Abstract
Remote and hybrid workers know fewer of their colleagues and have fewer strong workplace relationships. If strong relationships support knowledge sharing, workers will have a harder time getting knowledge they need. Prior research shows that digital communication technologies increase workers’ network-level knowledge of “who knows what” and “who knows who.” Yet, knowledge seekers may be hesitant to ask for knowledge, particularly when they have concerns that their relationship with a knowledge source is too distant. We conduct a dyad-level study of 141 instances of knowledge seeking among employees of a South American telecommunications company employing a hybrid work arrangement and using an enterprise social media called Chatter. We find that specific uses of the technology help develop what we call “relational confidence,” or the confidence that one has a close enough relationship to a colleague to ask and get needed knowledge. With greater relational confidence, knowledge sharing is more successful.
Lay Summary
By virtue of making relationships and expertise more visible, enterprise social media (ESM) platforms help increase workers’ knowledge of “who knows what” and “who knows who” across their organization. Even so, whether people act on this and actually ask a coworker for the knowledge they need is less clear. Knowing a coworker could be helpful is not the same as knowing a coworker would be helpful. People are in fact less likely to ask for knowledge if they do not feel confident their coworker would be helpful, which hampers productive knowledge sharing and stymies organizational learning. In this article, we find that interacting through an ESM platform just before asking someone for knowledge can help knowledge seekers build “relational confidence” in their coworker. With greater confidence, knowledge seekers are more likely to ask for knowledge and are more satisfied with the knowledge shared with them. The findings demonstrate that ESM can lubricate knowledge sharing in the time before people decide to ask for knowledge by helping them build relational confidence.
The rapid increase in remote and hybrid work across all manner of knowledge-intensive organizations has shifted how, when, and to what extent employees communicate with each other. Recent research on remote and hybrid work arrangements during the COVID-19 pandemic has shown that it is common for workers who increase their frequency of working remotely to strengthen ties with their immediate workgroup members and either weaken or abandon ties with individuals in other parts of the organization (Wu et al., 2021; Yang et al., 2021; Zuzul et al., 2021). Thus, some scholars speculate that in remote and hybrid work arrangements, workers will have fewer strong ties with individuals outside of their immediate workgroups but with whom they would benefit from receiving or sharing knowledge (Arena et al., 2022; Yang et al., 2021). The logic underwriting this speculation is based on empirical evidence showing that knowledge sharing is particularly instrumental when people connect with and learn from those in distant parts of the organizational network, which typically means with people in different workgroups, departments, divisions, and business units than their own (Borgatti & Cross, 2003; Hansen, 1999).
Of course, not all knowledge is created equal. Because it is easily documented and explained, explicit knowledge that is low in complexity can be easily communicated even to distant colleagues. Typically, a knowledge source does not have to devote much time to share knowledge that is not complex with a knowledge seeker; there is no major “ask” to be made that must be supported by a robust relationship between knowledge seeker and source (Hansen, 1999). Knowledge that is highly complex, which is often somewhat tacit in nature, generally cannot be easily communicated or explained. The knowledge seeker must devote time and effort to share such knowledge with a source, and the existing research shows that they are much more likely to be willing to do so if they report having a close relationship with the knowledge source (Levin & Cross, 2004). For this reason, prior studies have found support for a “matching hypothesis” (Reagans & McEvily, 2003, p. 262), in which knowledge sharing is perceived by those involved to be most successful when the closeness of their relationship matches the complexity of the knowledge to be shared: As knowledge grows more complex, the stronger the relationship between two individuals must be for knowledge sharing to be effective. The fact that recent evidence suggests people working remotely have fewer of the strong ties that reach out to distant parts of the organization may pose problems for sharing important knowledge.
One of those problems concerns a knowledge seeker’s propensity to ask for the knowledge they need. Because asking for knowledge from a person who is the source of knowledge can put the seeker in a vulnerable position, seekers tend to feel most confident asking a source with whom they have a close relationship (Borgatti & Cross, 2003). Asking for knowledge without knowing if the source will act benevolently can be daunting because the act of asking can reveal the seeker’s own inexperience or lack of expertise, which can damage their reputation (Abrams et al., 2003). Because the number of close workplace relationships appears to decline in remote and hybrid work arrangements, knowledge seekers will have fewer people in their networks from whom they feel confident that they can get the knowledge they need.
Recent research has documented the proliferation and rapid uptake in use of digital technologies for communication like Chatter, Microsoft Teams, Slack, and Basecamp—platforms that use embedded social features to enable structured and informal team-based interaction—during the pandemic (Aloisi & De Stefano, 2022; Leonardi, 2021). These technologies have been shown to help individuals in organizations more easily and accurately develop insights into the relevant knowledge that far-flung colleagues may have because anyone can see the communication occurring among others (Ellison et al., 2015). Consequently, in remote and hybrid work arrangements in which the entire organization is making use of digital technologies for communication, it may be easier than ever to find people from different parts of the organization with relevant knowledge, but knowledge seekers may still be reluctant to ask for that knowledge when they do not believe that the relationship they have with the source is close enough to generate a positive response (Kane et al., 2014).
Drawing on the theory of communication visibility (Kim et al. 2019; Leonardi, 2014, 2015; Rice et al., 2017; Treem et al., 2020a), we propose that the affordances of digital technologies broadly, and enterprise social media (ESM) more specifically, are not only useful for helping people find the knowledge they need in general (at the network level), but that they are also well-suited to helping knowledge seekers build the confidence to ask for particular knowledge (at the dyad level). We theorize that the visibility of communication on ESM helps knowledge seekers signal to a particular knowledge source in anticipation of asking for knowledge, by, for example, following a common group or file to demonstrate shared work-related interests and knowledge relevance. We also theorize that the visibility of communication on ESM helps knowledge seekers learn about the knowledge itself, by observing the knowledge source’s conversations, files, and work groups. We hypothesize that for these reasons, ESM communication with the source in anticipation of asking for knowledge will help the seeker feel more confident they can get the knowledge they need and ultimately, that knowledge sharing will be more successful. We find support for our theory using survey data on 141 instances of interpersonal knowledge sharing across 12 units of a medium-sized South American telecommunications company that employed a hybrid work arrangement in which individuals could choose to work remotely or from any office location in four different countries. The results demonstrate that specific, dyad-level uses of the ESM helped seekers to develop what we call “relational confidence,” or the confidence that one will get knowledge they want from a particular communication partner if they ask for it, and that in such cases, knowledge sharing was typically effective.
Beyond metaknowledge: the role of relational confidence
The theory of communication visibility proposes that digital communication technologies can help make people’s communication and interactions with others visible to a broader audience. In this theory, communication is not limited to a conceptualization of active interaction between individuals, but also includes passive activities such as liking someone’s content or following someone’s post-indirect communication that signals interest and affiliation between actors. Research in this tradition has mounted significant evidence that communication visibility is positively associated with increases in metaknowledge. Metaknowledge is typically defined as network-level knowledge about “who knows what” and “who knows whom” and has been shown, empirically, to be a valuable precursor to knowledge sharing (Ren & Argote, 2011), meaning “individuals sharing organizationally relevant information, ideas, suggestions, and expertise with one another” (Bartol & Srivastava, 2002, p. 65). In his development of communication visibility theory, Leonardi (2014, 2015) showed that use of ESM increased people’s accuracy at identifying “who knows what” and “who knows whom” by 31% and 88%, respectively, among members of their organization. Importantly, the visible nature of communication on the platform was the reason that most individuals showed such dramatic increases in the accuracy of their metaknowledge. Communication visibility allowed them to learn about the networks and knowledge of colleagues that worked in distant parts of the organization from their own and who they did not know well. In recent years, a number of other studies have validated these early findings about the link between communication visibility and metaknowledge accuracy, showing that communication visibility is useful because individuals are able to make inferences about one’s knowledge based on the fact that these technologies make other people’s messages transparent and their networks translucent (Engelbrecht et al., 2019; Liang et al., 2022; van Osch & Steinfield, 2018; van Zoonen & Sivunen, 2020; Zhao et al., 2020).
Although there is strong evidence that communication visibility can help people develop more accurate metaknowledge, and that they can eventually use this metaknowledge to improve knowledge acquisition, prior studies have overlooked how individuals get the knowledge they need once they identify that someone in their organization has it. In other words, it is unclear how knowledge seekers convert their newly acquired metaknowledge into knowledge acquisition, particularly when most people whom they ask are outside of their immediate workgroups and are often not close colleagues (Engelbrecht et al., 2019; Treem et al., 2020a).
Relational confidence and anticipatory communication
Knowledge seekers face a significant obstacle to getting the knowledge they need when they feel that they do not have a close enough relationship with a source, and so they lack the confidence that the source would be willing to share knowledge if asked. In such cases, most prior scholarship suggests that seekers should invest in building a closer relationship with the identified source over time by arranging to meet face-to-face (Abrams et al., 2003), by leveraging physically proximate mutual connections (Reagans & McEvily, 2003), and by establishing norms of reciprocity (Wasko & Faraj, 2005). After such an investment, the source may be more willing to explain the knowledge originally desired. Yet, such suggestions are not practical when people are working remotely and do not have ample informal occasions during which to build those relationships (Neeley, 2020). At the same time, asking the source for knowledge without a sufficiently close relationship is risky (Szulanski, 1996). Asking for knowledge means asking a source to dedicate considerable time to explain that knowledge; sharing knowledge requires “time devoted to personal interaction or thoughtful documentation of one’s expertise, or both” (Hinds & Pfeffer, 2003, p. 16). Therefore, asking for knowledge without a sufficiently close relationship may result in a rejection. Given the potential negative repercussions, it seems that in cases where it is not possible for seekers to ask a close source for needed knowledge, asking a weaker and more distant source for that knowledge requires a certain amount of confidence.
Confidence, broadly defined, is the feeling of self-assurance arising from an appreciation of one’s own abilities. At work, belief in one’s own capabilities to take action and respond to demands effectively, known as one’s self-efficacy, is positively related to individual performance (Gist, 1987). People with greater self-efficacy are more willing to take risks, which in turn helps them take advantage of opportunities and overall makes them more successful (Krueger & Dickson, 1994). We theorize that people who desire knowledge from someone they perceive as unwilling to explain that knowledge to them may take action to increase their relational confidence before they ask them for the knowledge. We define relational confidence as the feeling of self-assurance arising from an appreciation that one has a close enough relationship with a communication partner that they can ask and receive from them the knowledge or resources they need. Operationally, relational confidence is the knowledge seeker’s perceived surplus or deficit of relational closeness with a desired source of knowledge (where a perceived surplus means the knowledge seeker feels more than close enough to the knowledge source to get the knowledge they want, and a perceived deficit means the knowledge seeker feels not close enough). The evidence suggests that knowledge seekers generally recognize that they are more likely to successfully receive complex knowledge from a strong tie than they are from a weak tie (Krämer et al., 2014). Thus, if a knowledge seeker perceives that their relationship with a knowledge source is too weak to justify asking for complex knowledge (regardless of whether that is true or not), they may act differently than if they felt the relationship was strong enough. Because people’s perceptions of their social networks drive their communicative action (Ellison et al., 2007; Reiners & Alexander, 2013), a knowledge seeker may act to improve their relational confidence, meaning increase their own belief that the knowledge source would provide the needed knowledge if asked (which may or may not mean the knowledge source changes their perceptions as well). Thus, relational confidence is closely tied to the knowledge seeker’s own perceptions about their relationship with the source given the type of knowledge needed from them.
The visibility afforded by ESM may be ideal for developing relational confidence. Through the use of ESM, it is possible to learn more about what types of knowledge a colleague has, the types of work tasks or projects in which he/she/they have been involved in the past, or in which work groups the source participates (Kim et al. 2019; Wu, 2013). Such work-related visibility about the colleague can be useful also because it contains “identity information” (Ellison et al., 2015, p. 10). People feeling low relational confidence in a colleague may wish to gather information about the colleague’s projects or work tasks in anticipation of asking for knowledge, so they can have knowledge examples or details at hand. Information about a person’s workplace identity and tasks stored on social networking sites have been conceptualized as useful for quickly locating experts in a particular area (Ellison et al., 2015) or understanding others’ projects and work-related activities (Kim et al., 2019). This information gleaned from visible activity on ESM can become the fodder used for communicating with a knowledge source from whom one is anticipating asking for knowledge. “We define anticipatory communication as the communication a person engages in with another in advance of, and with the goal of, asking that person for knowledge or other resources.” Based on this definition and the potential relationship it portends with relational confidence, we conjecture that ESM are useful for people who want to learn a bit more about a colleague before asking them for knowledge.
H1: The lower a seeker’s relational confidence with a source who has needed knowledge, the more frequently the seeker uses ESM for anticipatory communication with the source.
Building relational confidence
In addition to learning about a colleague’s work, knowledge domains, and identity, people can use ESM to make their presence more salient in their colleague’s social network. The visibility and persistence of communication on the platform means that people can communicate in a way that leaves digital traces available for others to see (Kane et al., 2014). Anticipating that knowledge sources will see these visible traces of communication and interpret them as signals of a strong relationship, seekers may feel that a digital trace of notifications such as comments, joins, or edits will increase their visibility in the colleague’s network and may feel that traces make the colleague more likely to expect the knowledge request. People may even choose to engage in a way that leaves a digital trace visible widely to others across the site. Burt and Knez (1995) found that when two people communicate in public, where there is a third-party witness of the conversation, the two parties become concerned about what the third party will say about the interaction. Specifically, they found the awareness of third-party observers induces people involved in the communication to act consistently in the future with the way they acted in the interaction with the third-party observer. That is, as Cialdini (2001) describes, when people are observed engaging in actions in public, versus doing those same actions in private and unseen by others, they feel more concerned about how others will interpret their actions. This suggests that seekers may engage in communication in anticipation of asking for knowledge from a source that is strategically designed to increase the visibility of their interactions to third parties on ESM such that it is more likely the source pays attention to those interactions and feels a heightened sense of commitment to share the knowledge when asked.
We proposed earlier that the lower seekers’ feelings of relational confidence in a source, the more they will use ESM for anticipatory communication to learn about the source (H1). More than that, the existing literature suggests that anticipatory communication can leave public visible traces that the source may evaluate as signs of a strong relationship with the seeker. Of course, the source may never notice these communications from the seeker or interpret and reflect on visible interactions with the seeker as signals of a strong relationship. Nevertheless, we contend that the seeker will perceive that the source may take such activity as a signal of their interest and closeness. And because relational confidence is linked to the knowledge seeker’s perception of the source’s willingness to share particular knowledge, engaging in these activities is likely to lead the seeker to believe that their closeness has increased, and perhaps enough to facilitate the sharing of the knowledge the seeker desires. Thus, we propose:
H2: The more a knowledge seeker uses ESM for anticipatory communication with the source, the more they increase their feelings of relational confidence with the source.
Anticipatory communication on ESM and satisfaction with knowledge sharing
The visibility of communication afforded by ESM may also improve the outcome of knowledge sharing once a seeker asks for knowledge. The more complex the knowledge sought, the more difficult it is for a seeker to figure out exactly what questions to ask about it. But when a seeker knows about a specific project or document, they can simply ask: How did you do that? (Ambrosini & Bowman, 2001). The visible traces of someone’s work posted on ESM is a type of boundary object, or reference that can serve as a launching point for discussions and can make it easier for someone to ask pointed follow up questions or clarifications for deeper understanding (Carlile, 2002). A seeker’s ability to ask referential questions about needed knowledge can reduce the burden of explaining knowledge, perhaps by reducing the amount of time it takes to share the knowledge or at least by making the interaction more pleasant because it is clear what knowledge is sought. For these reasons, we suspect that the learning about a source’s knowledge through their communication made visible on ESM will improve the success of the knowledge request.
Moreover, studies have found that people draw on evidence of communication frequency on ESM as they quantify and evaluate relationships. According to Donath (2007, p. 238), “public comments and other communication [on digital collaboration platforms] … signal the strength and context of a relationship.” That is, the visible and persistent record of communication between parties on digital collaboration platforms can be interpreted as a relationship narrative, and in the relationship narrative, the content, frequency, and length of messages between users signal closeness. Lampe et al. (2012) found that the quantity and quality of interactions between individuals on ESM were a reliable signal of the resources that the people involved are willing to devote to the relationship in the future. As they explained, back-and-forth communication in a visible medium, “serves as a signal of relational investment and highlights their presence in that person’s network … and may increase the likelihood that those network members will respond to resource requests in the future” (p. 3197). Therefore, seekers may believe the digital traces they create on the platform will be observed by a source and that viewing visible evidence of interactions can create a sense of commitment in the source to the relationship in proportion to their content, frequency, and length. This perception of increased commitment can, perhaps, encourage seekers to become more willing to admit their own inexperience, ask more follow up questions, or request details or know-how than they would not have otherwise, without fear their colleague will reject or deride them. For all these reasons, we hypothesize:
H3: The more a knowledge seeker uses ESM for anticipatory communication with the source, the more the knowledge seeker will be satisfied with the knowledge received.
Methods
Sample and data collection
To test our hypotheses, data were collected from employees in two divisions (Divisions A and B) of a large South American telecommunications company that we call TeleMobile (a pseudonym). The two divisions employed 81 people, across 12 teams (six in each division: finance, operations, accounting, human resources, marketing, and IT). Although the headquarters was based in Lima, Peru, the company utilized a hybrid work arrangement in which individuals could choose to work remotely or from any office location in four different countries: Peru, Ecuador, Bolivia, or Uruguay. TeleMobile did not mandate that employees come to the office on any particular day, though every few months mandatory in-person staff meetings were held. On average, employees worked 2 days a week in their home-country office and 3 days a week at home. A few times a year (6 on average), employees would work from an office in a different country. To enable this hybrid work arrangement, TeleMobile’s senior leaders adopted Salesforce’s ESM, Chatter. Chatter combined social networking features with structured channels for collaboration among team members. Users develop profiles that provide work-related information on individual expertise and projects. The ESM also had a news feed that recorded time-stamped communications in conversational threads visible to all users in the company. Chatter users could “follow” files or groups in the organization, and the site also has algorithms that recommend relevant people, groups, and files. The ESM also allowed individuals to form groups or channels that would contain directed streams of work communication that were documented in time-stamped threads. Employees often joined channels for both their own work projects and broader company initiatives. All content on the ESM was available to all employees at the company. Employees of TeleMobile had access to other modes of communication, including email, instant messaging, phone, and face-to-face communication. Additionally, the work at TeleMobile was knowledge-intensive, involving tasks ranging from technological trouble-shooting to strategic marketing. These factors all made this setting ideal for examining employees’ relational confidence when asking for knowledge, their behaviors on Chatter, and the outcome of interpersonal requests of knowledge.
Data were collected through a survey distributed to all 81 employees, all of whom were active Chatter users at the time of data collection. The survey was discussed and previewed at a company off-site meeting facilitated by the second author and attended by the company’s CEO. The CEO asked all employees at the offsite to complete the survey. He distributed the survey link via email the day after the off-site ended. The survey and recruitment strategy received IRB approval. The survey followed the standard format of name generator surveys in network perception research (Levin & Cross, 2004; Reagans & McEvily, 2003). Each employee was asked for up to five examples of a “time you asked someone for work-related information, knowledge, or advice.” For each of the five examples given, respondents were asked to select, from the employee roster, the source of knowledge. All knowledge sources were also one of the 81 employees surveyed. Respondents then typed a brief description of the knowledge sought in each instance and additionally answered the three standard questions to evaluate knowledge complexity from codified to tacit (e.g., Hansen 1999; Levin & Cross, 2004). Table 1 shows examples of knowledge sought as described by the respondents, separated by whether they were rated as codified or tacit.
Examples of knowledge sought
| Low complexity (codified) | High complexity (tacit) |
|---|---|
|
|
| Low complexity (codified) | High complexity (tacit) |
|---|---|
|
|
Examples of knowledge sought
| Low complexity (codified) | High complexity (tacit) |
|---|---|
|
|
| Low complexity (codified) | High complexity (tacit) |
|---|---|
|
|
For each instance of knowledge sharing, we asked respondents to indicate whether they waited before asking the source for needed knowledge or whether they asked for knowledge right away at the point they decided they needed it. This separated the 141 instances of knowledge seeking included in our present analysis from the 256 instances collected in full (about three instances per respondent). For the focal sample of 141, we asked about the respondent’s anticipatory communication with the knowledge source. Following other studies that have asked about communication media choices with respect to different alters (Kim et al., 2007), we asked respondents about the frequency with which they communicated through different media (phone, email, face-to-face, instant messaging, and the ESM Chatter) “specifically in hopes of improving the chances <the knowledge source> would take the time to provide the knowledge when asked.” Finally, for each instance of knowledge sharing, respondents were asked two items to elicit their satisfaction with the knowledge they ultimately received from the source.
A total of 78 (96.3%) responses were received. Respondents were 38.5% female and 62.5% male with an average age of 35.5 years (SD = 6.6 years). Organizational tenure ranged from 1 to 15 years, with an average of 5.7 years of experience (SD = 3.5 years). The 141 instances in which the respondent indicated engaging in anticipatory communication with a colleague before asking them for knowledge is the main sample we use to test our theory of communication visibility and relational confidence. In this sample, 72% of instances occurred across the two company divisions and 78% of instances occurred across different functional teams, indicating that indeed people often seek knowledge from others outside their primary work group.
Variables
Relational confidence
Relational confidence is the knowledge seeker’s perceived surplus or deficit of relational closeness with a desired source of knowledge. A surplus indicates relational confidence because the seeker perceives that they are close enough to the source to get the knowledge they need, while a deficit indicates a lack of confidence because the seeker perceives they are not close enough to the source to get the knowledge they need. We measure relational confidence as the difference in response to two survey questions about respondent perceptions. The first is: “To what extent do you agree with the statement: ‘I had a close working relationship with [source name] when I decided I would need knowledge about [knowledge description] from him/her.’” The response scale consists of seven items ranging from 1 (strongly disagree) to 7 (strongly agree). This type of survey question has been used widely to solicit people’s perceptions of relational closeness in the literature on organizational knowledge sharing (Hansen, 1999; Levin & Cross, 2004; Reagans & McEvily, 2003). Moreover, in their extensive analysis of network measures, Marsden and Campbell (1984) concluded that closeness is the best single item indicator of relationship strength. The second question is: “To what extent do you agree with the statement: ‘For someone to be willing to explain [knowledge description] to me, he/she would have to consider me a close colleague.’” The response scale also consists of seven items ranging from 1 (strongly disagree) to 7 (strongly agree). This question is an adaptation of questions like those used by Connelly and Kelloway (2003) and Zhang and Jiang (2015) to solicit how participants speculate about the provision of resources in network relationships. The Relational Confidence variable was created by subtracting the respondent’s speculation about needed closeness (second question) from their perception of actual closeness (first question) (M = −2.15, SD = 1.29, Min = −6, Max = 1). Negative values indicate a perceived relational deficit (lacking relational confidence) and positive values indicate a perceived relational surplus (relational confidence). It is important to note that this operationalization means that relational confidence s a dyad-specific perceptual measure and not a general theoretical construct like those for which multi-item scales are typically used. More details about the specific calculation of the relational confidence variable are available in the Appendix.
Perceived change in relational confidence
We measure whether respondents perceived change in relational confidence between the time they decided they needed knowledge from the source and when they actually asked the source for it. We asked respondents the following relational perception question: “To what extent do you agree with the statement: ‘I had a close working relationship with [source name] on the day I asked him/her for [knowledge description].’” The response scale consists of seven items ranging from 1 (strongly disagree) to 7 (strongly agree). This question differs from the first question, which elicited perceptions of closeness before the respondent asked for knowledge, but is again consistent with the standard relational closeness question used widely in the literature (Hansen, 1999; Levin & Cross, 2004; Marsden & Campbell, 1984; Reagans & McEvily, 2003). If the respondent reports a higher value to this question than the first, it indicates that they perceive an increase in relational closeness, such as a smaller relational deficit or a change from deficit to surplus. The average Perceived Change in Relational Confidence in our sample is 1.92 (SD = 1.10), indicating the average respondent did perceive a positive change. In line with our theorizing, this measure captures whether or not knowledge seekers perceived a change in closeness to the source, which would indicate a change in confidence even if there is no underlying reason for them to be more confident (i.e., even if the knowledge source does not in fact feel closer to the knowledge seeker).
Anticipatory communication
Following other studies about communication media choices with respect to different alters (Kim et al., 2007), we asked respondents about their anticipatory communication behavior specifically with their intended knowledge source through five different media: phone, email, face-to-face, instant messaging, and the ESM Chatter: “To what extent do you agree with the statement: ‘I used <communication media> to communicate with [source name] to increase the chance he/she would view me as a close colleague and be willing to take the time to explain [knowledge description] to me.’” The response scale consists of five items ranging from 1 (strongly disagree) to 5 (strongly agree). Answers to this question generate five different variables, one for each medium, for each instance of knowledge sharing: (a) ESM anticipatory communication (M = 4.2, SD = 1.16), (b) email anticipatory communication (M = 3.07, SD = 1.16), (c) instant messaging anticipatory communication (M = 1.70, SD = 1.08), (d) phone anticipatory communication (M = 2.16, SD = 1.17), and (e) face-to-face anticipatory communication (M = 2.18, SD = 1.39). Similarly, we further asked respondents about how frequently they used different features on the company’s ESM, Chatter, for anticipatory communication. The response scale consists of five items ranging from 1 (very infrequently) to 5 (very frequently). This generated five additional variables about ESM feature use specifically for anticipatory communication: (a) following a common file (M = 3.07, SD = 1.41), (b) following a common group (M = 3.14, SD = 1.28), (c) looking at the colleague’s profile (M = 3.55, SD = 1.11), (d) messaging with the colleague (M = 4.04, SD = 0.84), and (e) commenting on/responding to the colleague’s post (M = 4.00, SD = 1.01).
Satisfaction with knowledge sharing
Respondents were asked, for each knowledge seeking instance, the extent to which they agreed with two statements: (a) “The knowledge I received from [source name] was exactly what I was looking for” and (b) “I was able to use the knowledge I received from [source name] to improve the quality of my work.” Both questions have the same response scale of five items ranging from 1 (strongly disagree) to 5 (strongly agree). The variable Satisfaction with Knowledge Sharing was calculated as the average of these two responses (M = 3.89, SD = 1.01).
Knowledge complexity
We measure knowledge complexity from codified to tacit. We adopt the standard three questions (see Hansen, 1999; Levin & Cross, 2004) to measure knowledge complexity: (a) “What portion of the information or knowledge about <knowledge description> could be sufficiently explained to you in writing (in written reports, manuals, emails, faxes, etc.)? [response scale of seven items ranging from 1 (none of it) to 7 (all of it)], (b) “How well documented could the information or advice be about <knowledge description>? [response scale of seven items ranging from 1 (very well documented) to 7 (not well documented)], and (b) what type of information or advice about <knowledge description> actually came from <knowledge source>? [response scale of seven items ranging from 1 (All reports, manuals, documents) to 7 (All personal, practical know-how, tricks of the trade)]. Averaging these questions gives Knowledge Complexity (M = 4.34, SD = 1.52).
Demographic controls
We also control for demographic similarities between the seeker and source. Same Sex equals 1 if yes, 0 if no (M = 0.62, SD = 0.49). We also control for whether they are from the same team and same unit. Same Team equals 1 if yes, 0 if not (M = 0.13, SD = 0.34) and the Same Division equals 1 if yes, 0 if no (M = 0.28, SD = 0.45). Finally, we calculate Tenure Gap in years (M = −0.86, SD = 4.76).
Analysis
Table 2 displays the summary statistics and correlations for the sample of 141 knowledge seekers who anticipated asking for knowledge. It is important to note that there were 105 instances in our larger dataset where knowledge seekers did not anticipate, but instead asked for knowledge right away. However, it is interesting to compare the average relational confidence between the two groups: −2.1 for those who anticipated asking for knowledge had lower relational confidence compared with 0.23 for those who did not (t-test p < .001). In other words, knowledge seekers with relational confidence tended to ask for knowledge right away whereas knowledge seekers with deficits in relational confidence elected to communicate more with the knowledge source before asking for knowledge. This provides some initial evidence to support our theory that relational confidence informs people’s communication decisions and thus perhaps knowledge sharing outcomes.
Summary statistics and correlations (N = 141)
| Variable | M | SD | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Knowledge sharing satisfaction | 4.13 | 0.84 | 1.00 | ||||||||||
| 2 | Relational confidence | −2.15 | 1.29 | 0.25** | 1.00 | |||||||||
| 3 | ESM anticipatory communication | 4.19 | 1.16 | 0.38** | 0.07 | 1.00 | ||||||||
| 4 | IM anticipatory communication | 1.70 | 1.08 | −0.20* | 0.00 | −0.16+ | 1.00 | |||||||
| 5 | Email anticipatory communication | 3.07 | 1.16 | −0.17* | −0.22** | −0.13 | −0.08 | 1.00 | ||||||
| 6 | Phone anticipatory communication | 2.16 | 1.17 | −0.06 | −0.12 | −0.07 | 0.16+ | −0.15+ | 1.00 | |||||
| 7 | Face-to-face anticipatory communication | 2.18 | 1.39 | 0.02 | −0.15+ | −0.28** | −0.07 | 0.09 | −0.03 | 1.00 | ||||
| 8 | Knowledge complexity | 4.34 | 1.52 | −0.01 | −0.63** | −0.06 | −0.10 | 0.11 | 0.16+ | 0.26** | 1.00 | |||
| 9 | Same sex | 0.60 | 0.49 | 0.25** | 0.11 | 0.03 | 0.08 | −0.13 | −0.03 | −0.11 | −0.04 | 1.00 | ||
| 10 | Same team | 0.13 | 0.34 | 0.03 | −0.15+ | −0.08 | 0.03 | 0.01 | 0.09 | −0.01 | 0.16+ | 0.11 | 1.00 | |
| 11 | Same unit | 0.28 | 0.45 | −0.11 | −0.03 | −0.01 | −0.03 | −0.01 | −0.03 | −0.09 | 0.06 | −0.07 | 0.03 | 1.00 |
| 12 | Tenure gap | −0.86 | 4.76 | 0.05 | −0.13 | −0.17* | −0.05 | −0.02 | 0.04 | 0.01 | 0.12 | −0.08 | −0.09 | 0.05 |
| Variable | M | SD | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Knowledge sharing satisfaction | 4.13 | 0.84 | 1.00 | ||||||||||
| 2 | Relational confidence | −2.15 | 1.29 | 0.25** | 1.00 | |||||||||
| 3 | ESM anticipatory communication | 4.19 | 1.16 | 0.38** | 0.07 | 1.00 | ||||||||
| 4 | IM anticipatory communication | 1.70 | 1.08 | −0.20* | 0.00 | −0.16+ | 1.00 | |||||||
| 5 | Email anticipatory communication | 3.07 | 1.16 | −0.17* | −0.22** | −0.13 | −0.08 | 1.00 | ||||||
| 6 | Phone anticipatory communication | 2.16 | 1.17 | −0.06 | −0.12 | −0.07 | 0.16+ | −0.15+ | 1.00 | |||||
| 7 | Face-to-face anticipatory communication | 2.18 | 1.39 | 0.02 | −0.15+ | −0.28** | −0.07 | 0.09 | −0.03 | 1.00 | ||||
| 8 | Knowledge complexity | 4.34 | 1.52 | −0.01 | −0.63** | −0.06 | −0.10 | 0.11 | 0.16+ | 0.26** | 1.00 | |||
| 9 | Same sex | 0.60 | 0.49 | 0.25** | 0.11 | 0.03 | 0.08 | −0.13 | −0.03 | −0.11 | −0.04 | 1.00 | ||
| 10 | Same team | 0.13 | 0.34 | 0.03 | −0.15+ | −0.08 | 0.03 | 0.01 | 0.09 | −0.01 | 0.16+ | 0.11 | 1.00 | |
| 11 | Same unit | 0.28 | 0.45 | −0.11 | −0.03 | −0.01 | −0.03 | −0.01 | −0.03 | −0.09 | 0.06 | −0.07 | 0.03 | 1.00 |
| 12 | Tenure gap | −0.86 | 4.76 | 0.05 | −0.13 | −0.17* | −0.05 | −0.02 | 0.04 | 0.01 | 0.12 | −0.08 | −0.09 | 0.05 |
p < .1,
p < .05,
p <.01.
Summary statistics and correlations (N = 141)
| Variable | M | SD | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Knowledge sharing satisfaction | 4.13 | 0.84 | 1.00 | ||||||||||
| 2 | Relational confidence | −2.15 | 1.29 | 0.25** | 1.00 | |||||||||
| 3 | ESM anticipatory communication | 4.19 | 1.16 | 0.38** | 0.07 | 1.00 | ||||||||
| 4 | IM anticipatory communication | 1.70 | 1.08 | −0.20* | 0.00 | −0.16+ | 1.00 | |||||||
| 5 | Email anticipatory communication | 3.07 | 1.16 | −0.17* | −0.22** | −0.13 | −0.08 | 1.00 | ||||||
| 6 | Phone anticipatory communication | 2.16 | 1.17 | −0.06 | −0.12 | −0.07 | 0.16+ | −0.15+ | 1.00 | |||||
| 7 | Face-to-face anticipatory communication | 2.18 | 1.39 | 0.02 | −0.15+ | −0.28** | −0.07 | 0.09 | −0.03 | 1.00 | ||||
| 8 | Knowledge complexity | 4.34 | 1.52 | −0.01 | −0.63** | −0.06 | −0.10 | 0.11 | 0.16+ | 0.26** | 1.00 | |||
| 9 | Same sex | 0.60 | 0.49 | 0.25** | 0.11 | 0.03 | 0.08 | −0.13 | −0.03 | −0.11 | −0.04 | 1.00 | ||
| 10 | Same team | 0.13 | 0.34 | 0.03 | −0.15+ | −0.08 | 0.03 | 0.01 | 0.09 | −0.01 | 0.16+ | 0.11 | 1.00 | |
| 11 | Same unit | 0.28 | 0.45 | −0.11 | −0.03 | −0.01 | −0.03 | −0.01 | −0.03 | −0.09 | 0.06 | −0.07 | 0.03 | 1.00 |
| 12 | Tenure gap | −0.86 | 4.76 | 0.05 | −0.13 | −0.17* | −0.05 | −0.02 | 0.04 | 0.01 | 0.12 | −0.08 | −0.09 | 0.05 |
| Variable | M | SD | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Knowledge sharing satisfaction | 4.13 | 0.84 | 1.00 | ||||||||||
| 2 | Relational confidence | −2.15 | 1.29 | 0.25** | 1.00 | |||||||||
| 3 | ESM anticipatory communication | 4.19 | 1.16 | 0.38** | 0.07 | 1.00 | ||||||||
| 4 | IM anticipatory communication | 1.70 | 1.08 | −0.20* | 0.00 | −0.16+ | 1.00 | |||||||
| 5 | Email anticipatory communication | 3.07 | 1.16 | −0.17* | −0.22** | −0.13 | −0.08 | 1.00 | ||||||
| 6 | Phone anticipatory communication | 2.16 | 1.17 | −0.06 | −0.12 | −0.07 | 0.16+ | −0.15+ | 1.00 | |||||
| 7 | Face-to-face anticipatory communication | 2.18 | 1.39 | 0.02 | −0.15+ | −0.28** | −0.07 | 0.09 | −0.03 | 1.00 | ||||
| 8 | Knowledge complexity | 4.34 | 1.52 | −0.01 | −0.63** | −0.06 | −0.10 | 0.11 | 0.16+ | 0.26** | 1.00 | |||
| 9 | Same sex | 0.60 | 0.49 | 0.25** | 0.11 | 0.03 | 0.08 | −0.13 | −0.03 | −0.11 | −0.04 | 1.00 | ||
| 10 | Same team | 0.13 | 0.34 | 0.03 | −0.15+ | −0.08 | 0.03 | 0.01 | 0.09 | −0.01 | 0.16+ | 0.11 | 1.00 | |
| 11 | Same unit | 0.28 | 0.45 | −0.11 | −0.03 | −0.01 | −0.03 | −0.01 | −0.03 | −0.09 | 0.06 | −0.07 | 0.03 | 1.00 |
| 12 | Tenure gap | −0.86 | 4.76 | 0.05 | −0.13 | −0.17* | −0.05 | −0.02 | 0.04 | 0.01 | 0.12 | −0.08 | −0.09 | 0.05 |
p < .1,
p < .05,
p <.01.
We test our hypotheses with both ordinal and OLS regression models. Our measures for anticipatory communication on the ESM, and through other communication media, are ordered but non-linear, and so we use an ordinal regression model. Satisfaction with knowledge sharing is a two-question average and assumed to be linear, and thus for this outcome we employ OLS regression models. For both model types, we cluster standard errors at the respondent level because we have multiple instances of knowledge sharing per respondent. Without clustered standard errors, repeated observations are a potential source of nonindependence, which may artificially reduce standard errors and inflate significance values (Hoff, 2005).
Results
Figure 1 displays the results of a series of ordinal regressions used to determine how the seeker’s relational confidence before asking for knowledge is associated with their anticipatory communication through different media. The coefficients for both email and face-to-face communication with the source are negative and significant, meaning that the lower one’s relational confidence in the source, the more they interacted with the source through email or face-to-face before asking for knowledge in hopes of increasing the chances the source would be willing to share the knowledge when asked for it. We hypothesized (H1) that relational confidence would similarly have a negative and significant relationship with ESM communication, but this effect is not evident overall (p = .73, AIC = 344.33).
Effect on anticipatory communication.
Notes: Coefficient estimates based on ordinal regression, standard errors in parentheses are clustered by respondent. +p < .1, *p < .05, **p < .01, ***p < .001.
We further analyzed how people used five different features on the ESM for anticipatory communication. Figure 2 displays the results. Of the five actions, two were significant: following files (p = .02, AIC = 467.2) and following groups the knowledge source also follows (p = .01, AIC = 447.2). However, the relationship is positive, in the opposite direction hypothesized in H1. Interestingly, the results suggest that people with a greater deficit in relational confidence communicate somewhat less with the knowledge source through the ESM, and significantly more face-to-face and through email, before asking for knowledge.
Effect on ESM anticipatory communication.
Notes: Coefficient estimates based on ordinal regression, standard errors in parentheses are clustered by respondent. +p < .l, *p < .05, **p < .01, ***p < .001.
In H2, we proposed that the more frequently a knowledge seeker communicates with the source on ESM, the more the seeker perceives an increase in relational confidence. Figure 3 displays the coefficients for all five media from the regression testing this effect. There is considerable variation across modalities. ESM anticipatory communication, as well as face-to-face anticipatory communication, is associated with significant increases in the seeker’s perceived relational confidence, in support of H2. The coefficient for ESM communication is larger than the coefficient for face-to-face communication, but not significantly.
Change in relational confidence before asking for knowledge.
Notes: Coefficient estimates based on OLS regression, standard errors in parentheses are clustered by respondent, controlling for same sex, same division, same unit, and tenure gap. +p < .1, *p < .05, **p < .01, ***p < .001.
Our third and final hypothesis explores whether anticipatory communication through ESM that increases perceptions of relational confidence will also be associated with a meaningful improvement in knowledge sharing satisfaction. Table 3 displays the results of a series of nested OLS regression models to test H3. Model 1 includes only demographic control variables to predict satisfaction. Only Same Sex is significant. Model 2 includes all forms of anticipatory communication except SNS. Interestingly, email and instant messaging anticipatory communication are negatively and significantly associated with satisfaction. Model 3 includes ESM anticipatory communication, testing H3. In support of H3, the coefficient of ESM anticipatory communication is positive and significant. In Model 3, face-to-face anticipatory communication is also positive and significant; however, the effect size is significantly smaller than the effect for ESM communication (Wald test p = .002). Model 4 includes the knowledge seeker’s perceived relational confidence when they asked for knowledge. Relational confidence (asking) is positive and significant, consistent with our theory that relational confidence when asking independently influences knowledge sharing outcomes. Model 5 includes Knowledge complexity as an additional control, and the effect is not significant. Importantly, the effect of ESM anticipatory communication remains positive and significant in Models 4 and 5, demonstrating the robustness of the direct effect of ESM anticipatory communication on knowledge sharing satisfaction and in strong support of H3.
Effects on knowledge sharing satisfaction
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | |
|---|---|---|---|---|---|
| Demographic controls | |||||
| Same sex | 0.42** | 0.41* | 0.42** | 0.39** | 0.39** |
| (0.15) | (0.16) | (0.14) | (0.14) | (0.14) | |
| Same division | 0.02 | 0.06 | 0.14 | 0.11 | 0.09 |
| (0.18) | (0.17) | (0.16) | (0.17) | (0.18) | |
| Same company | −0.17 | −0.18 | −0.16 | −0.12 | −0.13 |
| (0.16) | (0.16) | (0.14) | (0.13) | (0.14) | |
| Tenure gap | 0.01 | 0.01 | 0.02* | 0.02* | 0.02* |
| (0.01) | (0.01) | (0.01) | (0.01) | (0.01) | |
| Anticipatory communication | |||||
| −0.11* | −0.08 | −0.06 | −0.06 | ||
| (0.06) | (0.05) | (0.05) | (0.05) | ||
| Face-to-face | 0.02 | 0.09* | 0.07* | 0.06+ | |
| (0.04) | (0.04) | (0.03) | (0.03) | ||
| Instant messaging | −0.17* | −0.12+ | −0.10+ | −0.11+ | |
| (0.08) | (0.06) | (0.06) | (0.06) | ||
| Phone | −0.04 | −0.02 | −0.01 | −0.02 | |
| (0.06) | (0.05) | (0.06) | (0.05) | ||
| ESM | 0.29*** | 0.24*** | 0.24*** | ||
| (0.07) | (0.07) | (0.07) | |||
| Seeker perceptions | |||||
| Relational confidence (asking) | 0.21* | 0.21* | |||
| (0.09) | (0.09) | ||||
| Knowledge complexity | 0.02 | ||||
| (0.03) | |||||
| Intercept | Y | Y | Y | Y | Y |
| Clustered SEs | Y | Y | Y | Y | Y |
| R2 | 0.07 | 0.15 | 0.29 | 0.36 | 0.36 |
| F(df) | 2.74*(136) | 2.93**(132) | 5.82***(131) | 7.76***(130) | 6.59***(129) |
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | |
|---|---|---|---|---|---|
| Demographic controls | |||||
| Same sex | 0.42** | 0.41* | 0.42** | 0.39** | 0.39** |
| (0.15) | (0.16) | (0.14) | (0.14) | (0.14) | |
| Same division | 0.02 | 0.06 | 0.14 | 0.11 | 0.09 |
| (0.18) | (0.17) | (0.16) | (0.17) | (0.18) | |
| Same company | −0.17 | −0.18 | −0.16 | −0.12 | −0.13 |
| (0.16) | (0.16) | (0.14) | (0.13) | (0.14) | |
| Tenure gap | 0.01 | 0.01 | 0.02* | 0.02* | 0.02* |
| (0.01) | (0.01) | (0.01) | (0.01) | (0.01) | |
| Anticipatory communication | |||||
| −0.11* | −0.08 | −0.06 | −0.06 | ||
| (0.06) | (0.05) | (0.05) | (0.05) | ||
| Face-to-face | 0.02 | 0.09* | 0.07* | 0.06+ | |
| (0.04) | (0.04) | (0.03) | (0.03) | ||
| Instant messaging | −0.17* | −0.12+ | −0.10+ | −0.11+ | |
| (0.08) | (0.06) | (0.06) | (0.06) | ||
| Phone | −0.04 | −0.02 | −0.01 | −0.02 | |
| (0.06) | (0.05) | (0.06) | (0.05) | ||
| ESM | 0.29*** | 0.24*** | 0.24*** | ||
| (0.07) | (0.07) | (0.07) | |||
| Seeker perceptions | |||||
| Relational confidence (asking) | 0.21* | 0.21* | |||
| (0.09) | (0.09) | ||||
| Knowledge complexity | 0.02 | ||||
| (0.03) | |||||
| Intercept | Y | Y | Y | Y | Y |
| Clustered SEs | Y | Y | Y | Y | Y |
| R2 | 0.07 | 0.15 | 0.29 | 0.36 | 0.36 |
| F(df) | 2.74*(136) | 2.93**(132) | 5.82***(131) | 7.76***(130) | 6.59***(129) |
Note. Coefficient estimates based on OLS regression, standard errors in parentheses are clustered by respondent.
p < .1,
p < .05,
p < .01,
p < .001.
Effects on knowledge sharing satisfaction
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | |
|---|---|---|---|---|---|
| Demographic controls | |||||
| Same sex | 0.42** | 0.41* | 0.42** | 0.39** | 0.39** |
| (0.15) | (0.16) | (0.14) | (0.14) | (0.14) | |
| Same division | 0.02 | 0.06 | 0.14 | 0.11 | 0.09 |
| (0.18) | (0.17) | (0.16) | (0.17) | (0.18) | |
| Same company | −0.17 | −0.18 | −0.16 | −0.12 | −0.13 |
| (0.16) | (0.16) | (0.14) | (0.13) | (0.14) | |
| Tenure gap | 0.01 | 0.01 | 0.02* | 0.02* | 0.02* |
| (0.01) | (0.01) | (0.01) | (0.01) | (0.01) | |
| Anticipatory communication | |||||
| −0.11* | −0.08 | −0.06 | −0.06 | ||
| (0.06) | (0.05) | (0.05) | (0.05) | ||
| Face-to-face | 0.02 | 0.09* | 0.07* | 0.06+ | |
| (0.04) | (0.04) | (0.03) | (0.03) | ||
| Instant messaging | −0.17* | −0.12+ | −0.10+ | −0.11+ | |
| (0.08) | (0.06) | (0.06) | (0.06) | ||
| Phone | −0.04 | −0.02 | −0.01 | −0.02 | |
| (0.06) | (0.05) | (0.06) | (0.05) | ||
| ESM | 0.29*** | 0.24*** | 0.24*** | ||
| (0.07) | (0.07) | (0.07) | |||
| Seeker perceptions | |||||
| Relational confidence (asking) | 0.21* | 0.21* | |||
| (0.09) | (0.09) | ||||
| Knowledge complexity | 0.02 | ||||
| (0.03) | |||||
| Intercept | Y | Y | Y | Y | Y |
| Clustered SEs | Y | Y | Y | Y | Y |
| R2 | 0.07 | 0.15 | 0.29 | 0.36 | 0.36 |
| F(df) | 2.74*(136) | 2.93**(132) | 5.82***(131) | 7.76***(130) | 6.59***(129) |
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | |
|---|---|---|---|---|---|
| Demographic controls | |||||
| Same sex | 0.42** | 0.41* | 0.42** | 0.39** | 0.39** |
| (0.15) | (0.16) | (0.14) | (0.14) | (0.14) | |
| Same division | 0.02 | 0.06 | 0.14 | 0.11 | 0.09 |
| (0.18) | (0.17) | (0.16) | (0.17) | (0.18) | |
| Same company | −0.17 | −0.18 | −0.16 | −0.12 | −0.13 |
| (0.16) | (0.16) | (0.14) | (0.13) | (0.14) | |
| Tenure gap | 0.01 | 0.01 | 0.02* | 0.02* | 0.02* |
| (0.01) | (0.01) | (0.01) | (0.01) | (0.01) | |
| Anticipatory communication | |||||
| −0.11* | −0.08 | −0.06 | −0.06 | ||
| (0.06) | (0.05) | (0.05) | (0.05) | ||
| Face-to-face | 0.02 | 0.09* | 0.07* | 0.06+ | |
| (0.04) | (0.04) | (0.03) | (0.03) | ||
| Instant messaging | −0.17* | −0.12+ | −0.10+ | −0.11+ | |
| (0.08) | (0.06) | (0.06) | (0.06) | ||
| Phone | −0.04 | −0.02 | −0.01 | −0.02 | |
| (0.06) | (0.05) | (0.06) | (0.05) | ||
| ESM | 0.29*** | 0.24*** | 0.24*** | ||
| (0.07) | (0.07) | (0.07) | |||
| Seeker perceptions | |||||
| Relational confidence (asking) | 0.21* | 0.21* | |||
| (0.09) | (0.09) | ||||
| Knowledge complexity | 0.02 | ||||
| (0.03) | |||||
| Intercept | Y | Y | Y | Y | Y |
| Clustered SEs | Y | Y | Y | Y | Y |
| R2 | 0.07 | 0.15 | 0.29 | 0.36 | 0.36 |
| F(df) | 2.74*(136) | 2.93**(132) | 5.82***(131) | 7.76***(130) | 6.59***(129) |
Note. Coefficient estimates based on OLS regression, standard errors in parentheses are clustered by respondent.
p < .1,
p < .05,
p < .01,
p < .001.
Taking the results together, we find different effects for different communication media when it comes to relational confidence. Knowledge seekers seem to see ESM anticipatory communication—particularly following a common group or file—as particularly useful at the margin, when they perceive only a small deficit in relational confidence. Those with the lowest relational confidence, however, most often choose face-to-face and email anticipatory communication to help increase their confidence before asking for knowledge. Yet, ESM anticipatory communication is associated with a similar increase in relational confidence as face-to-face anticipatory communication, and ESM anticipatory communication has a more significant and larger effect on the seeker’s satisfaction with knowledge sharing compared with all other forms of anticipatory communication. These results demonstrate the unique role of ESM anticipatory communication and its effect on relational confidence for knowledge sharing outcomes, consistent with our theorizing about the role of communication visibility in building relational confidence, and further suggests that the effectiveness of ESM anticipatory communication may be underestimated by knowledge seekers.
Discussion
It is becoming clear that the future of work will be increasingly remote and hybrid. Employees will, no doubt, continue to gather together in the office at strategic intervals, but they will often communicate with their colleagues from their homes, cowork locations, client sites, or other corporate offices through digital communication technologies, like ESM. Although these technologies provide important affordances for remote work, nearly every study that has been done about the shift in employee communication networks as a result of using ESM suggests that when compared with the formation and maintenance of social networks in office environments where employees work predominantly face-to-face, employees in remote and hybrid work arrangements will maintain stronger ties with their immediate workgroups at the expense of the maintenance of strong ties with employees in other parts of the organization (e.g., Yang et al., 2021; Zuzul et al., 2021). This shift portends problems for knowledge sharing because when employees need knowledge from people outside their teams, they may not have close enough relationships to feel confident to ask for it.
To understand and potentially address these problems, we contribute the concepts of relational confidence and anticipatory communication. These are dyad-level concepts distinct, but related to, network-level concepts widely studied about relationships and communication. Compared with theories of communication that demonstrate the connection between visibility and network-wide understanding of people and relationships (e.g., Leonardi, 2014, 2015; Treem et al., 2020b; van Zoonen et al., 2022), our idea of anticipatory communication zooms in to how people communicate with a particular other with a specific goal in mind. Compared with theories of social networks that conjecture the role of actual relational closeness (e.g., Hansen, 1999; Levin & Cross, 2004; Reagans & McEvily, 2003), our idea of relational confidence suggests the microactions can elevate feelings of closeness regardless of whether relationship strength truly changes. This perspective on relational perceptions and communication choices enriches network-level theorizing with an understanding of how digital communication technologies matter for particular interpersonal communication, perceptions, and knowledge sharing.
With a dyad-level focus, our findings provide the first evidence showing that knowledge seekers in a hybrid organization can use ESM in ways that allow them to feel comfortable asking for the knowledge they need from a source of knowledge they do not know well. Importantly, the findings also show how relational confidence can be built and they demonstrate a strong empirical link between asking for knowledge with relational confidence and high satisfaction with the knowledge sharing process. In short, when knowledge seekers build relational confidence with a knowledge source, they are more likely not only to ask for knowledge, but they are likely to be satisfied with the knowledge they receive. These findings have implications for research on the future of work, generally, and for theory about communication visibility specifically.
Relational confidence and the future of work
In all work settings, but particularly so in hybrid and remote work arrangements, it is common to need knowledge from distant colleagues. Unfortunately, such knowledge is often hard to get. Knowledge is sticky and difficult to move from one area of an organization to another (Szulanski, 1996), particularly when people do not know each other well (Borgatti & Cross, 2003) and when the knowledge is very complex and difficult to explain to another (Hansen, 1999). Most prior research on knowledge sharing has focused on the mechanisms for effective knowledge search and transfer, but have overlooked an important process in between: How people decide whether or not to ask for the knowledge they need. Our findings demonstrate that one of the key challenges to moving knowledge among employees is due to a relational confidence problem. Even when organizations—such as TeleMobile—have developed a strong culture of knowledge sharing and implemented specific tools and technologies to support it, knowledge can be harder to get for people who perceive that they do not have a strong enough relationship with the person who has the knowledge they need. Our study suggests that organizational knowledge sharing outcomes can be improved when people come to feel more relationally confident before asking a colleague for knowledge.
Specifically, our study provides several important implications related to knowledge sharing in remote and hybrid work. First, anticipatory communication with a colleague (the knowledge source) is important for building a knowledge seeker’s relational confidence. Our results show that without the ability to communicate or interact with a colleague in advance of asking for knowledge (which is common when, in remote work settings, someone learns about knowledge outside of their work group; van Zoonen et al., 2022), people may not have the relational confidence to ask for the knowledge at all, and thus could instead choose to ask someone else who may be less knowledgeable or forgo it entirely. We also found that nearly all people in our remote and hybrid work sample communicate through ESM in anticipation of asking a colleague for knowledge, while those with the lowest levels of relational confidence additionally communicate with the colleague face-to-face and over email. This finding seems to indicate that when people have particularly low relational confidence, they choose modes of anticipatory communication that utilize rich media and which can perhaps better enable strong, close relational development. That is, people who have low relational confidence tend to adopt a strategy suggested by prior work: try to build stronger ties to better facilitate knowledge sharing (Hansen, 1999). But in contrast to their preferences to use “rich” media to build these relationships, the results show that the face-to-face communication preferred by people with lower relational confidence is more difficult to achieve in hybrid and remote work environments. The fact that ESM were the most useful tools for building relational confidence in our study indicates, in line with prior work on media richness (Daft & Lengel, 1986), that the success lies in choice of communicative medium affording appropriate cues. We present new evidence for how the indirect or transactional communicative acts that occur on ESM are not rich, but are very appropriate to the complex task of developing relational confidence for knowledge sharing.
Another implication suggested by these findings is that greater relational confidence improves a seeker’s satisfaction with the knowledge sharing process. Our results suggest that this occurs because people with greater relational confidence perceive that they have stronger ties with their colleagues and a better understanding of the knowledge sought and how to ask for it. Assuming that such perceptions are not always accurate, relational confidence appears to improve knowledge sharing outcomes even if the perceptions do not reflect reality. This finding adds to the emerging literature on the importance of network perceptions in driving effective organizational outcomes (Smith et al., 2020). Building on prior findings showing that perceptions of one’s own network activate certain relational moves, our results suggest that using ESM can aid knowledge seekers to shift their perceptions of closeness in ways that encourage them to ask for the knowledge they need. Relatedly, ESM are unique in their ability to help people build relational confidence in a way that improves the outcome of knowledge sharing. We theorize that this is because of the visibility of communication on ESM, which helps knowledge seekers publicly signal to the knowledge source that they have common interests and to learn about the knowledge the source has through observing their behavior, interactions, and followed files and work groups.
Our study also advances the theory of communication visibility (Kim et al., 2019; Leonardi, 2014, 2015; Rice et al., 2017; Treem et al., 2020a). First, our findings illuminate that the communication visibility afforded on social networking platforms can help for knowledge sharing beyond the development of metaknowledge. The assumption in much prior research has been that people who use ESM to improve their metaknowledge move straight from increased awareness of who has what knowledge to asking for knowledge (Liang et al., 2022; van Osch & Steinfield, 2018). We instead show that people engage in anticipatory communication to increase the chances that they will successfully receive the knowledge they need when they ask for it, particularly when they do not feel confident asking for it right away. Once we recognize that anticipatory communication happens before knowledge sharing, it is easier to understand how communication visibility plays a critical role.
The public and traceable nature of communication over ESM (Ellison et al., 2015) is an important reason why this mode of communication is uniquely effective at increasing relational confidence in a way that actually improves the outcome of knowledge sharing. Yet, it is interesting that we observe that people with lower levels of relational confidence are also communicating face-to-face and over email in anticipation of asking for knowledge, but these modes of anticipatory communication have less or no effect on knowledge sharing outcomes. The evidence we find indicates that ESM communication increases both feelings of relational confidence and knowledge sharing satisfaction, whereas face-to-face communication increases feelings of relational confidence but not knowledge sharing satisfaction. Perhaps the reason why is that anticipatory face-to-face communication improves people’s confidence in their personal connection to the knowledge source, whereas anticipatory ESM communication improves people’s confidence in their awareness of the knowledge itself, thus better facilitating knowledge sharing outcomes. This interpretation is corroborated by the evidence we find that the most instrumental behaviors on the ESM are following a common group or file. More research on relational confidence and anticipatory communication is needed to test further our preliminary evidence.
Limitations and opportunities for future research
This study is only a first step toward understanding relational confidence and the way that communication visibility afforded by ESM can help people with low relational confidence acquire the knowledge they need when they are separated from others via time and space. One limitation to our study is that we use only a single item to measure perceptions of relational closeness. Although this is sound practice (Marsden & Campbell, 1984), we encourage future research to enhance our measures and further test their validity. A second limitation of our study is its cross-sectional survey design. Without the ability to survey employees at multiple points in time, we designed a survey to elicit past instances of knowledge sharing, past perceptions about relational confidence in a colleague, and past communication with a colleague who had needed knowledge. This retrospective cross-sectional design could create bias in variable measurements. The most significant source of potential bias is that people’s perceptions of past relational confidence in a colleague are not accurate because they are influenced by their later communication and knowledge sharing outcome. If systemic bias indeed exists, we would expect that our measures of relational confidence in a colleague, anticipatory communication with the colleague, and the outcome of knowledge sharing with that same colleague would together be high or low (i.e., more confidence, more anticipatory communication, and better outcome). This is not what we find; we observe the relationships to relational confidence are in opposite directions for different media, and moreover, that among the different communication media, only ESM use is significantly and consistently associated with a better outcome of knowledge sharing. Therefore, although we cannot eliminate the possibility of biases, the data do not evidence systematic biases in our measures. We encourage future longitudinal studies to assess changes in relational confidence and anticipatory communication in real time.
Another limitation of our study is that we do not have two-sided measures to assess the tie strength between colleagues. Therefore, we cannot say how people’s perceptions of relational confidence align (or not) with actual measures of tie strength. We suspect, however, that perceptions of relational confidence in a colleague may diverge quite significantly from tie strength with a colleague. After all, relational confidence is different even from perceived tie strength; relational confidence is specific to the knowledge sought and whether people feel a colleague would be willing to share that specific knowledge when asked. Therefore, our study speaks to prior work, such as Borgatti and Cross (2003), which shows strong ties are critical for knowledge sharing (particularly complex knowledge sharing). Future research should design studies to compare the effect of tie strength versus relational confidence on knowledge sharing outcomes. It might be that relational confidence matters more, and if so, this would have transformative implications for how to manage and improve knowledge sharing in organizations.
A final important limitation of our study is that we analyze only instances where people ultimately asked a colleague for knowledge. In our survey, we find that those with high relational confidence in a colleague asked the colleague for knowledge right away, without any anticipatory communication. Future research should also seek to uncover how relational confidence in different colleagues motivates who people choose to ask for knowledge and who they do not, and perhaps also whether they decide to ask for knowledge at all. Further, it would be illuminating to better understand how different communication modalities, and specifically visible communication on ESM, differentially help people increase their relational confidence enough to actually ask for knowledge. Such studies could further advance theory at the intersection of communication visibility and knowledge sharing.
Conclusion
Our study points to new insights for those managing or researching hybrid and remote work. Until now, efforts to improve knowledge sharing in organizations with hybrid or remote workers have been focused on ESM and networking platforms to help improve employees to develop more accurate metaknowledge, and perhaps also to codify knowledge in digital files and repositories. Until now, the relational confidence people have in their colleagues and their anticipatory communication before asking a colleague for knowledge were not widely recognized as important antecedents of effective knowledge sharing. Our study demonstrates that these same technologies—because of the communication visibility they afford users—are uniquely positioned to afford people the opportunity to build their relational confidence in a way that improves knowledge sharing outcomes. We suspect that changes in relational confidence may be a microfoundation of wider social and communication networks. If relational confidence changes knowledge sharing outcomes, it might also change relationships themselves and how networks evolve, and thus have large implications for work in remote and hybrid organizations.
Data availability
The data presented in this article cannot be shared publicly due to concern for the privacy of individuals that participated in the study and an agreement made with the organization that gave access to this research. The data will be shared on reasonable request to the corresponding author.
Funding
Funds for this research were made possible by grants from the National Science Foundation (SES-2051896 and SES-1922266).
Conflicts of interest: None declared.
Acknowledgments
The authors wish to thank Nicole Ellison, Nancy Baym, and the anonymous reviewers for providing comments that improved this manuscript.
