further investigate how cultural differences could influence organizational knowledge sharing and hiding and consequently, TMS development. Indeed, previous research suggests that certain cultural orientations, such as collectivism compared to individualism, might facilitate knowledge sharing (Witherspoon et al., 2013). It will be beneficial to examine the direct effects, as well as potential moderating effects, of national or cultural characteristics on team members’ knowledge sharing and hiding. For example, it might be possible that the negative influence of social relationships on knowledge hiding found in this study could be strengthened in a collectivist culture or among collectivism-oriented individuals, but weakened in an individualistic culture or among individualism-oriented individuals.

Third, this study does not focus on the outcomes of organizational knowledge sharing and hiding. In their meta-analysis of 72 published research, Mesmer-Magnus and DeChurch (2009) confirmed the importance of knowledge sharing to team performance, cohesion, decision satisfaction, and knowledge integration. Further, while much of the existing research revealed the positive outcomes of knowledge sharing (Ren & Argote, 2011) and negative effects of knowledge hiding (Witherspoon et al., 2013), more research is needed to pinpoint the circumstances under which knowledge sharing might spur negative outcomes and knowledge hiding might be productive. Indeed, previous research suggests that excessive knowledge sharing and the receipt of unsolicited information could reduce team members’ competitive performance (Haas & Hansen, 2005) and job satisfaction (Su et al., 2010). On the other hand, appropriate knowledge hiding may reduce the cognitive overload within knowledge intensive work teams. It is imperative for future research to provide more insights into individual, team, and organizational outcomes as a result of knowledge sharing and hiding.

Finally, as stated earlier, a major limitation of the ERGM analysis employed by this study is that all data input were restricted to binary data. The dichotomization of continuous network data sacrifices the richness and subtle variance of the network ties under study. Further, unlike multiple regression analysis that predicts the relationship between one dependent variable and multiple independent variables, the ERGM analysis performed in the XPNet is only able to assess the relationship between two networks (variables). Therefore, the findings of this research should be interpreted with caution. It is hopeful that with the advancement of social network analysis techniques, future ERGM development will offer more powerful and accessible solutions to address these limitations.