Expanding organisational knowledge online: the role of bridging members in knowledge expansion in online groups

ABSTRACT

In this paper we hypothesise about the role of bridging members – people who connect otherwise unconnected online groups – on an enterprise social network (ESN) and in what manner their role affects their group’s knowledge expansion. Analysing data from an ESN of a Dutch health insurance company, we show how online groups can benefit from their bridging members who take advantage of the affordances of the digital platform. We find that a larger proportion of bridging members within a group mitigates the decrease in knowledge expansion over time. This research contributes to the literature on online knowledge sharing and social networks by showing the effect of bridging members on knowledge expansion in a digital environment. Our study has implications for organisations who want to leverage knowledge sharing for innovation.

KEYWORDS:

• Digital platforms
• knowledge sharing
• knowledge expansion
• brokerage
• social networks
• bridging members
• digital affordances
• enterprise social networks
• multiple team membership

Introduction

Digital technologies in organisations provide new possibilities for them to leverage their innovative potential. Organisations and organisational scholars are particularly interested in the role of online knowledge flows. This is because knowledge and knowledge sharing are key to bringing about innovation in organisations (Kim & Chung, 2017; Nambisan et al., 2017; Yoon et al., 2017), which many organisations consider to be at the core of their business model and strategies. The creation and diffusion of knowledge in organisations is an important theme in organisational research (Baralou & Tsoukas, 2015; Cross et al., 2001; Haas & Hansen, 2007; Hansen et al., 2005; Kane et al., 2014; Lee & Cole, 2003; Nonaka, 1994). While a strong stream of research has been (and continues to be) devoted to the study of knowledge in organisations, we yet have to fully understand the role of digital technologies in organisational knowledge flows (Faraj et al., 2011, 2016; Von Krogh, 2012). Within this stream of research, we find considerable evidence for the importance of knowledge expansion specifically (Dyer & Nobeoka, 2000; Tortoriello et al., 2015); that is, it is important that the knowledge that is being shared within a group adds to the existing knowledge generated by the group. Knowledge expansion is one aspect of knowledge management (Ponis & Koronis, 2012) and involves acquiring, generating and selecting knowledge, and creating linkages within and between organisational groups to foster creativity and networking (Davenport & Prusak, 2000). Such knowledge expansion can improve organisational outcomes such as innovation (Tortoriello et al., 2015; Xue, 2018) and creativity (Mannucci & Yong, 2018).

An important avenue through which digital technologies are thought to benefit organisations, in particular in terms of knowledge expansion, is through lowering the threshold for inter-organisational communication. Here, questions raised in the stream of research devoted to the study of knowledge in organisations are combined with network research on organisations to better understand the role of digital technologies in organisational knowledge flows (Faraj et al., 2011, 2016; Von Krogh, 2012). Employees in knowledge-intensive organisations increasingly make use of enterprise social networks or ESNs (Dimicco et al., 2008; O’Leary, 2016) which are widely adopted by large organisations because they allow for collaboration between people independent of place and time (Faraj et al., 2011). In ESNs, individuals can voluntarily team up with others in groups, where they share knowledge that is indefinitely accessible for anyone in the organisation (Leonardi, 2017). Despite the potential that a digital platform like an ESN has to offer in terms of knowledge management in general and knowledge expansion in particular, acquiring and implementing an ESN requires substantial investments in terms of time and money. Therefore, it is important to understand the drivers of effective ESN use with regards to knowledge expansion.

In this paper, we argue that the brokerage position of bridging members results in brokering behaviour of those bridging members. Bridging members’ behaviour, in turn, influences interactions within and between the bridged groups and as such influences knowledge expansion within these groups. Consequently, we extend an emerging research stream addressing brokerage and brokering (Halevy et al., 2019; Kwon et al., 2020). Here, scholars distinguish between the structural argument – brokerage – that is focused on the position of knowledge brokers in the overall online network; and the processual argument – brokering –, where the actual process of knowledge sharing that contributes to knowledge expansion is central (Obstfeld et al., 2014; Soda et al., 2018). We address the digital affordances of the ESN that renders possible how employees may exhibit brokerage behaviour such as integrating knowledge (Grigoriou & Rothaermel, 2014), facilitating interactions (Lingo & O’Mahony, 2010), motivating cooperative behaviour (Nakashima et al., 2017), and transferring information (Obstfeld et al., 2014). The research aim is to gain understanding of the effect of bridging members on knowledge expansion in online groups within the context of an ESN.

To answer the research question, we develop and test two hypotheses. The hypotheses build on the core idea that on ESNs, group members have to bring in knowledge from outside of the focal group (Austin, 2000), in a digital environment where many ESNs offer the functionality to create and join online groups where knowledge can be posted and discussed. We argue that bridging members, who are members of multiple online groups simultaneously, are those who bring in knowledge from other ESN groups and as a result contribute to knowledge expansion. We build on the brokering behaviour that bridging members may exhibit, enabled by their position, and explain how the bridging members can exert their position to mitigate the decline in knowledge expansion over time. They are able to leverage digital technologies in ways that lead to the online expansion of knowledge. In particular, bridging members are able to profit from their position between groups where they are early to learn about and connect other groups’ activities (Burt & Soda, 2017). They are also optimally able to use the ESNs affordances of visibility, persistence, editability, and association (Treem & Leonardi, 2013), where knowledge can be shared in different ways than in offline settings (Howison et al., 2011; Leonardi, 2014; Moser et al., 2017).

In the first hypothesis, we address the duration, or functional age, of the group and the effect thereof on knowledge expansion. Groups that continue to share knowledge over time experience an increased sense of community (Preece, 2001), but also have to deal with a decline in knowledge expansion over time due to increased redundancy of already shared knowledge (Moskaliuk et al., 2012). In the second hypothesis, we argue that bridging members, i.e., people who connect otherwise unconnected online groups, have a mitigating effect on the decline in knowledge expansion over time. Taken together, we posit that bridging members are crucial for online knowledge flows, specifically for the expansion of the knowledge in online groups that persists over time.

We draw on ESN data from a Dutch health insurance company to test our hypotheses. We find that bridging members are key players in online ESN groups and that they mitigate the decrease in the expansion of knowledge that is shared in these groups over time. We contribute to the literature on online knowledge sharing and social networks by showing the effect of bridging members on knowledge expansion in a digital environment. In addition, we assume that bridging members leverage the affordances of visibility, persistence, editability and association that characterise digital platforms. We also provide practical implications for organisations using ESNs by shedding light on how online groups can benefit from their bridging members on such digital platforms.

Theoretical background

Enterprise social networks

An ESN is a digitally enabled social network situated within an enterprise. Usually, an ESN is an online platform accessible from a web browser. It is designed to support social relations amongst employees within an organisation, by providing the tools to share content in the form of messages or documents, and find answers to specific questions (Ellison et al., 2015). One of the core principles of an ESN is self-organisation, where people who share business interests can share knowledge on that topic with each other. This means that anyone with access to the online platform regardless of role, tenure, department, or physical location can start a group, invite others, reply to posts, or use other functionalities that are enabled in the ESN. Some participants may have specific roles such as moderator or administrator. In contrast to online social networks, like Facebook or LinkedIn, which are also accessible from a web browser and have similar features, an ESN is only accessible for those involved with the specific organisation, either as employees or for external people invited to specific groups. Consequently, when an employee leaves the organisation, it means that access to the ESN is withdrawn for this person. The documents and other content that is shared on an ESN, is secured and is accessible only for authorised users.

ESNs are of great importance and relevance to organisations. A 2013 report from Deloitte predicted that by the end of that year more than 90% of Fortune 500 companies would have partially or fully implemented an ESN, a 70% increase compared to 2011 (Lee et al., 2013). Bughin et al. (2015) found that 58% of companies surveyed in 2013 planned to make investments into ESNs over the next three years. Bughin et al. (2016) report that tools that can enable collaboration amongst employees are appreciated, since 80% of executives, up from 69% in 2014, stated that their companies used these tools for internal purposes. These reports indicate that ESNs are widespread in organisations, and are growing in numbers and usage.

Knowledge expansion

Organisational knowledge is an important organisational resource (Alavi & Leidner, 2016). Knowledge management in organisations can be seen as a process which refers to building a dynamic work environment and learning to foster the continuous generation, collection, and use of individual and collective knowledge to discover new values for the organisation (Patalas-Maliszewska, 2013). We yet have to fully understand the role of digital technologies in organisational knowledge (Faraj et al., 2011, 2016; Von Krogh, 2012). ESNs are often used to aid in knowledge management not only to collect already existing knowledge, but also to create new knowledge through collaborations between employees (O’Leary, 2016). In other words, ESNs foster knowledge expansion, i.e., acquiring, generating, and selecting knowledge. Knowledge expansion takes place through social interactions between employees of an organisation or between them and people from outside the organisation (Nonaka & Takeuchi, 1995) resulting in a larger organisational knowledge base. Knowledge expansion is one of three pillars of the process-based approach of knowledge management, the other pillars being knowledge codification and knowledge transfer (Davenport & Prusak, 2000). Knowledge expansion involves four basic steps: (1) putting into place the release of resources, for example, by creating R&D centres to generate knowledge, (2) acquisition or seeking of knowledge within the organisation, (3) linking individual people or departments, and (4) networking such as the creation of communities and professional networks (Davenport & Prusak, 2000). ESNs can play an important role in this process of knowledge expansion, because the four steps can easily be realised: the ESN is a resource where employees can share and find knowledge, link employees and the groups they engage in, and can be used as a community or professional network where knowledge can be generated. Technology is often adopted by lead users and passed along through networks, which eventually leads to new practices and becomes gradually part of the active knowledge capital of the organisation (Davenport & Prusak, 2000).

An ESN is often put in place to foster knowledge management within an organisation, however we still lack a thorough understanding of the underlying mechanisms of knowledge expansion in organisations by means of an ESN. While the amount of shared knowledge increases after every posted message or reply, it is likely that knowledge becomes redundant over time because the same knowledge is shared over and over again. Redundant knowledge is useful for various reasons, including easier interpretation (Ter Wal et al., 2016), community building (Gulati et al., 2012) and confirmation of group norms (Centola & Macy, 2007), but not so much for knowledge expansion (Sosa, 2011). This is because knowledge is expanded only when new knowledge is added to the group (Kim & Chung, 2020).

The role of bridging members on ESNs

The building blocks of ESNs are online groups which have group members, who are contributing knowledge to the group by posting messages and replying to existing posts. ESN participants can easily engage with multiple groups. Usually, there is no restriction of the number of groups that one can be a member of and contribute to. Participants who are active in multiple groups have knowledge about what is going on in each of these groups. That means that these bridging members become key players in the network of groups, as they are in a position to transfer knowledge from one group to the other (Borgatti, 2006).

Existing literature has discussed similar concepts, most notably knowledge brokers (Burt, 2005) and boundary spanners (Lee & Cole, 2003). These roles are labelled as key structural positions in networks (Long et al., 2013). Brokers can span boundaries, but not all boundary spanners play the broker role (Fleming & Waguespack, 2007). Where the broker might keep unconnected parties separated for her/his own benefit (Burt, 1992), which is also known as the tertius gaudens strategic orientation (Obstfeld, 2005), the boundary spanner brings others together to foster cooperation between otherwise unconnected parties (Cross & Prusak, 2002), also known as the tertius iungens strategic orientation (Obstfeld, 2005). Therefore, the difference between broker and boundary spanners is their motivation for how to deal with structural holes, i.e., gaps in the social network. The broker following the tertius gaudens orientation likes to control the structural hole for his or her own benefit, whereas the boundary spanner following the tertius iungens orientation closes the structural hole to facilitate coordination or knowledge flows across the hole (Long et al., 2013; Obstfeld, 2005).

Both the broker and the boundary spanner direct the flow of knowledge between separate people or groups. The bridging member fulfils a role that is similar to the broker or boundary spanner. The bridging member spans boundaries between online groups by posting messages or replying to others’ messages in both groups. In contrast to the broker and the boundary spanner, the bridging member has no preference for either keeping apart (broker) or connecting (boundary spanner) others. As such, a bridging member creates overlap between two or more groups. The dual membership leads to a structural role similar to a structural fold as defined by Vedres and Stark (2010). The digital affordances of an ESN facilitate bridging member’s behaviour that influences the flow of knowledge within and between both groups, where the latter has previously also been identified as team boundary spanning (Marrone, 2010). Depending on the number of bridging members that bridge two groups, the two groups can overlap in such a way that the overlap itself mimics a group in its own right where knowledge can flow within the group. Bridging members on an ESN occupy the ‘fold’ between two (or even more) groups which has been identified in relational sociology as an important network position (Nollert, 2010).

Structural brokerage and processual brokering

We argue that bridging members are, by structurally connecting groups and enabling knowledge flows between groups, simultaneously structural and processual brokers. The distinction between structural and processual brokerage is a recent advancement of the brokerage literature (Halevy et al., 2019; Kwon et al., 2020; Obstfeld et al., 2014; Soda et al., 2018). Here, scholars argue that prior work on brokers in networks has paid insufficient attention to the actual processes that brokers engage in, also known as brokering (Halevy et al., 2019). Prior work has established a connection between network position and knowledge sharing (Cross & Cummings, 2004; Gargiulo et al., 2009; Reagans & McEvily, 2003; Tortoriello et al., 2015, 2012). Building on the extant literature on knowledge sharing in digital environments, we believe that shifting the focus to the brokering activities undertaken for knowledge sharing helps to disentangle the effects of structure and process on knowledge outcomes.

First, we elaborate on structural brokerage with regards to ESNs. On an ESN, we define a bridging member as somebody who establishes a connection between two or more groups. Some online groups are active for a short period, for example, when the group is created for a one-time event, and some groups are active for a longer period. As a consequence, a member of an online group can be a bridging member for variable time periods. After bridging, the participant may be inactive for the next period and connect two groups again later, thus becoming a bridging member again. As such, the bridging member builds bridges between many groups within the ESN, but not necessarily all at the same time. This results in flexible intergroup structures. Therefore, ESN bridging members fulfil an important role in terms of structural position and degree of connectedness between groups, which changes the intergroup network. Having many connections to other groups is beneficial for the group in terms of its innovativeness (Ahuja, 2000; Zaheer & Bell, 2005) and knowledge expansion (Lecoutre, 2019).

Second, we develop the processual brokering argument and argue that bridging members can apply different brokering strategies. Some studies of the processes associated with knowledge sharing and knowledge expansion on digital platforms have pointed towards the important role of those most active in the network (Füller et al., 2007). Due to the digital affordances of an ESN, posted messages are immediately visible to many employees and retrievable for any member at any time. In an open online environment, knowledge is never exclusively shared with a single individual but rather with the community as a whole. This has important implications for brokering. Because bridging members occupy important structural positions in the network of groups, they are better aware of their options to act (Kwon et al., 2020), so they can intend to broker or span additional boundaries for the overall benefit of the group. In the case of a digital environment such as an ESN, we argue that a tertius iungens strategic orientation of the bridging member enables knowledge sharing (Obstfeld et al., 2014; Quintane & Carnabuci, 2016) by spanning boundaries that either connect new groups, or strengthen existing connections between groups. For example, they can initiate a bridge to introduce people from two groups, thus showing a tertius iungens brokering strategy (Obstfeld, 2005). In online groups, it is more difficult for a bridging member to keep groups and their knowledge separated by following a tertius gaudens strategy, because others can create the same bridge, and become aware of the shared knowledge in those online groups as well. Due to the evolving knowledge in the community, connecting activities are as important as connecting positions.

In the following, we develop hypotheses about the different mechanisms that influence knowledge expansion. The conceptual model is depicted in Figure 1.

Figure 1. Conceptual model.

Hypotheses

Group duration and knowledge expansion

The composition of groups on digital platforms can vary along several dimensions. One important dimension is group duration, i.e., a group’s functional age. In general, groups that are active for a longer period generate more posts and replies. While some posts and replies might contain roughly the same type of content, we expect that groups that are active for a longer period expand their knowledge over time, i.e., acquiring, generating and selecting knowledge, for two reasons. First, long-living groups have had more time to attract members who participate in the group and bring in new knowledge. Second, because online collaboration enables asynchronous interaction, there is simply more opportunity and available time for others to respond to already existing messages (Siemon et al., 2019).

Yet, many posts and replies do not necessary translate to knowledge expansion, as we will explain in the following. In ESNs, groups that are new or active attract more attention (Guo & Saxton, 2018). For example, new groups are visible on the homepage of an ESN so others become aware of it. Also, recent messages from different groups are often added to users’ activity stream (Guy et al., 2016), which is usually part of the homepage of an ESN to inform users of recent activity. Consequentially, because new messages are exposed to a greater audience this way, a positive feedback loop will kick in which reinforces knowledge consumption and collaboration (Kane & Ransbotham, 2016). This means that many groups are very active in their early stages and expand knowledge quite fast. However, participation can decline when the needs of members are rather functional instead of social, resulting in less activity in online groups over time (Wang & Fesenmaier, 2004).

Due to the variety of groups in terms of goals and knowledge repositories, the knowledge that is shared within groups differs from one group to another. How individuals within a group are connected to each other influences their knowledge sharing and seeking behaviour, because connected group members determine how trustworthy the group’s knowledge is (He et al., 2009). Active knowledge exchange creates a sense of community and as such adds to the knowledge expansion of the group (Davenport & Prusak, 2000). The downside is that these within-team relations may lead to in-group bias. This observation from social psychologists addresses the tendency of group members to systematically overvalue members and undervalue non-members (Tajfel & Turner, 1986). As a result, group members value their own knowledge more than that of members from other groups. This may result in groupthink and rejection of knowledge from others outside of the group (Janis, 1972). Such processes are reported in terms of convergence of language used for homogenous groups, leading to overestimation and recirculation of their knowledge (Van Swol & Kane, 2019).

The above described scenario restricts the inflow of new knowledge and reinforces commonly held beliefs. Group members may have increased awareness of each other’s knowledge and may steer other group members to focus on this body of knowledge. This further reduces the chances that the group will seek knowledge from sources other than their own (Hansen et al., 2005). Following these arguments, a within-group discussion will lead to a decline in knowledge expansion in the group over time.

Hypothesis 1: The longer an online group exists, the lower the average knowledge expansion will be.

Bridging members and their role in knowledge expansion

The decline in knowledge expansion over time within an online group might be mitigated through processual brokering of bridging members. This is because bridging members, through their connections with different groups, are exposed to different knowledge. Being connected to more other groups can also lower the risk of groupthink (Janis, 1972). Connections between groups increase the likelihood of group members to seek knowledge from outside of their group, which has a positive impact on knowledge expansion (Davenport & Prusak, 2000). These connections, or weak ties, provide access to information and resources beyond one’s immediate social circle (Granovetter, 1973) and can speed up the search for knowledge across groups (Borgatti & Cross, 2003). The role of weak ties is an important aspect in terms of knowledge expansion, not only to locate knowledge from outside of the group, but also to collaboratively construct new knowledge (Davenport & Prusak, 2000; Lecoutre, 2019). Over time, when more bridging members connect the same groups, weak ties might develop into strong ties. Through this stronger linking of groups, the knowledge sharing capacity between groups increases, and the knowledge within each of the connected groups might expand. Members who connect other groups with other members simultaneously, are together also better able in interpreting information from both groups (Ter Wal et al., 2020). This form of collaboration can help to determine the usefulness of new knowledge. Thus, knowledge expansion is explained by the structural position of bridging members, and how they utilise this position in terms their brokering behaviour (Halevy et al., 2019; Kwon et al., 2020; Obstfeld et al., 2014). This form of brokering increases the knowledge flows between, and through the influx of new knowledge also within, groups.

The intergroup network of the group members can be enlarged in two ways. First, groups can attract new group members that have connections outside of the group. Second, existing group members can create new connections by becoming active in other groups. In digital settings such as an ESN, people frequently connect with others they did not know before. Being used to connecting with unfamiliar others makes it easier to create weak ties that might develop over time into stronger ties. At the same time, online groups use text-based communication, which contributes to possible recirculation of knowledge within the group. This recirculation strengthens group norms and creates a sense of community (Doerfel & Moore, 2016). Over time, a shared language develops (Deichmann et al., 2020; Hansen, 2002; Van Swol & Kane, 2019). While these aspects are important for the group as a whole, they limit knowledge expansion at the same time. We argue that for groups to increase their possibilities for knowledge expansion, they can benefit from members who generate knowledge within the group itself; however, beyond a certain point they benefit from members who connect their group to others. This opens up possibilities to generate knowledge by combining knowledge from multiple sources. The four identified steps of knowledge expansion (Davenport & Prusak, 2000) are therefore important in different ways during the different stages of an online group.

As bridging members are active in other groups as well, they have more opportunities and motivation to contribute to their focal group, because they are aware of and take part in what is discussed in multiple groups. They are often more experienced in using virtual communities which makes them more capable to generate new knowledge (Hung & Cheng, 2013) because they know how to best use the platform’s digital affordances (Treem & Leonardi, 2013). This increases bridging members’ possibilities for interaction, cooperation, and engagement with those groups (Hansen et al., 2005). There are two main reasons for an increase in interaction, cooperation, and engagement, which should influence knowledge flow between, and thus subsequent knowledge expansion within, the groups. First, when people become bridging members, they are structurally embedded between two groups. This position in a digital setting such as an ESN allows them to share a group’s knowledge with members of the other group. When there are multiple bridging members in an online group, there are even more opportunities to share knowledge between groups. Consequently, the more bridging members a group has, the better knowledge can flow between the groups, thus expanding the knowledge of the group. Second, and following the tertius iungens approach, bridging members may invite others to become bridging members as well. This is because from the tertius iungens perspective, bridging members are inclined to close structural holes, i.e., connect otherwise unconnected groups. They do so by inviting people they know from other groups to join their own group, thus enabling new linkages. Taken together, bridging members enable the recombination of knowledge by bringing together different knowledge domains (Deichmann et al., 2020) which then leads to within-group knowledge expansion.

In addition, bridging members are in a position to optimally leverage the affordances of the digital environment, such as visibility, persistence, and association (Treem & Leonardi, 2013). Here, bridging members share diverse knowledge with a focal group. Such diverse knowledge may be new to the group as a whole, but the bridging member who shares the knowledge can reinforce and affirm that knowledge with the help of other bridging members that they are familiar with from other groups where they share membership (Ishiyama, 2016). In digital settings, bridging members can easily switch their attention between groups (Stuetzer et al., 2013). This makes digital settings suitable for bridging members to share new knowledge between their groups. Besides sharing new knowledge with the bridged groups early on (Burt & Soda, 2017), we expect that this knowledge needs to be discussed and negotiated more thoroughly within the groups than is the case with more familiar or redundant knowledge. In that sense, the bridging members can use the norms and the shared history to contribute to a shared language within the group (Doerfel & Moore, 2016). This may strengthen linkages between group members, because they can more easily relate to the subject, and expand and elaborate on it. Over time, more and larger overlaps between groups can develop because of the processual brokerage of bridging members. So as group duration increases, each of the different knowledge expansion steps may become more feasible when the proportion of bridging members increases as well. These arguments lead to the following hypothesis:

Hypothesis 2: The proportion of bridging members in an online group will have a mitigating effect on the negative association between group duration and knowledge expansion in such a way that the extent of knowledge expansion over time declines less strongly when a group has relatively more bridging members.

Methodology and setting

Setting

Our analysis is based on ESN data from a Dutch health insurance company with more than 2000 employees and 247 active groups. Our unit of analysis is the online group. Our data covers a period of almost two years, from the start of the ESN in June 2012 until and including February 2014, which coincided with the acquisition of the company that hosted the ESN. We had written approval to continue using the dataset we had collected so far, but could not get an export of the ESN data at a later point in time. The dataset comprises information about employees, groups, group membership (including the date when an employee joined a group), group threads, messages and replies to messages in threads (including a time stamp), thread summaries, the number of years the employees have worked within the organisation, ownership of groups, and starting dates of groups. The owner of the group could close the active thread and start a new thread within the group. Only one thread at a time could be active within a group.

Measures

Dependent variable: knowledge expansion

Inspired by previous work (Kim & Chung, 2020), we measure the average knowledge expansion by clustering the total number of unique meaningful lemmas. Lemmatisation is a normalisation and clustering process in which different morphological variants of a word are mapped into the same underlying lemma so they can be analysed as a single term or concept (Korenius et al., 2004). A lemma is the canonical form, dictionary form, or citation form of a set of words. In English, for example, ’run’, ‘runs,’ ‘ran’ and ‘running’ are forms of the same lexeme, with ‘run’ as the lemma. By reducing the total number of distinct terms, lemmatisation decreases the complexity and ambiguity of the analysed text, and therefore improves and facilitates text processing (Liu et al., 2012). We only include meaningful lemmas, therefore we exclude stop words, articles, and punctuations. The text messages in the groups are in Dutch. For the Dutch language, the lemmatisation process was performed using a morphosyntactic tagger and dependency parser for Dutch called Frog (Van den Bosch et al., 2007). This tool analyses Dutch text and produces FoLiA, a specific XML format for linguistic annotation (Van Gompel & Reynaert, 2013). With this format, we could extract all meaningful lemmas and operationalise the dependent variable. Meaningful lemmas extracted through clustering are an important measure of knowledge expansion, because they are used to recognise similar knowledge with high precision (Kim & Chung, 2020). We use the average number of meaningful lemmas per day as a proxy for group knowledge expansion.

Independent variable: group duration

This is a count variable and measures the duration of the group in days.

Moderator variable: proportion of bridging members

This variable measures the proportion of bridging members as the total number of bridging members divided by the total number of active group members. A bridging member is considered as such when s/he is active in the focal group and at least one other group during a period of 30 consecutive days. Individual people are counted only once in calculating this variable; that is, although a bridging member might bridge between two or multiple groups, we count this individual only once. As a result, the proportion of bridging members does not exceed 100%.

Using this window of 30 consecutive days is based on the assumption that transferring knowledge from one group to another must be done within a given timespan, because the relevance of the shared knowledge declines over time (Reagans et al., 2015). In line with the literature on deriving social network measurements from trace data which states that the choices in relation to time sensitivity of the accumulation of data are crucial to the inference of a network structure (Howison et al., 2011), we made the decision for the 30-day cut-off point. Using this cut-off point results in an average proportion of bridging members of 66%. This percentage corresponds with other studies in different settings. In a study of offline knowledge networks, about 68% of the people were identified as brokers (Kauffeld-Monz & Fritsch, 2013). In another study on collective online brokering with regards to learning (Chen et al., 2010), two thirds of participants were brokers in that setting. This is similar to the percentage reported in a study by Fleming et al. (2007) about innovation by collaborative brokerage in patenting. Their data show that 67% of the collaborators acted as brokers. We ran robustness checks to see whether our results would change with different cut-off points. Running all regression analyses with cut-off points of 5, 10, 15, 30, 60, and 90 days, and even without a cut-off point, produced similar results. In the situation reported here, using a 30-day cut-off point, every group has at least one bridging member, and in our final sample we observe an average proportion of bridging members of 66%.

Control variables: group size

This is the total number of active group members. In general, larger groups are more active with more posts and replies, which can have positive effects on knowledge expansion. But group size can also have negative effects. Large online groups suffer from social loafing problems (Piezon & Donaldson, 2005), which can reduce participants’ motivations to contribute (Wang et al., 2013) and fosters free-riding and inactivity. Also, in larger groups, there can be less interaction between members, as well as lower quality of interaction (Kirkman & Shapiro, 1997), and cooperation tends to decline as a group grows in size (Kerr & Bruun, 1981). In smaller groups, members feel their contributions are more important and relevant because their posts are more visible, and as a result members of smaller groups engage more in the discussions (Piezon & Donaldson, 2005). Connected groups. This refers to the number of groups the focal group is connected to through its bridging members. The same proportion of bridging members can connect only one other group, when all bridging members only connect to the same other group, or many, when the bridges to other groups are more distributed. The higher the number of connected groups, the more opportunities the group might have to expand its knowledge (Oh et al., 2006). Tenure. This control variable refers to the average tenure in the organisation of the group members measured in years. Longer tenured employees can be expected to be active in more online groups because on average they would know more colleagues and be more confident to speak up, and as a result form a bridge between more groups.

Analysis

We analysed 247 groups using linear regression analysis to test Hypothesis 1 and 2. We investigated the variables for skewedness and found that all variables except tenure were skewed. One of the assumptions of linear regression analysis based on ordinary least squares (OLS) is that the estimation errors are homoscedastic, meaning equally variable conditioned on the predictor variables (Hayes, 2012). Therefore, we report significance levels based on heteroscedasticity consistent standard errors to control for any residual heteroscedasticity (Long & Ervin, 2000), using the HCREG macro from Hayes and Cai (2007).

Results

In Table 1, we report the means and standard deviations of the measures as well as their correlations. Correlations are generally not high, except for connected groups with duration (0.73), but some are significant; therefore, we checked the variance inflation factors (VIF’s) of all variables. The maximum value was 3.44, well below the recommended maximum value of 10 (Field, 2009).

Table 1. Descriptive statistics and correlation matrix

Variable		Mean	S.D.	Min.	Max.	1		2		3		4		5
1.	Group size	13.47	27.21	1.00	352.00
2.	Connected groups	29.98	34.94	1.00	97.00	0.51	**
3.	Avg. tenure (yrs)	8.81	4.87	0.00	28.57	0.11		−0.02
4.	Avg. knowledge expansion per day	15.48	20.40	0.52	131.50	−0.03		−0.23	**	−0.10
5.	Duration (days)	98.44	125.75	2.00	348.00	0.24	**	0.73	**	0.06		−0.39	**
6.	Bridging member proportion (BM-pr)	0.66	0.25	0.08	1.00	−0.03		0.42	**	−0.20	**	−0.20	**	0.33	**

N = 247, ** p < 0.01, * p < 0.05

Hypothesis 1 states a linear and negative association between group duration and knowledge expansion. Model 2 in Table 2 shows the negative and significant direct effect (β = −.08, SE = .13, p ≤ .001, R² = .17), which indicates that for groups that have existed longer, the average knowledge expansion declines. Thus, hypothesis 1 is supported. Hypothesis 2 states that the proportion of bridging members mitigates the negative association between group duration and average knowledge expansion. Model 4 shows the result for the interaction effect for the proportion of bridging members on the association between duration and knowledge expansion (β = .22, SE = .07, p ≤ .01, R² = .22). The result of this interaction effect is plotted in Figure 2 and shows that for online groups that have a high duration, knowledge expansion decreases least for groups with relatively many bridging members. In addition, Figure 3 depicts the conditional indirect effect for different levels of the proportion of bridging members, ranging from minus one standard deviation to plus one standard deviation from the mean. However, for groups that have a short duration, knowledge expansion is highest for groups with relatively few bridging members. Hypothesis 2 is supported; however, the results show some nuances in regards to the observed linear relation, because groups that have a short duration exhibit higher average knowledge expansion when the proportion of bridging members is low.

Table 2. Results of regression analyses predicting average knowledge expansion for group per day (New Lemmas)

Variables		Model 1			Model 2			Model 3			Model 4
Constant		23.76	***	(2.87)	23.67	***	(2.74)	30.82	***	(6.65)	39.35	***	(8.27)
Group size		0.10		(0.24)	0.04		(0.19)	0.01		(0.17)	0.06		(0.24)
Connected groups		−0.17	*	(0.08)	0.05		(0.08)	0.10		(0.08)	0.03		(0.11)
Avg. tenure (yrs)		−0.50	*	(0.21)	−0.32		(0.20)	−0.40		(0.21)	−0.37		(0.21)
Duration (days)					−0.08	***	(0.13)	−0.08	***	(0.01)	−0.25	***	(0.06)
Bridging member proportion (BM-pr)								−11.08		(7.90)	−21.97	*	(10.09)
Duration x BM-pr											0.22	**	(0.07)
R2		0.07			0.17			0.18			0.22
F		6.66	***		12.21	***		10.66	***		11.52	***
∆R2					0.09			0.01			0.04
∆F					26.73	***		3.87			13.17	***

N = 247 groups

*** p ≤.001, ** p ≤.01, * p ≤.05

Standard error in parentheses

Figure 2. Curvilinear moderating effect of proportion of bridging members on the linear group duration – knowledge expansion relationship (two-way interaction with quadratic moderator).

Figure 3. Conditional indirect effect for moderating effect of bridging member proportion.

Discussion

In this paper, we demonstrate that people who bridge online groups in digital environments mitigate the decrease of their groups’ knowledge expansion over time as measured by the meaningful lemmas used in the groups. In online groups that exist for a short period, knowledge expands more in a situation with relatively few bridging members. For long-living groups, however, having more bridging members is beneficial. In the latter case, the negative effect of group duration on knowledge expansion is mitigated by the number of bridging members in the online group. For both cases, the visibility and the type of knowledge shared by bridging members might drive the underlying mechanism. This is because bridging members are less prone to recirculate redundant knowledge as they are in a position to access and circulate knowledge that is new to at least some of their groups. This might be explained by the use of more diverse words in online groups by high status members (Reysen et al., 2010). In addition, bridging members can be more aware of how to articulate their insights obtained from other groups, because they are better able to combine knowledge from the multiple groups that they are linked to (Hargadon, 2002).

Theoretical implications

Past network research has demonstrated the importance of balancing two types of members in a group. In terms of creativity and innovation, a group benefits from members that have connections to other groups, which increases their exposure to diverse opinions and practices (Burt, 2009; Burt & Soda, 2017). These members enable the flow of knowledge between groups and are considered brokers. In terms of alignment and efficiency, however, a group benefits from having tightly connected members. This duality of brokerage and closure is important for groups because it can help them to increase their social capital (Burt, 2009) and effectiveness (Oh et al., 2004). This study adds to this literature by highlighting the dual role of bridging members and the consequences for knowledge expansion of their group, which enables both the recombination of knowledge (Nonaka, 1994) and creates new linkages between knowledge domains (Deichmann et al., 2020; Mannucci & Yong, 2018). In particular, we theorise and empirically show that more connections between groups, i.e., a larger proportion of bridging members, mitigate the decrease in the expansion of knowledge that is shared in the focal group over time.

We hypothesised that a larger proportion of bridging members would mitigate the decline in knowledge expansion over time, which is supported by our analysis. However, we also found that groups that only exist for a short period expand more knowledge when there are relatively few bridging members active in the group. This finding is important as it demonstrates the joint influence of two mechanisms on knowledge expansion: the duration of the group, and the proportion of bridging members in the group. Groups that only exist for a short period expand more knowledge when there are relatively few bridging members active in the group. One explanation for this finding is that group members are fairly comfortable discussing online with people that they know better. In this situation, people are closely connected and together form a ‘clique’ (Krackhardt, 1999). Their close connections increase stability and contribute to a shared understanding (Tortoriello & Krackhardt, 2010) between the group members. As bridging members are also part of this clique, they are trusted sources of external information. We previously argued that when bridging members are active in multiple groups, they are more able to utilise the digital affordances of an ESN and contribute knowledge, and, in turn, motivate others to do the same. This brokering behaviour of the bridging member might be a key explanation in why the decrease in knowledge expansion can be mitigated over time. This is relevant because it takes time to increase the proportion of bridging members in an online group. When, over time, the group attracts or seeks more bridging members relative to the total number of members, the knowledge within the group can expand and the continued input of new knowledge is likely to keep all members engaged. This relates to a comparable finding where structural folds, or the overlaps between groups, have a positive effect on group creativity when values do not diverge (Lehtimäki et al., 2020).

With this study, we contribute to an emerging research stream addressing brokerage as a process (Obstfeld et al., 2014; Soda et al., 2018), or brokering (Halevy et al., 2019; Kwon et al., 2020). Specifically, we argue that ‘structural brokers’ are ideally positioned to increase knowledge flows between groups; and that ‘processual brokerage,’ i.e., how these bridging members leverage affordances of digital technologies in an effort to expand knowledge, is a necessary activity to realise this increase. The research thus shows that structural and processual arguments work in tandem in explaining knowledge sharing outcomes on digital platforms. We also assume that bridging members leverage the affordances of visibility, persistence and association (Treem & Leonardi, 2013) that characterises such digital platforms.

Our research also has implications for research on multiple team membership. The common notion is that multiple team membership has negative consequences for team performance and productivity (Baralou & Tsoukas, 2015; Bertolotti et al., 2015; O’Leary et al., 2011; Pluut et al., 2014). In contrast, in our setting, we find that over time, multiple group membership has a positive effect on the expansion of knowledge that is being shared in the groups. We believe that our findings might be explained by the digital environment as well as the voluntary character of contributing to online groups on an ESN. First, in traditional, i.e., offline, settings, people who are members of multiple teams soon meet their limit in terms of cognitive load and role strain. Becoming part of too many teams or groups will soon lead to cognitive and emotional overload, instead of increased performance. Attention becomes fragmented, and the constant switching between different roles and responsibilities can become too much of a burden. In digital settings, however, bridging members can more easily switch their attention between groups (Stuetzer et al., 2013). This increases the connectivity between groups and enables groups to keep expanding their knowledge (Davenport & Prusak, 2000). In addition, digital communication is more distant and easier than face-to-face communication in offline environments, which might be an additional advantage of the digital environment. Second, contributions to digital platforms such as our ESN is, in most organisations, still voluntary. Employees have no formal responsibilities with regards to their online messages or replies; i.e., their online behaviour is not being monitored as opposed to their behaviour and performance in other respects. Employees, and especially bridging members who are typically highly motivated, might, therefore, enjoy their digital participation more than their other work routines.

Practical implications

Our study has implications for managerial practice. Organisations increasingly use ESNs (Dimicco et al., 2008; O’Leary, 2016), and also ESNs require substantial investments of time and money when being implemented. Bughin et al. (2015) found that 58% of companies surveyed in 2013 planned to make investments into ESNs over the next three years. Bughin et al. (2016) reported that tools that can enable collaboration amongst employees are appreciated, since 80% of executives, up from 69% in 2014, stated that their companies use these tools for internal purposes such as knowledge sharing. One of the benefits of these digital environments is that sharing knowledge is much easier than in offline environments. However, organisations and their employees keep struggling to reap these benefits (Chin et al., 2015; Wehner et al., 2017). Our study shows that bridging members are important in connecting online groups in terms of knowledge expansion. This is because they enable groups to share more and more diverse knowledge by mitigating the decline in knowledge expansion in groups over time. These insights are beneficial for organisations that make use of ESNs, for consulting organisations that assist in implementing ESNs in other organisations, and for ESN manufacturers or similar digital platforms. Our insights may help them to find ways to increase the number of bridging members in groups that are active for a longer time, and distribute them in such a way that knowledge expansion is maximised.

Limitations

Some limitations of our research provide fruitful avenues for future research. First, there is reason to assume that some bridging members building on their brokerage behaviour create sequentially bridges to other groups over time (Sasovova et al., 2010). New entrants, and in particular new bridging members, can change the subject of the group, add to its knowledge expansion, and change the structure of groups (Summers et al., 2012). Future research might therefore take into account the changes that might occur in how groups are connected by bridging members over time, in particular the way in which bridging members engage with the internal network structures of groups. Also, individual bridging members leaving groups and as a result stop acting as a bridge between groups, might have specific consequences. At a later point in time, these members might return to the online group and function as a bridge again. How these specific dynamic impacts knowledge exchange in online groups is new ground to explore. Another suggestion is to look into the bridging behaviour of the bridging members in terms of the actual content that they share over time, to be able to assess the quality of the knowledge instead of the sheer expansion of knowledge (De Luca & Cano Rubio, 2019). Bridging members mitigate knowledge expansion over time; however, if and how knowledge expansion relates to knowledge quality is unclear.

Conclusion

Our study shows that bridging members, i.e., people who connect otherwise unconnected groups in a digital environment, over time can mitigate the decrease in the expansion of the knowledge that is shared within these groups. Using data from a large Dutch health insurance company, we find that bridging members generally increase knowledge expansion in online groups. However, in the early stages of an online group this effect disappears: younger groups expand more knowledge when the proportion of bridging members is smaller. Only after some time do these groups benefit from having relatively more bridging members to expand knowledge. We argue that structural and processual brokerage arguments are simultaneously at play in knowledge sharing outcomes on digital platforms, and that bridging members may be able to leverage the affordances that characterise such digital platforms. Future research should further investigate the interplay of structure, process, and affordances of digital platforms to shed light on the social mechanisms that underlie the creation and diffusion of knowledge in organisations.

Disclosure statement

No potential conflict of interest was reported by the author(s).