Multiple team membership and job performance: The role of employees’ information‐sharing networks

Individuals in contemporary work organizations are often involved in multiple teams at the same time. This study uses a social capital perspective to propose that employees' multiple team memberships (MTM) offer the potential for individual performance benefitsanddetriments, depending on the characteristics of an employee's information-sharing network. To test our predictions, we gathered both archival and survey data at an organization for applied research in the Netherlands. We found that individual MTM was indirectly associated with an employee's overall job performance by increasing the size of his or her information-sharing network. As expected, however, this indirect relationship was contingent on the average strength of an employee's network ties (i.e., the frequency of the respective interactions), such that MTM only improved overall performance when network ties were relatively weak. The indirect relationship between MTM and individual job performance was negative, by contrast, when an employee's network ties were relatively strong. Together, these findings advance our understanding of the mechanisms and contingency factors that shape the performance consequences associated with individuals' concurrent membership in multiple teams. Practitioner points An employee's membership in multiple teams at the same time increases the size of his or her information-sharing network within the organization. The performance consequences associated with this increased information-sharing network hinge on the characteristics of an employee's information-sharing network. If the respective information-sharing linkages are based on relatively infrequent interactions with colleagues, an employee's multiple team membership indirectly benefits his or her overall job performance. If the respective information-sharing linkages are based on relatively frequent and intense interactions with colleagues, however, an employee's multiple team membership indirectly diminishes his or her overall job performance.

Modern organizations increasingly use team-based structures to bring together employees with different knowledge bases and diverse types of expertise to solve complex problems (e.g., project teams; Edmondson & Harvey, 2017;Wuchty, Jones, & Uzzi, 2007). Within these structures, it is rather common that individual employees participate in more than one team at the same time, balancing the requirements of a particular project with responsibilities in other teams (Cummings & Haas, 2012;Mortensen, 2014). Studies estimate that such multiple team membership (MTM) occurs among at least 65% of employees across Europe and the United States (Mortensen, Woolley, & O'Leary, 2007;O'Leary et al., 2011).
Recognizing this salience of MTM in many organizations, several studies have considered its consequences for key work outcomes (Cummings & Haas, 2012;Rapp & Mathieu, 2018). Conceptual research, for example, has suggested that organizations introduce MTM-based structures to increase individual employees' efficiency and effectiveness on the job (O'Leary et al., 2011). More recent longitudinal research has found that higher MTM was associated with subsequent, long-term improvements in an employee's overall job performance (i.e., his or her general task accomplishment across teams; Van de Brake, Walter, Rink, Essens, & Van der Vegt, 2018). 1 Scholars have prominently invoked social network mechanisms to provide a theoretical explanation for such performance benefits (Mortensen, Woolley, & O'Leary, 2007;Van de Brake et al., 2018). Specifically, MTM may extend an employee's social network across multiple teams, allowing him or her to effectively transfer key resources (e.g., work-related information and knowledge) between teams and, thus, to achieve superior individual performance outcomes across tasks and teams (O'Leary et al., 2011;Wageman, Gardner, & Mortensen, 2012). Importantly, however, the empirical MTM research has not examined such network mechanisms to date. Further complicating matters, there are good theoretical reasons to expect that MTM's social network implications and, thus, its overall performance consequences are not universally positive. It is possible, in particular, that the expanded social network resulting from high MTM creates a complex and demanding work context that overburdens an employee's capacities (Mortensen et al., 2007). Some scholars have argued, accordingly, that MTM may decrease (rather than improve) an employee's overall job performance (Kauppila, 2014;Pluut, Flestea, & Curs ßeu, 2014;Zika-Viktorsson, Sundstr€ om, & Engwall, 2006).
Hence, despite burgeoning scholarly interest in MTM (Margolis, 2020), our conceptual understanding of this construct's consequences for an individual employee's overall job performance remains incomplete and ambiguous. It is unclear, in particular, (1) whether prevalent social network explanations indeed provide a viable account of individual MTM's performance implications and (2) whetheror when -MTM's beneficial versus detrimental consequences are most likely to prevail. These open questions limit our knowledge of how this common type of contemporary work arrangement shapes key performance outcomes. Consequently, we believe it is important to empirically examine the social network mechanisms that may explain MTM's consequences for individual employees' overall job performance and, in doing so, to consider key boundary conditions that may either accentuate MTM's advantages or emphasize its downsides. The present research combines insights from social capital theory (Lin, 1999(Lin, , 2002 and the social network literature (Borgatti & Foster, 2003;Kwon & Adler, 2014) to address these issues.
First, individual MTM requires employees to cooperate with other individuals across multiple distinct teams (O'Leary et al., 2011). Within each of these teams, members need to utilize and share resources (e.g., task-related knowledge and expertise) to complete tasks and achieve common goals (Rost, 2011;Van der Vegt, Van De Vliert, & Oosterhof, 2003). Hence, we draw on social capital theory, one of the most prominent perspectives on interpersonal resource transactions (Lin, 1999(Lin, , 2002, to suggest that MTM may expand an employee's information-sharing network. More specifically, higher MTM may expose an employee to a greater number of team members that can provide access to a wide variety of information resources (e.g., useful ideas and work practices; Krackhardt, 1992). Subsequently, a focal employee may utilize these resources to achieve superior job performance (Vedres & Stark, 2010). Our conceptual model therefore casts an employee's number of information-sharing network ties (i.e., the size of his or her respective network) as a key mediating mechanism that explicates MTM's performance implications.
Second, we amend this social capital perspective with insights from social networks research to posit that information-sharing relationships (also known as network ties) between individuals differ in the extent to which they can provide relevant and useful information (Borgatti & Foster, 2003). A critical factor, in this regard, is the strength of a network tie (i.e., the frequency of the respective interactions; Krackhardt, 1992). Scholars have frequently used Granovetter's (1973) theory on the strength of weak ties to analyse the role of interpersonal connections for individuals' resource attainment and subsequent performance outcomes (Hansen, 1999;Perry-Smith & Shalley, 2014). On this basis, we propose that the strength of an employee's network ties functions as a key boundary condition in the relationship between his or her information-sharing network size and overall job performance.
Drawing this argumentation together, we argue that MTM offers the potential for indirect performance benefits and detriments by increasing the size of an employee's information-sharing network (see Figure 1). Moreover, we propose that the direction of the indirect MTM-performance relationship critically hinges on the strength of the respective network connections. When MTM results in a greater number of informationsharing ties and, on average, these ties are relatively weak, this should offer the potential for overall performance benefits by increasing an employee's access to novel and useful information resources (Granovetter, 1973;Reagans & Zuckerman, 2001). When the information-sharing network ties emanating from higher MTM are relatively strong, by contrast, this may substantially drain an employee's time and energy resources without providing the informational benefits associated with weaker ties, therefore diminishing his or her job performance (Hobfoll, 1989;Mehra, Kilduff, & Brass, 2001). By examining these notions, we aim to show the relevance of a social network perspective in MTM research and to integrate seemingly contradictory arguments that previous research has put forth about individual MTM's performance benefits and detriments. In doing so, this study offers new theoretical insights into the complex performance consequences that ensue when an employee's work is distributed across multiple concurrent teams, explicating both why and when an employee's performance is likely to improve or to suffer in such modern work arrangements.

Theory and hypotheses
Scholars have depicted individual MTM as a complex construct that comprises different dimensions, including the number of concurrent teams that an individual belongs to as well as these teams' variety (e.g., in terms of tasks, functions, or technologies; O'Leary et al., 2011). Following prior research (Pluut et al., 2014;Van de Brake et al., 2019), the present study focuses on the former aspect. Hence, we define individual MTM as the number of teams to which an employee allocates working time during a specific period (e.g., per week; Van de Brake et al., 2018). When such MTM is higher, an employee is simultaneously involved in a greater number of teams, whereas an individual with lower MTM spends his or her working time in only one or a few concurrent teams. We focus on this dimension because previous studies (1) have shown such MTM to potentially shape employees' overall job performance across teams and tasks (Van de Brake et al., 2018;Zika-Viktorsson et al., 2006) and (2) have suggested that an increase in MTM number (rather than variety) may substantially increase an employee's social network on the job (O'Leary et al., 2011;Van de Brake et al., 2018, 2019. We note, however, that we controlled for MTM variety in our hypotheses tests and further explored this alternative dimension in supplementary analyses.

MTM and the size of an individual's information-sharing network
Recent MTM research proposed that this work practice may come with key 'social capital advantages' that may drive the relationship between MTM and overall job performance ( Van de Brake et al., 2018, p. 1228see also O'Leary et al., 2011). Social capital theory (Lin, 1999(Lin, , 2002 suggests that an individual's information-sharing network entails valuable resources, such as knowledge, instrumental support, and expertise (Thompson, 2005). By consequence, an individual's information-sharing network is described as an important source of social capital and, thus, as a key antecedent of his or her job performance (Cross & Cummings, 2004;Kwon & Adler, 2014). Following prior MTM research (O'Leary et al., 2011;Van de Brake et al., 2018), we therefore propose that individual MTM enlarges an employee's information-sharing network and, thus, may enable him or her to achieve superior performance levels.
Specifically, we expect that higher MTM increases the number of coworkers with whom an employee can share information about work-related topics (Mehra et al., 2001;Perry-Smith, 2006). MTM implies, after all, that an individual employee simultaneously collaborates with various coworkers across multiple teams (Mortensen et al., 2007). Although memberships may overlap between simultaneous teams to some extent (with teams sharing two or more members; Vedres & Stark, 2010), it is unlikely that each team within an organization can accomplish its unique tasks and meet its specific demands with the same set of individuals (O'Leary et al., 2011). Hence, each additional team membership increases the likelihood for an employee of encountering new informationsharing opportunities with previously unfamiliar colleagues.
It is important to note, in this regard, that the additional colleagues encountered with increasing MTM are distributed across a number of different teams. Within each of these teams, members are likely to work interdependently, such that they need to share information about their task progress, deadlines, procedures, and available resources to achieve joint goals (Mesmer-Magnus & DeChurch, 2009;Van der Vegt, Van De Vliert, & Oosterhof, 2003). Consequently, higher MTM may give an employee a more extensive overview of the job-relevant information sources available within his or her task environment (Lewis, 2004;Lin, 1999), and he or she may be able to access and use the unique information embedded within different teams with relative ease (Choi & Thompson, 2005;Perry-Smith, 2006). Together, this suggests that higher MTM increases an employee's access to potentially useful information resources and, as such, greatly expands his or her information-sharing network.
When an employee's MTM is less pronounced, by contrast, his or her informationsharing network is likely to remain confined to only one or a few teams. In such circumstances, the respective employee will collaborate with a relatively limited number of coworkers across a small number of teams, restricting his or her access to other teams' informational resources. Hence, we expect that employees' information-sharing networks are smaller when their MTM is lower. Altogether, we therefore suggest: Hypothesis 1. An employee's multiple team membership is positively related to the size of his or her information-sharing network.

Information-sharing network size and tie strength
We argue that fully understanding MTM's performance consequences requires moving beyond the size of an individual's information-sharing network to also consider the strength of the respective network ties. Indeed, network scholars have emphasized that besides the number of an individual's information-sharing ties (i.e., the size of one's information-sharing network), the strength of these ties critically determines a network's utility for a focal individual (Granovetter, 1973). Weaker information-sharing ties are characterized by less frequent interactions (e.g., with acquaintances or distant colleagues), whereas stronger network ties reflect closer connections between individuals with more frequent interactions (e.g., with friends or close colleagues; Krackhardt, 1992). Individuals usually develop strong ties with team members holding values and attitudes that are similar to their own, whereas information-sharing ties with members holding dissimilar attitudes and values may remain weaker (Borgatti & Foster, 2003;Klein et al., 2004). Moreover, individuals may differ in their approaches towards within-team cooperation and coordination, with some individuals preferring more frequent interactions than others to align joint task accomplishment and exchange relevant information (Mesmer-Magnus & DeChurch, 2009). Hence, even though higher MTM is likely to increase the size of an employee's information-sharing network, we believe an individual's MTM should be largely unrelated with the average strength of the respective connections, such that a multi-teamer's information-sharing network may entail both relatively strong and relatively weak ties. Although strong ties have a variety of intuitive benefits, including the opportunity to garner emotional support and develop a deep understanding of others' attitudes, beliefs, and assumptions (Hansen, 1999;Krackhardt, 1992), we propose that weaker information-sharing ties will be particularly useful for employees because these ties typically connect with relatively dissimilar others that have access to different sources of information than the focal individual (Granovetter, 1973). The novel and non-redundant information accessible through such weak network ties may help employees to solve non-routine problems and may provide them with alternative task approaches and opportunities for knowledge integration across multiple teams (Hansen, 1999;Reagans & Zuckerman, 2001). Moreover, an employee should be able to maintain weak information-sharing ties with minimal time expenditure and high efficiency, because such ties are based on relatively infrequent interactions (Levin, Walter, & Murnighan, 2011). Accordingly, research has demonstrated that weaker network ties are more likely to spark new ideas, to promote effective and efficient task accomplishment and, by consequence, to foster individuals' performance (Perry-Smith & Shalley, 2014). Building on this conceptual fundament, we anticipate the strength of the ties within an individual's informationsharing network to moderate the linkage between the size of the respective network and an employee's performance on the job.
More specifically, we propose that an employee's overall job performance is most likely to benefit if his or her information-sharing network is relatively large and when the respective network ties are, on average, relatively weak. In MTM contexts, in particular, this type of network configuration should provide efficient access to many non-redundant informational resources across multiple teams (Krackhardt, 1992). As such, a large information-sharing network that, on average, comprises relatively weak ties may allow an employee to integrate many diverse insights from separate teams to create innovative work approaches and develop new and workable solutions to complex problems (Amabile, Conti, Coon, Lazenby, & Herron, 1996;Edmondson & Harvey, 2017;Grant, 1996).
By contrast, stronger (rather than weaker) network ties require substantially more time and attention due to their higher interaction frequency and are less useful for providing the unique, non-redundant types of information required for superior job performance (Krackhardt, 1992). A key reason is that individuals tend to form stronger connections with people that are similar to themselves (e.g., with similar expertise and perspectives) and that individuals who interact more frequently tend to become even more similar in terms of resources and experiences over time (Marsden & Friedkin, 1993). Accordingly, social network research has shown that stronger ties are less likely to provide access to novel information that is otherwise not available in one's social surroundings (Krackhardt, 1992;Perry-Smith & Shalley, 2014;Reagans & Zuckerman, 2001).
Drawing from this logic, we argue that a multi-teamer's large information-sharing network that, on average, contains relatively strong ties is likely to hold redundant information and, consequently, has limited potential to improve the respective employee's performance across multiple teams (Granovetter, 1973). In fact, maintaining a large network of strong, high-frequency information-sharing ties may distract a focal employee from core task responsibilities, diminishing the time and effort the employee can invest to perform efficiently and effectively on the job (Mehra et al., 2001;Rook, 1984). Interestingly, the notion that strong relationships may deplete scarce personal resources (e.g., time, attention) is also echoed in conservation of resources theory (Hobfoll, 1989). Indeed, the resource expenditure required to nurture and maintain a large number of strong network ties may deplete a multi-teamer's energy and potentially evoke fatigue and decreased task efficiency (Mehra et al., 2001;Van de Brake et al., 2018). Logically, then, we expect that a larger information-sharing network decreases a multiteamer's overall job performance when the respective network ties are, on average, relatively strong. Hence, we hypothesize: Hypothesis 2. Average tie strength moderates the relationship between the size of an employee's information-sharing network and his or her overall job performance. This relationship is positive when an employee's information-sharing ties are weaker, and negative when his or her informationsharing ties are stronger.

MTM's indirect performance consequences
The above reasoning suggests that individual MTM positively associates with the size of an employee's information-sharing network (Hypothesis 1). A greater information-sharing network, in turn, may improve an employee's overall job performance, provided that these network's ties are relatively weak, on average. When the respective informationsharing ties are stronger, by contrast, a larger network is expected to decrease an employee's performance (Hypothesis 2). Logically, then, we would expect that MTM is indirectly related to an employee's overall job performance, through the size of the employee's information-sharing network, in a pattern of moderated mediation (cf. Preacher, Rucker, & Hayes, 2007). This indirect relationship should be positive if the information network ties emanating from an employee's MTM are relatively weak. In such circumstances, higher MTM may enable employees to efficiently access non-redundant informational resources in their organizational environment (Perry-Smith, 2006), to integrate insights from separate teams into new and unique perspectives on important issues (Amabile et al., 1996) and, thus, to achieve superior job performance outcomes across teams.
If the information-sharing network ties resulting from MTM are relatively strong, by contrast, the associated advantages should remain limited (Granovetter, 1973) and the resulting overall performance consequences may reverse. Under these circumstances, higher MTM adds redundant ties that, even though they span multiple teams and social contexts, (1) do not provide many novel and useful information resources and (2) potentially distract from more important tasks and responsibilities (Krackhardt, 1992;Lin, 1999). As depicted in Figure 1, we therefore hypothesize a conditional indirect relationship between individual MTM and overall job performance, with the size of an employee's information-sharing network as a key mediating mechanism and the average strength of the respective network ties as an important moderating factor.
Hypothesis 3. MTM is indirectly related to an employee's overall job performance, by increasing the size of his or her information-sharing network on the job. This indirect relationship is positive when an employee's information-sharing ties are weaker, on average, and negative when his or her informationsharing ties are stronger.

Method
Sample and procedure We examined our hypotheses in a sample of employees from an organization of applied research in the Netherlands. Within this organization, work is mainly conducted in projects formed around specific assignments (e.g., contract research for the Dutch government, military, or private companies). The organization consists of 23 locations across the country, and we specifically targeted researchers from one location in the midwest of the Netherlands (N = 109). This location housed three departments that carried out projects in related research areas (i.e., perception and cognition, training and learning, and human behaviour). We selected this particular location because (1) the respective project teams were predominantly staffed with employees from the three departments located at this location, thus allowing us to obtain a relatively detailed and complete overview of individual employees' information-sharing networks, and (2) employees at this location regularly participated in more than one project at the same time. Altogether, this provided a viable context for examining our conceptual model. We collected time-lagged data from three independent sources and at different time points to minimize common method concerns. First, we used weekly work hour registrations from the host organization's personnel records to capture each participant's MTM. Employees were obliged to register the number of work hours spent on different projects in a very detailed manner. The organization used this information for billing purposes and to calculate project costs, and project managers monitored the accuracy of these registrations. The organization provided these registrations for the 17 weeks before the second stage of our data collection (i.e., the first 17 weeks of 2015). This allowed us to capture respondents' prolonged participation in multiple teams (cf. Van de Brake et al., 2019) and to include only MTM experiences that happened before the measurement of the other study variables (e.g., network characteristics). In the second stage (during week 18 of 2015), a trained assistant approached all employees at the target location to ask for their cooperation in completing a questionnaire that assessed their information-sharing network characteristics (i.e., size and strength). Seven of the potential participants could not be reached due to longer term absences. The remaining 102 employees were informed that participation was voluntary and responses confidential. In the third and final stage (in week 20 of 2015), we distributed another survey among the three department supervisors at the target location, asking them to rate their individual subordinate employees' overall job performance. In accordance with European privacy laws, supervisors only rated the performance of employees who had agreed to participate in the study.
A total of 76 out of the 102 eligible employees (representing a 75 per cent response rate, a percentage similar to prior network studies; e.g., Hansen, 1999;Rost, 2011) completed the social network survey. These individuals (42 males, 36 females) were highly educated (i.e., they had at least a master's degree), mostly worked on a full-time basis (average hours per week = 36), had a mean age of 43 years, and had been working with the organization for an average of 12 years. Moreover, MTM was relatively common among these employees, as they were simultaneously involved in approximately 3 teams per week during the study period. 2

Measures
Multiple team membership Following prior research (Pluut et al., 2014;Van de Brake et al., 2018), we measured MTM as the number of project teams in which an employee was actively and concurrently involved. Specifically, as noted before, we used the organization's weekly work hour registrations to capture the number of teams to which an individual allocated time during each specific week. Hence, the present MTM measure is based on an employee's average number of active teams per weekaligning the time interval used to assess an employee's simultaneous team memberships with typical rhythms and temporal structures in the host organization (as recommended by O'Leary et al., 2011). To construct the final MTM measure, we averaged these scores across the 17 weeks available. 3 Information-sharing network characteristics Consistent with prior studies (Baer, 2010;Hansen, 1999;Perry-Smith, 2006), we used a two-step approach to assess the size and strength of an employee's information-sharing network. In doing so, all participants received an alphabetical list of the 109 researchers employed at the target location. In a first step, we asked the respondents to mark each individual with whom they had exchanged information about work-related topics during the past six months (Mehra et al., 2001;Perry-Smith, 2006). We used this measure to calculate network size. Specifically, to enhance measurement accuracy and fully utilize the information available, we averaged an individual respondent's total number of selfreported (i.e., outdegree) and incoming (i.e., indegree) nominations to assess the size of his or her information-sharing network. We note, however, that an alternative operationalization based solely on outdegree ties (cf. Baer, 2010) yielded virtually identical results for all hypotheses tests. Moreover, we did not require specific network ties to be reciprocated on a dyadic level because participants may have reported contacts that did not respond to our network survey (Perry-Smith, 2006).
In the second step, respondents assessed how frequently they had been in contact with each individual marked as an information-sharing tie in Step 1, using a 5-point scale (1 = 'once in the past six months', 2 = 'once per month', 3 = 'several times a month', 4 = 'several times a week', 5 = 'several times a day'). We used this measure to calculate the strength of an individual's information-sharing network, again averaging his or her selfreported and incoming frequency ratings. As for network size, using an alternative operationalization based solely on self-reported ties left the results for all hypotheses tests virtually unchanged.
Overall job performance Each of the three departments in our sample was led by a different supervisor. These supervisors interacted with our sample employees on a regular basis, and they were formally required, by the organization, to closely monitor their employees' performance. Hence, we asked these supervisors to evaluate the performance of each of the individual employees within their respective departments (Rotundo & Sackett, 2002). Supervisors received a list with the names of their direct subordinates and were asked to rate how each employee performed on three separate criteria, namely overall performance, work efficiency, and work quality (Ancona & Caldwell, 1992;Liden, Wayne, & Sparrowe, 2000). The response scale for these items ranged from 1 (far below average) to 7 (far above average). Cronbach's alpha for the three-item measure was .84.

Control variables
We considered a number of potential controls in our analyses. First, we included two department dummy variables because respondents were nested in three departments (cf. Snijders & Bosker, 1999). Second, research suggests that performance evaluations are often biased by employees' gender (e.g., favouring males over females; Inesi & Cable, 2014), organizational tenure (e.g., favouring employees with lower tenure; Ng & Feldman, 2010), and contractual work hours (e.g., favouring those with more working hours; Van de Brake et al., 2019). We therefore considered these demographics as possible covariates. Third, employees' Big Five personality traits may shape their information-sharing networks (Mehra et al., 2001;Thompson, 2005) and overall job performance (Fang et al., 2015). We therefore assessed these personality characteristics using a Dutch translation of the Big Five Inventory (Denissen et al., 2008). Cronbach's alpha was .77 for openness (10 items), .77 for conscientiousness (nine items), .79 for extraversion (eight items), .74 for agreeableness (nine items), and .78 for neuroticism (eight items).
Fourth, as outlined earlier, we considered MTM variety (O'Leary et al., 2011) as a control variable to illustrate the unique role of our present conceptualization of MTM (i.e., an individual's number of concurrent team memberships). The existing literature does not offer a viable operationalization of MTM variety. Hence, we discussed this issue with a subject-matter expert from the host organization. This person informed us that the organization assigned each project team to a specific 'knowledge domain' (e.g., urban mobility, weapon systems, health technology), with teams operating in different domains exhibiting greater diversity in tasks, technologies, and tools than teams in the same domain. We obtained this information for all project teams in which our respondents worked during the study period. On this basis, we used a Blau Diversity Index to capture an individual's MTM variety (Blau, 1977). This measure ranged from 0 (if all of an employee's teams during the study period were in the same domain) to 1 (if all teams were in different domains). An alternative operationalization in which we first calculated MTM variety during each week during the study period and then averaged across these weeks yielded virtually identical results. 4 Finally, we captured two additional information-sharing network characteristics. MTM may go along with a certain degree of network overlap (i.e., when working with the same colleagues in different teams; O'Leary et al., 2011), and such overlap may affect MTM's performance consequences (Vedres & Stark, 2010). We assessed network overlap by calculating the average number of teams that a respondent shared with his or her reported network contacts (e.g., a score of 2 indicates that, on average, a respondent shared 2 teams with his or her network ties). Moreover, employees with higher MTM may be more likely to span 'structural holes' between otherwise unconnected teams and, as such, experience unique advantages (Borgatti & Foster, 2003). To control for this alternative network mechanism (i.e., beyond network size), we used the Igraph package in R to assess respondents' information-sharing network constraint, (i.e., one of the most commonly used measures to assess structural holes; see Burt, 1992, pp. 50-71, for details). Importantly, network constraint was highly correlated with network size (r = À.88)as is to be expected, because network size is a root construct of network constraint (i.e., constraint is determined by network size, density, and hierarchy; Burt, 1992). Also, controlling for constraint significantly decreased the fit of our overall structural equation model (Δv 2 = 92.54, Δdf = 4, p < .01). Following prior research (Ballinger, Cross, & Holtom, 2016), we therefore excluded this variable from our hypotheses tests. We note, however, that all subsequent results remained robust when network constraint was included.
We repeated all hypotheses tests (1) including all control variables, (2) without any control variables, and (3) including only potent control variables that were significantly related with at least one of the outcome variables in our initial model. We note that the results and conclusions remained virtually identical across these analyses, supporting the robustness of our findings. For reasons of parsimony, we report a model that only includes control variables that are significantly related to respondents' network size and/or overall job performance. Results including or excluding all control variables are available from the first author.

Analytical strategy
We used structural equation modelling in Mplus version 7 (Muth en & Muth en, 1998) to test the hypotheses. Job performance was included as a three-item latent construct, whereas all other variables were included as single-item measures. As recommended by Hayes (2009, see also Maxwell & Cole, 2007, we computed 10,000 bootstrapped parameter estimates and associated 95% confidence intervals to test the conditional indirect relationship between MTM and performance (through the size of an individual's information-sharing network) at higher and lower levels of average network tie strength.
To ease interpretability and be consistent with prior MTM studies (Bertolotti et al., 2015;Van de Brake et al., 2019), we standardized all predictor variables before testing the hypotheses. Table 1 presents means, standard deviations, and correlations for all study variables. As expected, MTM was positively related to the size of an employee's information-sharing network (r = .36; p < .01), and unrelated to the strength of the respective network ties (r = .19; p > .05). Moreover, MTM was positively associated with job performance (r = .43; p < .01). Regarding the control variables, individuals' performance was significantly related with the two department dummies (r = À.46 and .52; p < .01) as well as agreeableness (r = À.25; p < .05), and network size was correlated with gender (r = .28; p < .05), openness (r = .35; p < .01), and network overlap (r = .26; p < .05).

Hypotheses testing
The overall structural equation model (i.e., including the hypothesized relationships, as depicted in Figure 1, and controlling for respondents' departments, gender, work hours, and openness) provided a good fit with the current data (v 2 = 31.16, df = 20, p > .05; RMSEA = .09, CFI = .95, SRMR = .04; cf. Hu & Bentler, 1999). As shown in Table 2, individual employees' MTM was positively associated with the size of their informationsharing networks in this model (B = .39; p < .01). Hence, we found support for Hypothesis 1. Further, as predicted in Hypothesis 2, the relationship between information-sharing network size and overall job performance was contingent on the average strength of an employee's network ties, as indicated by a significant interaction coefficient (B = À.23; p < .01; Aiken, West, & Reno, 1991). Figure 2 depicts the pattern of this interaction at +1 SD and À1 SD values of average tie strength. More specifically, regions-of-significance analyses (using the Johnson-Neyman technique; Preacher et al., 2007) revealed that the linkage between information-sharing network size and job performance was positive and significant at any value of average tie strength lower than 1.8 SD below the mean (i.e., at tie strength values lower than 2.00; estimate at À1.8 SD = 0.34; 95% CI = 0.007-0.680). Conversely, the linkage between network size and overall job performance was negative among employees whose information-sharing network was, on average, relatively strong. This relationship was negative and significant at any average tie strength value greater than 0.7 SD above the mean (i.e., at tie strength values greater than 2.93; estimate at +0.7 SD = À0.23; 95% CI = À0.505 to À0.009). 5 To test Hypothesis 3, we examined the conditional indirect relationship between MTM and job performance, via information-sharing network size, at varying levels of average tie strength. We found this conditional indirect relationship to be positive when average tie strength was relatively low. Again, the positive indirect relationship between MTM and job performance, as mediated by information-sharing network size, reached conventional significance levels at any value of average tie strength lower than 1.8 SD below the mean (indirect estimate at À1.8 SD = 0.13; 95% CI = 0.002-0.303). The conditional indirect relation was negative, by contrast, when average tie strength was .08 Average tie strength À.14 .07 Network size 9 average tie strength À.23** .08 Note. N = 76 individuals. Predictors were standardized and unstandardized coefficients are reported. Non-significant covariates were excluded. *p < .05.; **p < .01. 5 Our approach towards measuring network size and average tie strength also enabled alternative operationalizations of relevant network characteristics, based on a direct count of individuals' relatively weak vs. relatively strong information-sharing ties. We therefore used this approach to examine the robustness of our findings. Specifically, we coded tie scores of 4 ('several times a week') and 5 ('several times a day') as strong, and lower tie scores as weak (cf. Perry-Smith, 2006). Results from the analyses using this alternative operationalization of strong and weak ties corroborated our main findings. Detailed results are available upon request from the first author.
relatively high. The negative indirect linkage between MTM and job performance was significant, in particular, at any value of average tie strength greater than .7 SD above the mean (indirect estimate at +0.7 SD = À0.09; 95% CI = À0.212 to À0.003). In sum, these results offer support for Hypothesis 3. Parenthetically, we note that the direct relationship between MTM and job performance remained significant (B = .37; p < .01) even after accounting for the social network mechanisms postulated in our theorizing (see Table 2). As such, the present results point towards a pattern of partial mediation. Aside from the interaction of network size and tie strength, other mechanisms seem to be present that also account for MTM's performance consequences. We will further address this issue in the Discussion section.

Discussion
The present study aims to provide new insights into why and when individual employees' MTM relates to their overall job performance. Building on the literature on social capital and social networks, we found that individual MTM was positively associated with the size of an employee's information-sharing network. The performance consequences associated with a larger information-sharing network, however, were contingent on the average strength of the respective network ties. As such, the role of MTM for an individual's job performance was double-sided. On the one hand, there was a positive indirect relationship between MTM and job performance, through the size of an employee's information-sharing network, when the respective ties were relatively weak. On the other hand, we found a negative indirect relationship between MTM and job performance when an employee's information-sharing network ties were relatively strong.

Theoretical implications
Together, these findings make several important contributions to the literature on teams, MTM, and social capital in organizations. With many employees' work in modern organizations being distributed across multiple concurrent teams, a burgeoning literature has developed on the performance consequences of such individual MTM (Margolis, 2020). In this regard, a prominent assumption in the MTM literature is that social network mechanisms represent a key explanatory factor (Mortensen et al., 2007;O'Leary et al., 2011). It is important to note, however, that the empirical MTM research has not examined such mechanisms to date. Hence, scholars have called for empirical studies on the social network structures associated with individuals' concurrent memberships within different teams to better understand why MTM may shape an employee's performance outcomes (O'Leary et al, 2011;Van de Brake et al., 2018). The present investigation responds to these calls, illustrating that MTM's performance consequences result, at least partially, from the implications such work arrangements have for the size of an employee's information-sharing network. Hence, our findings contribute to theory advancement in the MTM literature, highlighting the usefulness of a social network perspective in explicating this construct's individual-level implications.
Moreover, our findings address ambiguity in the current literature on the performance benefits and detriments of individual MTM. On the one hand, our study supports the notion that MTM can promote employees' job performance by providing them with important social capital resources, as indicated by a large information-sharing network (cf. Van de Brake et al., 2018, 2019. Importantly, however, these advantages only materialize if the respective network's ties are relatively weak, on average, thus offering access to novel, unique, and nonredundant information. On the other hand, our study also points towards potentially negative consequences of MTMnamely if the larger information-sharing network resulting from higher MTM is characterized by relatively strong ties. This finding is consistent with the critical sentiments some scholars have voiced against MTM, arguing that such work arrangements can distract employees from effective task accomplishment by focusing their efforts on relationship maintenance and adjustment to different team contexts (Leroy, 2009;Pluut et al., 2014). Hence, beyond explicating why MTM may shape individual employee's overall job performance, our findings illustrate when MTM's performance advantages or disadvantages are most likely to prevail. The present study therefore promotes a more integrative perspective on individual MTM's performance consequences, reconciling seemingly contradictory perspectives advanced in prior research.
Finally, our findings contribute to theory development on the origins of social capital in organizations. To date, most research on this issue has focused on employees' personality traits, demographic characteristics, and behavioural tendencies (Ibarra, 1993;Marsden & Friedkin, 1993;Mehra et al., 2001), whereas only a few studies have examined the role of organizational practices (for an exception, see Zaheer & Soda, 2009). The present investigation addresses this shortcoming, highlighting MTM as a key factor that may enable individuals to enlarge their information-sharing network on the job, with important implications for their overall job performance. These results show that, beyond individual factors, contextual aspects play a relevant role in social capital development, broadening our understanding of the social capital construct.

Limitations and future research directions
We acknowledge some limitations that future MTM research could address to further develop this field of inquiry. Although our study has methodological strengths (e.g., multi-source, time-lagged data), for example, our relatively small sample came from a single organization in one country, the Netherlands. This may limit the extent to which our results can be generalized to other organizational and cultural contexts. Also, our social network survey assessed respondents' information-sharing interactions in the preceding six months (i.e., November-April 2015), whereas the archival data we received from the host organization enabled us to calculate MTM for a four-month period (i.e., January-April 2015). Clearly, it would have been preferable if our MTM and social network data fully covered the same time frames. Importantly, however, we believe the supplementary analyses, as reported in Footnote 3, alleviate associated concerns to some extent, illustrating that our findings are robust when utilizing different time frames for the MTM measure (i.e., between one and four months).
Further, the social network survey did not capture information-sharing relationships with coworkers at other host organization locations. This is a consequence of our rosterbased measurement approach, which required us to present respondents with a limited list of potential contacts (see also Perry-Smith, 2006). We note that this approach has considerable advantages in comparison to alternative, free-recall methods of collecting social network data (where respondents may forget relevant contacts; Marsden & Friedkin, 1993;Scott, 2017). Also, we deliberately focused on a location within the host organization that primarily staffed its project teams with internal employees, such that potential information-sharing ties with other locations were less relevant. Nevertheless, future research may benefit from capturing more distal information-sharing ties as well (e.g., by allowing respondents to list additional contacts to the roster), thus obtaining a more complete picture of individuals' information-sharing networks.
More generally, our social network approach is based on the notion that a larger number of network ties is more beneficial if these ties are relatively weak (rather than strong), for example, because weaker ties can provide more novel and unique information resources, whereas stronger ties require greater maintenance. We note that this logic is consistent with a considerable body of social network theory and research (Borgatti & Foster, 2003;Granovetter, 1973;Kwon & Adler, 2014). Nonetheless, it is clear that we did not directly examine these mechanisms. Hence, future research specifically investigating the resource and social exchange processes underlying our suggested relationships (e.g., by measuring employees' access to task-related information, novel knowledge sources, and useful materials; Krackhardt, 1992) may provide further confidence in our theorizing.
Moreover, we note that we assessed individuals' overall performance on the job, thus potentially neglecting more nuanced aspects of an employee's performance within specific teams (cf. Rapp & Mathieu, 2018). With MTM enabling an employee to transfer new information from one team to another, a team may benefit more if it receives (rather than provides) such information (Vedres & Stark, 2010). As a result, multi-teamers' contributions to joint task accomplishment may be more pronounced in teams to which they provide novel information, therefore enabling them to experience greater performance advantages in such teams (as opposed to the teams from which they obtain information). Future research differentiating such team-specific performance aspects may advance a more detailed understanding of the possible benefits and disadvantages associated with individual MTM.
Finally, although our study illustrated information-sharing network size and average tie strength as important factors in the MTM-performance linkage, these social network characteristics only partially explained MTM's performance consequences (i.e., MTM's direct relationship with job performance remained significant in our moderated mediation model). Hence, the relevance of network aspects notwithstanding, future research could consider additional mediating mechanisms to provide a more comprehensive perspective on the linkage between individual MTM and performance.

Practical implications
Although many managers recognize the increasing prevalence of MTM, few have a thorough understanding of how it affects their employees (see Mortensen & Gardner, 2017). The present study addresses this problem and, as such, has direct implications for the design and management of contemporary jobs. We found that a key consequence of MTM is that it increases individual employees' information-sharing networks. An employee's ability to effectively navigate such increased networks, however, critically depends on the extent to which the respective connections are either relatively weak (i.e., based on relatively infrequent, low-intensity interactions) or strong (i.e., more intense).
More specifically, our findings suggest that managers must encourage employees with high MTM to develop relatively weak information-sharing interactions across their teams, thus offering the potential for substantial performance advantages from such work arrangements. Managers could achieve this by regularly conducting social network assessments among their employees, which would give them a detailed overview of the organization's information-sharing network (for practical examples, see Kleinbaum & Tushman, 2008;Valente, 2012). If this network points to employees who tend to develop stronger and more intense information-sharing ties, managers should either keep these employees' MTM relatively limited or assign them to tasks that require more modest amounts of interpersonal information-sharing.

Conclusion
The present research uncovered important social network mechanisms that explicate how an employee's MTM may shape his or her overall job performance. Our findings demonstrate that MTM's network implications are a double-edged sword that can both improve and harm an employee's functioning on the job, thus contributing to a more complete understanding of MTM's consequences for individual employees. We hope these findings will stimulate further research on employees' MTM, in general, and on their social network characteristics within and across teams, in particular, helping organizations to more effectively manage complex multi-team arrangements.

Data availability statement
The data used in this manuscript are not publicly available. Participation in the study was voluntary and based on the assurance that individual responses would be treated confidentially by the members of the research team. Given the size of the sample and the presence of individual-level demographics, it could be possible to identify single participants. We can thus not make the data publicly available to protect the anonymity of research participants. Furthermore, we draw from confidential organizational archival data (i.e., project team information, billable work hours) to operationalize multiple team membership. However, we have permanently deposited the data and study materials on the servers of the University of Groningen for possible later quality assessment.