Content area
This research applies and integrates transactive memory systems (TMS) theory and the Big Five personality traits model to investigate the performance dynamics of dyadic teams engaged in virtual collaborative problem-solving (CPS). Specifically, this study examines how the personal attributes of team members, including their expertness and Big Five personality traits (extraversion, agreeableness, openness, conscientiousness, and neuroticism), as well as the resultant diversity in expertness and Big Five personality traits within teams, influence both team-level and individual-level performance gain from virtual collaboration. Studying 377 dyadic teams composed of 754 individuals working on an online collaborative intellective task, this research found that dyads with high expertness diversity had greater performance gain from virtual collaboration than dyads with low expertness diversity. Further, dyads, where both members scored low on agreeableness, showed the most significant improvement in team performance. At the individual level, a team member who had a low expertness level but was paired with a high-expertness teammate demonstrated the greatest performance gain from virtual collaboration. The integration of TMS theory and the Big Five personality traits model provides a richer and more nuanced understanding of how individual attributes and team dynamics contribute to successful virtual CPS outcomes.
Who benefits from virtual collaboration?
The interplay of team member expertness and Big Five personality traits
Collaborative problem-solving (CPS) is widely recognized as a critical 21st-century skill and is essential for team success in today’s workplace (Fiore et al., 2017). CPS teams can collaborate in person (Xu et al., 2023), virtually (Yang et al., 2022), or in hybrid formats (Pramila-Savukoski et al., 2023). Notably, the prominence of virtual and hybrid CPS modes has surged in the wake of the COVID-19 pandemic (Karl et al., 2022). Compared to face-to-face communication and collaboration, virtual CPS generally entails decreased levels of information richness (Aritz et al., 2018) and social presence (Fonner and Roloff, 2012), which may affect how knowledge can be accurately recognized (Su, 2012), effectively coordinated (Kanawattanachai and Yoo, 2007), and efficiently retrieved (Yuan, Carboni, et al., 2010) within the team. Although there has been a growing research emphasis on virtual collaboration and team dynamics, much of the focus has been on virtual communication processes such as verbal communication patterns (O’Bryan et al., 2022), nonverbal exchange of expertise cues (Yoon and Hollingshead, 2010), task-oriented communication and knowledge coordination (Kanawattanachai and Yoo, 2007), and modes of computer-mediated communication (Tang et al., 2014). Therefore, there remains a pressing need for further investigation into how personal attributes and the resultant diversity of these attributes impact virtual collaboration outcomes in CPS teams.
A personal attribute pivotal to CPS outcomes is individual expertise, which refers to the type and level of knowledge a person possesses (Di Gangi et al., 2012). In order to achieve optimal CPS outcomes, today’s work teams are typically assembled to include individual members who possess different types of expertise across multiple knowledge domains (conceptualized as expertise diversity) and who hold different levels of expertise within the same knowledge domain (conceptualized as expertness diversity) (Lee et al., 2022; Martins et al., 2013; Van der Vegt et al., 2006). While the effects of expertise diversity (i.e., inter-domain expertise difference) on team performance are well-documented (Lee et al., 2022; Zheng et al., 2016), especially through the lens of Transactive Memory Systems (TMS) theory (Lewis and Herndon, 2011; Wegner, 1986, 1995), the impact of expertness diversity (i.e., intra-domain expertise difference) on team performance remains largely understudied.
In a comprehensive review of 64 TMS-related empirical studies, Yan et al .(2021) asserted that existing research on TMS overly emphasizes expertise-related factors while neglecting other individual attributes that also play a critical role in affecting TMS development and team collaboration. Among these attributes, individuals’ personality traits stand out as crucial factors that significantly influence team performance and collaboration outcomes (Stadler et al., 2019; Zhong et al., 2019). Despite extensive research on personality traits across various contexts (see Huang et al., 2014), little attention has been given to investigating their impact on team performance through the operation of TMS development (Pearsall and Ellis, 2006). To address this gap, this study proposes an integrated theoretical framework that leverages both TMS theory and the Big Five personality traits model to explore the performance dynamics of virtual CPS teams. Specifically, the goal of this research is to conceptually analyze and empirically test how team members’ expertness and Big Five personality traits (extraversion, agreeableness, openness, conscientiousness, and neuroticism), as well as the resultant diversity in expertness and Big Five personality traits within teams, could influence CPS performance changes through their impact on TMS mechanisms. The integration of TMS theory and the Big Five personality traits model provides a richer and more nuanced understanding of how individual attributes and team dynamics contribute to successful virtual CPS outcomes.
Theories and research questions
Transactive memory systems (TMS) Theory
Transactive memory systems theory originated to examine the processes and effects of expertise recognition and knowledge coordination in dyads and small groups (Wegner, 1986, 1995). A transactive memory system (TMS) is developed through team members’ perceptions of others’ credibility (who is a trustworthy source of expertise), each member’s knowledge specialization (who is an expert in what knowledge domains), and knowledge coordination within the team (with whom to process and exchange knowledge) (Lewis and Herndon, 2011). In essence, a TMS is composed of the distributed knowledge possessed by each individual member and a shared understanding of “who knows what” in the team (Yan et al., 2021).
A TMS can develop into two different types of structures: differentiated and integrated (Gupta and Hollingshead, 2010; Wegner, 1986). When a team TMS is highly differentiated, individual members hold specialized expertise in different knowledge domains (Wegner, 1986). In other words, knowledge in a particular domain is possessed by only one or very few individuals rather than being widely distributed among all team members (Gupta and Hollingshead, 2010; Wegner, 1986). This allows team members to focus on their areas of expertise, reducing their workload and avoiding expertise overlap within the team (Hollingshead, 2000). While each member focuses on one’s expertise domains, the team collectively gains access to a broader range of expertise, leading to improved team performance (Liang et al., 1995; Moreland and Myaskovsky, 2000).
An integrated TMS, in contrast, forms “when the same items of information are held in different individual memory stores, and the individuals are aware of the overlap because they share label and location information as well” (Wegner, 1986, p. 204). This means in a team with an integrated TMS structure, team members possess expertise in the same knowledge domains and information is broadly distributed among team members (Gupta and Hollingshead, 2010; Wegner, 1986). While the integrated TMS structure leads to overlapped knowledge within the team, it can be beneficial to those teams in which each individual member is required to carry out every function of the job activity without relying on teammates’ expertise (Wegner, 1986) and those teams aiming to solve intellective problems with clear, uncontroversial answers or solutions (Gupta and Hollingshead, 2010).
Existing TMS research has delved into the mechanisms that foster the development of differentiated and integrated TMS structures, respectively. Hollingshead (2001) found that a TMS became most differentiated when individuals had incentives to remember different information and most integrated when individuals had incentives to remember the same information. A later study showed that teams with mixed-gender friendships were more likely to develop a differentiated TMS, while teams with strong feelings of closeness tended to create an integrated TMS (Iannone et al., 2017). Further, it was found that teams with role-specific preparation developed more differentiated TMS, which led to better performance compared to teams whose members received cross-role preparation (Linton et al., 2018).
An abundance of TMS scholarship has also shed light on how differentiated and integrated TMS structures affect team performance in various contexts. While both differentiated and integrated TMS structures yield similar results on memory-based tasks (e.g., recall tasks), an integrated TMS leads to faster and more accurate performance on intellective tasks (e.g., tasks requiring critical thinking and reasoning) compared to a differentiated TMS (Gupta and Hollingshead, 2010). In addition, an integrated TMS fosters collaboration and mutual support among team members, while a differentiated TMS provides a clear division of responsibilities (Gupta and Hollingshead, 2010). Additionally, Gupta (2012) found that team members in an integrated TMS tended to cover work tasks for the noncontributors, whereas those in a differentiated TMS were likely to facilitate other team members in performing their jobs. Moreover, a differentiated TMS was found to exacerbate the negative effects of polychronicity diversity (wherein some team members prefer single-tasking while others prefer multitasking) on team performance (Mohammed and Nadkarni, 2014). Lastly, a differentiated TMS enhances centralized information seeking when specialized expertise is needed, but an integrated TMS promotes team cohesion and social interactions within the team (Yan et al., 2021).
Expertise vs. expertness diversity
The classification of differentiated and integrated TMS structures in TMS scholarship has provided a theoretical foundation for the conceptualization of two forms of task-related cognitive diversity in team research: expertise and expertness diversity (Martins et al., 2013; Van der Vegt et al., 2006). While expertise diversity refers to the variation of individuals’ expertise in different knowledge domains (i.e., inter-domain expertise difference), expertness diversity represents team members’ differences in the level of expertise within the same knowledge domain (i.e., intra-domain expertise difference) (Lee et al., 2022; Van der Vegt et al., 2006). Both expertise and expertness diversity are common and desirable for teamwork because today’s work teams need the breadth of knowledge to handle wide-ranging tasks and the depth of expertise to tackle complex problems effectively. For example, a website development team may be composed of members who are specialized in different web-design knowledge domains such as front-end HTML and CSS coding, graphic design, search engine optimization, and server-side scripting skills, which reflect the expertise diversity within the team. Meanwhile, within the specific domain of graphic design, there may exist some degrees of expertness diversity such that one member is a very skillful and experienced graphic designer whereas another member may only have entry-level knowledge about graphic design.
Both expertise and expertness diversity are closely related to the development of differentiated and integrated TMS structures (Lewis, 2003). On the one hand, without the distribution of expertise across multiple knowledge domains (i.e., expertise diversity), the differentiated TMS structure would not have become feasible and sustainable within the team (Littlepage and Mueller, 1997). On the other hand, expertness diversity reflects how an integrated TMS structure can function within a particular knowledge domain (Van der Vegt et al., 2006). Consider a team with a well-developed integrated TMS structure; while team members possess similar expertise in shared knowledge domains, it is unlikely for all team members to have exactly the same level of expertise in one knowledge domain. Thus, expertness diversity delineates how team members differ in their level of expertness in a single knowledge domain within the context of an integrated TMS structure (Martins et al., 2013). This perspective offers deeper insights into how an integrated TMS structure can manifest at the domain level, supplementing and extending traditional TMS research that primarily focuses on cross-domain knowledge distribution (Lewis and Herndon, 2011).
Prior research suggests that expertise and expertness diversity may influence team performance through different mechanisms (Martins et al., 2013; Mathieu et al., 2008). While expertise diversity can foster team learning by exposing members to new knowledge and perspectives, expertness diversity may stimulate individual learning when less experienced members seek guidance from more skilled teammates (Van der Vegt et al., 2006). However, it is worth noting that high-expertness members may not always be willing and committed to helping low-expertness members (Van der Vegt et al., 2006). The influence of these factors can be further affected by team dynamics. For example, when team members’ psychological safety was low, expertise diversity could hinder team performance, while expertness diversity could improve team performance (Martins et al., 2013). Further, team members’ expertness diversity could moderate the relationship between expertise diversity and team performance such that the positive effect of expertise diversity on team performance diminished when team members’ expertness diversity was high (Zheng et al., 2016).
Despite a considerable body of research that has documented the effects of expertise diversity on team performance (e.g., Lee et al., 2022; Shi and Weber, 2018; Zheng et al., 2016), the exploration of expertness diversity’s impact on team performance, particularly virtual CPS outcomes, has been extremely limited. Therefore, a primary goal of this study is to address such a research gap by examining how team members’ expertness level, as well as the expertness diversity within the team, could affect their performance in virtual CPS settings. To this end, we will begin by elucidating the key characteristics of virtual CPS and examining their potential impact on the development and operation of TMS within virtual CPS teams. Subsequently, we will explore how the expertness diversity within the team may affect TMS dynamics and ultimately influence virtual CPS outcomes.
Virtual collaborative problem-solving
Collaborative problem-solving (CPS) is the process of two or more individuals working together to understand, analyze, and resolve problems, typically through interpersonal communication, cooperation, and collective decision-making (Xu et al., 2023). CPS entails two critical dimensions: collaboration and problem-solving (Fiore et al., 2017). Collaboration encompasses the communicative, social, and interactive elements of CPS, while problem-solving pertains to the task-oriented, cognitive reasoning, intellective, and decision-making aspects of CPS (Fiore et al., 2017). These two dimensions necessitate the application of TMS theory to better understand the processes and outcomes of CPS because the success of CPS is, by and large, influenced by effective responsibility allocation, expertise recognition, knowledge retrieval, and information coordination, all of which are fundamental mechanisms of TMS development (Austin, 2003).
Previous research on CPS and TMS has studied a wide array of CPS teams, investigating how factors like team size, task characteristics, participant backgrounds, and communication modalities impact TMS development and team performance. Many of these studies, particularly early TMS research, focused on dyadic teams (i.e., teams composed of two members only). It was found that in dyadic teams aiming to complete memory recall tasks, intimate relationships and access to nonverbal or paralinguistic cues supported TMS development and improved team performance (Hollingshead, 1998a, 1998b). Moreover, in dyadic teams engaged in collaborative quiz questions regarding job-related knowledge, team performance improved when team members had highly diverse expertise levels and allocated more work to the higher-expertness member (Littlepage et al., 2008). In educational settings where students collaborated on team projects, knowledge-based learning, and social communication facilitated TMS development and enhanced team performance (Jackson and Moreland, 2009; Zhang et al., 2016). For more formal, non-ad hoc teams such as those in project development for new products (Akgün et al., 2005) or information systems (Hsu et al., 2012), frequent formal team communication emerged as a crucial factor for TMS development and team success. The nature of CPS tasks also plays a significant role, as research has shown that for exploratory tasks (i.e., searching for new knowledge to resolve problems), informal and face-to-face communication positively correlates with TMS development, while tasks driven by exploitation (i.e., using existing knowledge for problem-solving) benefit from formal and computer-mediated communication (Tang et al., 2014).
Given the ubiquitous implementation of remote and online collaborative work escalated by the COVID-19 pandemic (Maurer et al., 2022), this study focuses on the context of virtual CPS. Virtual CPS involves two or more geographically dispersed individuals working together to solve problems by using ICTs (information and communication technologies) (Jiang et al., 2023; Xu et al., 2023). Virtual CPS teams commonly utilize online tools such as video conferencing platforms (Karl et al., 2022), collaborative document editing software (Breuer et al., 2016), and project management platforms (Soboleva et al., 2021) to facilitate their communication, collaboration, and coordination. Team members can be situated anywhere locally and globally, and the affordances of virtual technologies can facilitate their boundary management by overcoming temporal and geographical constraints (Navick and Gibbs, 2023; Sivunen et al., 2023). While virtual CPS often adheres to a structured problem-solving framework, its execution allows for flexibility, allowing asynchronous participation and adapting to different work styles and time zones (Yang et al., 2022).
The virtual CPS’ unique characteristics can create challenges that hinder the positive effects of TMS development and expertness diversity on team performance for the following reasons. First, due to the absence or decreased levels of information richness (Aritz et al., 2018) and social presence (Fonner and Roloff, 2012) in virtual CPS, communication and coordination may be more difficult and time-consuming between high-expertness and low-expertness members, especially when they need to transfer complex knowledge to tackle equivocal problems. Second, when team members lack face-to-face interactions, there may be limited social opportunities for them to observe and process communication cues that are essential in shaping accurate assessments of other members’ expertness level (Su, 2012) and fostering efficient knowledge retrieval from the high-expertness members (Yuan, Carboni, et al., 2010). Thirdly, in large virtual teams characterized by expertness diversity, there is an elevated risk of social loafing (Price et al., 2006), wherein certain members withhold their contribution by freeriding other members’ contributions, resulting in further reluctance among high-expertness members to assist their less knowledgeable counterparts (Van der Vegt et al., 2006), consequently impeding overall team performance.
Despite these challenges, existing research suggests that TMS can be developed within virtual teams (Kanawattanachai and Yoo, 2002), which bolsters the potential for expertness diversity to enhance team performance in virtual CPS settings. While virtuality may complicate the accurate identification of team members’ expertise, past research found that leveraging digital knowledge repositories within work teams can mitigate the negative impact of virtual work on accurate perception of team members’ expertness levels (Su, 2012). Furthermore, research indicates that in geographically dispersed virtual teams, the use of ICTs facilitates knowledge exchange between team members with varying levels of expertness, promoting learning and growth within the team (Shi and Weber, 2018). Additionally, various features available on virtual collaboration platforms have been found to positively influence multiple dimensions of TMS development (Yoon and Zhu, 2022). Notably, the virtual technologies’ visibility and searchability functions can enhance TMS accuracy, while the awareness and pervasiveness features promote TMS sharedness, and the editability and self-presentation capabilities contribute to TMS validation (Yoon and Zhu, 2022). These findings resonate with social information processing theory (Walther, 1992, 2011), which asserts that in virtual communication settings, individuals have the motivation and capabilities in compensating for the absence of nonverbal cues by harnessing remaining information cues (e.g., texts, emojis, punctuation, and timing of responses) to help them better process and interpret information messages. In addition, these studies support the theory of communication visibility (Treem et al., 2020, 2024), which suggests the affordance of digital communication tools can enhance the visibility of team member communication and interaction, thereby promoting knowledge sharing and coordination within the team.
Drawing upon these research findings, this study posits that virtual CPS teams with varying degrees of expertness diversity might experience different performance outcomes. In teams characterized by high expertness diversity (i.e., comprising both high-expertness and low-expertness members), a reciprocal learning dynamic may occur. On the one hand, high-expertness members can validate and refine their existing expertise through coaching and clarifying knowledge to low-expertness members. On the other hand, low-expertness members can learn and acquire knowledge from high-expertness members, fostering their intellectual growth and problem-solving skills. The knowledge transfer and exchange between high- and low-expertness members not only mitigates the risks of groupthink (Janis, 1971; McCauley, 1998), where uniformity of knowledge stifles innovation and critical inquiry but also enhances the entire team’s problem-solving capabilities. Further, while the affordances of virtual technologies can make high-expertness members’ knowledge more visible and accessible (Treem et al., 2020), the transfer of deeply embedded or complex knowledge (Haas and Hansen, 2007) from high-expertness members to low-expertness members may be time-consuming and ineffective. Additionally, the absence of face-to-face communication may exacerbate the difficulty of managing conflicting viewpoints stemming from expertness diversity. These challenges may ultimately hinder overall team performance in virtual CPS settings.
Virtual CPS teams with low expertness diversity can embody two possible team compositions: homogeneous high-expertness and homogeneous low-expertness. In teams comprised of homogeneous high-expertness members, experts can utilize their collective knowledge to identify and resolve problems efficiently. However, the uniformity of expertness can create a culture of overconfidence, which might blind these experts to potential errors. Such culture may also exacerbate confirmation bias (Tschan et al., 2009), leading team members to favor information that confirms existing beliefs and hindering knowledge sharing within the team. Moreover, in virtual CPS teams where social interactions and rapport-building opportunities are limited, high-expertness members may be more likely to engage in competitive and ego-centric behaviors such as knowledge hiding (Connelly et al., 2019), ultimately undermining the team’s overall performance.
Teams composed solely of low-expertness members face significant challenges with complex problem-solving and decision-making due to their limited knowledge base. The absence of high-expertness members within the team can result in prolonged searches for optimal solutions that may require higher levels of expertness. In addition, low-expertness members may face protracted debates over the validity and merits of existing ideas, as they lack sufficient levels of expertness to make informed judgment calls. However, these teams often exhibit openness to novel ideas and a strong desire to acquire new knowledge, making them more adaptable and resilient in dynamic environments where conventional expertise may prove inadequate to address ever-changing problems. Further, while virtual collaboration platforms offer opportunities to tap into external expertise to compensate for internal deficiencies, the lack of face-to-face communication in virtual CPS environments can make it more difficult to build trust (Choi and Cho, 2019), which is essential for collective learning and error correction among low-expertness members.
Therefore, considering the potential positive and negative effects of varying degrees of expertness diversity on team performance in virtual CPS, this study proposes the following research question:
RQ1: At the team level, how is the team’s performance gain from virtual collaboration influenced by the team’s expertness diversity?
While prior TMS and cognitive diversity research has primarily focused on team-level performance outcomes, it is evident that each individual team member performs and benefits from CPS differently (Lee et al., 2022). Specifically, an individual member’s performance gain from virtual collaboration may depend on not only this member’s own expertness level but also the expertness level of the teammates. In a virtual CPS setting, pairing teammates with varying degrees of expertness may influence individual performance through two key mechanisms of TMS development: knowledge storage and information retrieval. First, individuals with higher expertise in a specific domain can contribute deeper and more accurate knowledge to the TMS in that area, enabling the establishment of a high-quality knowledge repository to be utilized by the low-expertness members. However, in face-to-face settings, these experts may face challenges managing overwhelming knowledge requests from low-expertness members, leading to stress and distractions that hinder their productivity (Su et al., 2010). Conversely, in virtual CPS settings, high-expertness members can leverage knowledge management systems such as cloud storage and databases to disseminate their expertise efficiently, thereby saving time on direct communication with low-expertness members and enhancing their productivity (Yuan et al., 2007).
Second, when it comes to information retrieval, virtual CPS presents both opportunities and challenges for low-expertness members. On the one hand, virtual collaboration platforms offer specialized repositories and designated channels for low-expertness members to retrieve knowledge from experts without being constrained by geographical or temporal barriers (Di Gangi et al., 2012). Enhanced search functionalities and clear metadata tags streamline the process for lower-expertness members to identify expertise locations and access knowledge efficiently. On the other hand, low-expertness members may encounter difficulties in obtaining timely feedback and clarification from high-expertness members, particularly when retrieving and applying complex knowledge for problem-solving (Haas and Hansen, 2007). Moreover, excessive reliance on virtual communication tools for information retrieval can induce fatigue and stress (Luebstorf et al., 2023), potentially impeding the productivity and efficiency of low-expertness members.
Therefore, an individual’s expertness level, together with the expertness level of the teammate, is expected to intricately impact one’s performance outcomes in CPS settings. Given the limited research on the impact of these dynamic environments on individual performance in virtual CPS, this research proposes the following research question:
RQ2: At the individual level, how is a team member’s performance gain from virtual collaboration influenced by one’s own expertness level and that of the teammate?
Big five personality traits
While the above research questions delve into how team members’ expertness levels, as well as their expertness diversity, could affect virtual CPS performance through the operation of TMS, other personal attributes, such as personality traits, can also influence TMS development and consequently lead to different team collaboration outcomes (Pearsall and Ellis, 2006). However, the scholarship on the relationship between personality traits and TMS development is extremely scarce. Thus, it is unclear how team members’ personality traits, as well as their personality traits diversity, could influence their virtual CPS performance from the TMS perspective. Therefore, to further enrich the theoretical framework, this study integrates the investigation into how each of the Big Five personality traits, a well-established personality traits model, can affect individual and team CPS performance through their effects on TMS development in virtual settings.
The Big Five personality traits model (Goldberg, 1990; McCrae and Costa, 2008) provides a multidimensional theoretical framework and assessment instrument to characterize individual personality traits into five categories: openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism (i.e., the lack of emotional stability). The first trait, openness to experience, reflects an individual’s tendency to be curious, imaginative, and open to new ideas and experiences (Judge et al., 2002). Previous research found that individuals high in openness to experience were more likely to learn and explore diverse information sources and engage in knowledge sharing (Yin et al., 2023). Thus, in teams whose members share the openness trait, TMS may be developed more quickly and effectively. Further, openness to experience may enhance an individual’s adaptability in virtual environments, because individuals high in this trait readily embrace new collaboration tools and processes, navigating the inherent ambiguity and uncertainty with greater comfort in unfamiliar settings (Huang et al., 2014). These factors collectively contribute to a potential positive impact of the openness personality trait on virtual CPS performance by promoting intra-team knowledge sharing and learning, as well as team members’ ability to explore uncharted territories and tackle challenging tasks. However, a team with diverse openness levels, where some members are highly open and others less so, can also encounter challenges. As individuals with lower openness may be less inclined to explore new information or collaborate closely on innovative tasks (McCrae and Costa, 2008), teams with greater diversity in openness levels may face hurdles in virtual collaboration compared to teams where members are uniformly high in openness.
The conscientiousness trait represents a person’s level of organization, self-discipline, and reliability. Individuals high in conscientiousness tend to be detail-oriented, dependable, and meticulous in task performance (Sackett and Walmsley, 2014). Previous studies found that highly conscientious individuals were effective learners, which helped them develop a higher level of expertise (Poropat, 2009). From a TMS perspective, highly conscientious team members excel at organizing and storing information, making it easier for others to retrieve and utilize such information for task completion. They also contribute to establishing clear communication and documentation practices that facilitate team members’ expertise recognition and directory updating, two crucial components of TMS development (Yuan, Fulk et al., 2010). Further, in a virtual collaboration setting, the conscientiousness trait may be instrumental in creating precise structures for information sharing and knowledge retrieval through technology-mediated platforms, boosting the effectiveness of virtual TMS development. Conscientiousness may also urge an individual member to meet deadlines and follow through on tasks in a virtual setting where direct supervision may be less present. However, overly conscientious individuals may become fixated on flawless execution, hindering collaboration and innovation (Robert and Cheung, 2010). Research also suggests that team performance may be impeded in teams of highly diverse conscientiousness because highly conscientious members may be frustrated and irritated by working with those with lower conscientiousness (Shoss et al., 2015).
The next dimension, extraversion, maybe the most researched personality trait among all Big Five personality traits. This trait reflects a person’s level of sociability, talkativeness, and assertiveness. Extraverted individuals are more likely to ask questions, share information, and initiate conversations (Wilmot et al., 2019), fostering knowledge exchange that is crucial for TMS development. Indeed, it was found that extraversion had the greatest impact on knowledge sharing among all Big Five personality traits (Yin et al., 2023). Previous research has also revealed that critical team members’ assertiveness, a subdimension of the extraversion trait, facilitated TMS development and consequently improved team performance because assertive members could effectively stimulate the development of specialization, credibility, and coordination within the team (Pearsall and Ellis, 2006). However, overly extroverted team members may dominate discussions, hindering contributions from introverted colleagues (McCord et al., 2014). Further, extraverts may find it more challenging to build rapport and maintain engagement in a virtual environment where nonverbal cues are limited. In sum, when completing tasks that require frequent and extensive interpersonal interactions, such as CPS, teams composed of mostly extraverts are likely to excel (Sackett and Walmsley, 2014). However, the advantages of extraversion may be less pronounced in teams with highly diverse extraversion levels or those composed mainly of introverts (Wilmot et al., 2019).
The agreeableness trait signals an individual’s tendency to be cooperative, trusting, and helpful (Judge et al., 2002). Highly agreeable individuals are inclined to foster collaboration, maintain positive dynamics, and promote trust (Huang et al., 2014), all of which facilitate TMS development within the team. Specifically, agreeable members embrace group harmony and readily share information and expertise, creating a foundation for TMS development. Their cooperative nature and conflict-resolution skills contribute to open communication, essential for the effective operation of TMS (Neff et al., 2014). Further, agreeableness fosters trust, lowering the barriers to seeking and retrieving information from others within the team’s knowledge base. However, agreeableness can also present challenges. For example, overly agreeable individuals may prioritize maintaining positive relationships over task completion (Fang et al., 2015), which may impede the team’s ability to tackle urgent issues in a timely manner. They are more likely to shy away from difficult conversations that are sometimes crucial for finding solutions to complex problems. This limitation may be particularly salient in virtual CPS as nonverbal cues and immediate reactions are diminished, making it even harder to gauge teammates’ sentiments and even longer to achieve team cohesion. Similar to the conscientiousness trait, teams with high agreeableness diversity might experience decreased productivity. Highly agreeable members may find it challenging to navigate interactions with less agreeable teammates, potentially leading to greater agitation and more struggles.
Neuroticism, the last Big Five personality trait, indicates an individual’s susceptibility to negative emotions like anxiety, fear, and sadness. Individuals high in neuroticism are more prone to communication difficulties, as anxiety and negativity can hinder clear and effective communication, impacting team collaboration (Silvester et al., 2014). A tendency towards negativity can make it challenging to build trust, a crucial foundation for information sharing within a team. Further, fear of judgment or criticism might lead them to withhold valuable knowledge (Hernaus et al., 2019), hindering the development of TMS. The negative impact of neuroticism on TMS development and team performance may be amplified in virtual environments. The physical disconnection and limited social cues inherent in virtual work can exacerbate anxiety for those high in neuroticism (Huang et al., 2014). In addition, the absence of nonverbal cues can increase the risk of misinterpretations and misunderstandings, further straining communication. Finally, managing stress effectively can be harder in a virtual setting, potentially leading to decreased performance for highly neurotic individuals (Judge et al., 2002).
In sum, the diversity of Big Five personality traits within a team may exert potential positive and negative impacts on TMS development, thereby affecting overall team performance. In addition, an individual’s personality traits, together with the personality traits of the paired teammate, are expected to influence one’s performance outcomes in CPS settings. These dynamics are further complicated by virtual CPS as the virtual environment may hinder the full expression and perception of personality traits during communication and collaboration processes. Considering this interplay between individual personality traits and team personality diversity, this study proposes the following research questions:
RQ3: At the team level, how is the team’s performance gain from virtual collaboration influenced by the team’s Big Five personality traits diversity?
RQ4: At the individual level, how is a team member’s performance gain from virtual collaboration influenced by one’s own Big Five personality traits and those of the teammate?
Method
Participants and procedures
To address these questions, this research conducted an experimental study of 377 ad hoc dyadic teams. We chose to focus on dyads instead of larger teams (i.e., teams comprised of 3 or more members) because a dyad represents the smallest team size suitable for studying expertise recognition (Littlepage and Silbiger, 1992), knowledge coordination (Littlepage and Silbiger, 1992), CPS (Xu et al., 2023), and TMS development (Wegner, 1986). Dyadic teams facilitate more focused and intimate interactions between two members, making communication dynamics easier to observe and analyze within a shorter timeframe. In contrast, larger teams introduce greater complexity with increased interpersonal dynamics, varied communication patterns, and potential role overlaps. Studying dyads allows researchers to isolate and examine specific collaboration aspects effectively, without the complexities of larger teams (Hollingshead and Poole, 2011). Insights gained from studying dyads can then inform understanding of collaboration mechanisms in more complex settings such as multi-person teams and organizations (Wegner et al., 1985).
Participants to form the dyadic teams were recruited through Amazon Mechanical Turk. One thousand individuals were originally invited to participate in this study. Each participant was asked to complete a general science knowledge test, the Ten Item Personality Inventory (TIPI) survey, and a short survey on the participant’s basic demographic background. Then, the participants were randomly divided into 500 dyadic teams to complete an online simulation-based collaborative task. After removing those teams and individuals with incomplete responses to the general science knowledge test, personality survey, or the online collaborative task, we obtained a final dataset of 377 dyadic teams consisting of 754 individuals. Each of these individuals participated in one and only one team. Among them, 366 (49%) identified themselves as male, and the rest 388 (51%) identified themselves as female. The majority (77%) of the participants were white, 8% were black, 7% were Hispanic, 6% were Asian, and the rest 2% were American Indian or Pacific Islander. Regarding the language background, 98% of the participants were native English speakers, and 2% were non-native English speakers. The self-reported age of the participants ranged from 18 to 68, with 70% between 18 and 35, 29% between 35 and 60, and 1% older than 60. All participants indicated that they had at least 2 years of college education per our recruitment requirement.
The virtual CPS task assigned to each team involved an online simulation-based task in the domain of general science (a sample screenshot is shown in Fig. 1). Similar to the study by Littlepage et al. (2008), this research designed a eureka-type intellective task, where dyadic teams were instructed to provide uncontroversial, correct answers to thought-provoking questions. However, unlike previous studies, this task was conducted entirely virtually in our study. In this simulation, virtual collaboration between dyadic team members was exclusively conducted through online synchronous text chat. Team members chatted with each other to complete a general science knowledge quiz focused on volcanoes, answering a series of seven questions. This method of online, text-based collaboration has been utilized in prior research involving simulated problem-solving (Kanawattanachai and Yoo, 2007). Compared to communication modes with greater media richness and social presence, such as videoconferencing, text-based communication can reduce the pressure and distractions stemming from social and non-verbal cues, enabling participants to concentrate more on task-related content and discussions (Oviedo and Fox Tree, 2021). Moreover, text chat provides a detailed record of conversations, allowing teams to easily revisit prior discussions, track decisions and actions, and monitor progress. This enhances accountability and facilitates knowledge sharing within the team. Additionally, text-based communication requires less bandwidth and technical expertise compared to videoconferencing, making it more accessible and easier to implement for research purposes, especially in laboratory settings (Hollingshead and Poole, 2011). Instead of using pre-existing teams, this study recruited participants, who may be strangers to each other, to form ad hoc teams for the CPS task, which is a common practice, especially in experiments and assessments on CPS skills (Xu et al., 2023). In our task, participants were assigned screen names for the collaborative task and were not encouraged to disclose their real identities to each other. Randomly assembled teams minimize the effects of potential confounding factors and the impacts of differences in previous collaboration experiences on the new task.
Fig. 1 [Images not available. See PDF.]
Screenshot of virtual CPS task.
Measures
Individual expertness
Before beginning the CPS task, each team member underwent a preliminary knowledge assessment by completing a general science knowledge test comprising a composite inventory of 37 multiple-choice questions. These questions were adapted from the Scientific Literacy Measurement (SLiM) instrument (Hao et al., 2017). This method of using a composite inventory to measure team members’ expertness level is a similar approach to how domain-specific expertise was measured in past research (Van der Vegt et al., 2006). Each team member’s test score in this test indicates the level of expertness of this member in the general domain of science. All participants’ test scores ranged from 0 to 36 (mean = 27.40 and median = 29). The reliability of this test measured by Cronbach α was 0.89, which indicated an acceptable level of internal consistency among all question items. The distribution of all participants’ expertness levels, measured by their general science knowledge test scores, is shown in Fig. 2.
Fig. 2 [Images not available. See PDF.]
Distribution of individual expertness levels (general science knowledge test scores) (N = 754).
To classify a team member’s expertness level, we divided each participant’s general science knowledge test score into two categories: High-expertness (H) and Low-expertness (L), based on the median value of all participants’ scores. If a team member’s test score exceeded the sample median of 29, s/he was classified as High-expertness (H); if a team member’s test score equaled or fell below 29, s/he was classified as Low-expertness (L). This dichotomization process resulted in 446 individuals labeled as L and 308 individuals labeled as H in terms of their expertness levels within the general domain of science. Next, for each individual embedded in a dyadic team, s/he can be assigned into one of four possible categories in reference to her/his teammate’s expertness levels: H collaborating with H (labeled as HwH), H collaborating with L (labeled as HwL), L collaborating with H (LwH), and L collaborating with L (labeled as LwL).
Expertness diversity
Previous research has measured expertness diversity by computing the standard deviation (Martins et al., 2013) or variation (Lee et al., 2022) of participants’ expertness scores in a given knowledge domain. However, we contend that this approach might only capture the absolute degree of expertness diversity while overlooking the nuanced distinction between two possible scenarios in teams exhibiting low degrees of expertness diversity: where all team members possess either high or low levels of expertness. Therefore, we decided to treat expertness diversity as a categorical variable rather than a continuous variable. In this way, we are able to discern whether teams with low expertness diversity consist of members with predominantly high or low expertness levels.
In our study, when a dyadic team was composed of a member with H level of expertness and a member with L level of expertness, this team was coded to have a high degree of expertness diversity. When a team consisted of two members who both had H levels of expertness or L levels of expertness, this team was coded to have a low degree of expertness diversity. Among the 377 teams, 134 teams had low degrees of expertness diversity with both members being low-expertness (labeled as LL), 65 teams had low degrees of expertness diversity with both members being high-expertness (labeled as HH), and 178 teams had high degrees of expertness diversity (labeled as LH) comprising a high-expertness and a low-expertness member. The distribution of team members’ expertness levels and their expertness diversity within the team are summarized in Table 1.
Table 1. Distribution of participants and team diversity based on expertness and personality traits.
L Individual | H Individual | LL Team | LH Team | HH Team | |
Total N = 754 | Total N = 377 | ||||
Expertness | 446 | 308 | 134 | 178 | 65 |
Openness to experience | 515 | 239 | 177 | 161 | 39 |
Conscientiousness | 460 | 294 | 144 | 172 | 61 |
Extraversion | 417 | 337 | 115 | 187 | 75 |
Agreeableness | 457 | 297 | 144 | 169 | 64 |
Neuroticism | 412 | 342 | 120 | 172 | 85 |
Big Five personality traits and personality traits diversity
To measure each participant’s Big Five personality traits, we used the Ten Item Personality Inventory (TIPI) test (Gosling et al., 2003). This test included 10 question items on a 5-point Likert scale (ranging from 0—Extremely disagree to 5—Extremely agree). The TIPI test assessed each category of the Big Five personality traits using two opposite items: one standard item and one reverse-scored item. The final measurement of each personality trait was the mean of the standard item and the reverse-scored item after recoding.
The distribution of all participants’ Big Five personality traits scores is shown in Fig. 3. The highest frequencies of the responses for four of the Big Five personality traits (except for “extraversion”) were around 4 (“Agree”), which was also the median of these four traits. For extraversion, the median was 2.5 (between “Disagree” and “Neither agree nor disagree”). Similar to the measurement of individual expertness, for each of the five personality traits, we used the median of all participants’ test scores to classify each team member as either High (H) or Low (L) in that particular personality trait category. Next, each individual was categorized into one of four possible compositions based on the individual’s and her/his teammate’s personality traits: High collaborating with High (labeled as HwH), High collaborating with Low (labeled as HwL), Low collaborating with High (LwH), and Low collaborating with Low (labeled as LwL).
Fig. 3 [Images not available. See PDF.]
Distribution of Big Five personality traits (N = 754).
Within each Big Five personality trait (Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism), dyadic teams with one member scoring H and the other L were coded as having high personality diversity for that specific trait. Conversely, teams, where both members scored either H or L were coded as having low personality diversity for that trait. The distribution of individual participants’ Big Five personality traits and the diversity of teams’ Big Five personality traits are summarized in Table 1.
Individual and team performance gain from virtual collaboration
We implemented a computerized system to measure the performance change resulting from virtual collaboration (i.e. the difference between task performance before and after virtual collaboration). First, the system asked each team member to answer the seven science task questions individually and graded their individual responses against correct answers. These graded individual responses were denoted as the initial score by the system but were not made available to the participants. Then, the system instructed team members to collaborate with each other by sharing and discussing their answers through the online text chat. Each team was asked to decide on a final response after their virtual collaboration. Subsequently, each team member was asked to submit a revised response individually if s/he wanted to change her/his initial response (graded and denoted as the revised score by the system). Finally, one of the team members was randomly chosen by the system to submit the team answer upon which both members agreed (graded and denoted as the team representative score by the system). In this study, we focused on the initial and revised scores because we observed that most of the team representative scores were identical to the team members’ revised scores. Based on the initial and revised scores for each question, we developed the following variables to measure individual- and team-level performance gain from virtual collaboration, which were used as dependent variables in the subsequent data analysis.
At the team level, we first computed the team's initial score by summing up the initial scores of all question items responded to by both team members in a team prior to virtual collaboration. Then, we computed the team score change by subtracting the sum of the team's initial scores from the sum of the revised scores of all question items. A positive value of the team score change indicated the team level performance gain from virtual collaboration, whereas a negative value of the team score change indicated the team level performance loss as a result of virtual collaboration.
At the individual level, we first computed the individual initial score by summing up a team member’s initial scores of all question items prior to virtual collaboration. Then we computed the individual score change by subtracting an individual member’s initial scores of all questions from this member’s revised scores. A positive value of the individual score change indicated the individual performance gain from virtual collaboration, whereas a negative value of the individual score change indicated the individual performance loss as a result of virtual collaboration.
Analysis
Both RQ1 and RQ3 inquire how team-level characteristics, namely the team’s expertness diversity (RQ1) and Big Five personality traits diversity (RQ2), could influence team performance gain from virtual collaboration. To address these two questions concurrently, we ran ANCOVAs with the team score change as the dependent variable. The independent variables were the team expertness diversity and Big Five personality traits diversity. We also added the team's initial score as a covariate variable to control for the effect of each team’s initial performance. For the variable that was found to be statically significant in the ANCOVA results, we ran the post-hoc Tukey’s Test to assess the statistical differences among the three subgroups (i.e. HH, LH, and LL) regarding the team performance gain from virtual collaboration.
RQ2 investigates how a team member’s performance gain from virtual collaboration can be influenced by one’s own expertness level and that of the teammate. Likewise, RQ4 explores how a team member’s performance gain from virtual collaboration can be affected by one’s own Big Five personality traits and those of the teammate. To answer these two questions concurrently, we ran ANCOVAs with the individual score change as the dependent variable. The independent variables included an individual member’s expertness and Big Five personality traits. We also added the individual initial score as a covariate variable to control for the effect of an individual member’s initial performance. For the variable that was found to be statistically significant in the ANCOVA results, we ran the post-hoc Tukey’s test to further test the differences among four different subgroups (i.e. HwH, HwL, LwH, and LwL) regarding the individual’s performance gain from virtual collaboration.
Results
Team performance gain from virtual collaboration
Table 2 summarizes the ANCOVA results to address RQ1 and RQ3. After controlling for the effect of the team's initial score (p < 0.05), the team expertness diversity and the agreeableness personality trait diversity remained significant predictors of the team score change. Further, the post-hoc Tukey’s test results (presented in Fig. 4) showed that teams with a high degree of expertness diversity (labeled as LH) had significantly greater score changes than teams with low degrees of expertness diversity (both HH and LL teams). In addition, teams composed of two members who both had low agreeableness traits (LL in agreeableness) had the greatest score changes compared with the other two types of teams (HH and LH in agreeableness).
Table 2. ANCOVA results with team score change as the dependent variable.
Independent variable | Sum of squares | df | Mean square | F | p |
|---|---|---|---|---|---|
Team initial score | 5.60 | 1 | 5.60 | 4.15 | 0.04 |
Expertness diversity | 11.56 | 2 | 5.78 | 4.29 | 0.01 |
Openness to experience diversity | 3.82 | 2 | 1.91 | 1.42 | 0.24 |
Conscientiousness diversity | 1.69 | 2 | 0.84 | 0.63 | 0.54 |
Extraversion diversity | 5.45 | 2 | 2.73 | 2.02 | 0.13 |
Agreeableness diversity | 10.49 | 2 | 5.24 | 3.89 | 0.02 |
Neuroticism diversity | 0.12 | 2 | 0.06 | 0.05 | 0.96 |
Residuals | 489.65 | 363 | 1.35 |
Fig. 4 [Images not available. See PDF.]
Team-level comparison based on post-hoc Tukey’s test results.
a Expertness diversity and score changes; (b) Agreeableness diversity and score changes.
Individual performance gain from virtual collaboration
Table 3 summarizes the ANCOVA results to address RQ2 and RQ4. After controlling for the effect of the individual initial score, the individual team member’s expertness level remained the significant predictor of the individual score change. Further, the post-hoc Tukey’s test results (presented in Fig. 5) showed that low-expertness individuals had the greatest score changes when collaborating with a high-expertness teammate (labeled as Lw/H). However, the pairwise differences in individual score changes among the other three groups (Hw/L, Lw/L, and Hw/H) were not significant. The agreeableness personality trait was significant in the full model, but all subgroup comparisons were not significant in the post-hoc Tukey’s test and thus were not reported nor included for further discussion.
Table 3. ANCOVA results with individual score change as the dependent variable.
Independent variable | Sum of squares | df | Mean square | F | p |
|---|---|---|---|---|---|
Individual initial score | 115.63 | 1 | 115.63 | 183.89 | <0.001 |
Expertness | 11.91 | 3 | 3.97 | 6.32 | <0.001 |
Openness to experience | 2.60 | 3 | 0.87 | 1.38 | 0.25 |
Conscientiousness | 0.59 | 3 | 0.20 | 0.32 | 0.81 |
Extraversion | 2.42 | 3 | 0.81 | 1.27 | 0.28 |
Agreeableness | 5.15 | 3 | 1.72 | 2.73 | 0.04 |
Neuroticism | 0.31 | 3 | 0.10 | 0.17 | 0.92 |
Residuals | 461.56 | 734 | 0.63 |
Fig. 5 [Images not available. See PDF.]
Individual-level comparison based on post-hoc Tukey’s test results.
Discussion
This study applies and integrates TMS theory and the Big Five personality traits model to uncover the potential benefits of virtual collaboration within dyadic teams. At the team level, it investigates the impact of a team’s expertness diversity and diversity in Big Five personality traits (extraversion, agreeableness, openness, conscientiousness, and neuroticism) on the team’s performance gained from virtual collaboration. At the individual level, the study examines how a team member’s performance gain from virtual collaboration is affected by one’s own expertness level and Big Five personality traits as well as those of the teammate. This research conducted an experimental study, analyzing 377 dyadic teams comprising 754 individuals engaged in a virtual CPS activity, namely an online simulation-based collaborative task. The results revealed that dyadic teams characterized by high expertness diversity, encompassing members with high and low levels of expertness, exhibit greater performance gains from virtual collaboration compared to teams with low expertness diversity, where both members had either low or high levels of expertness. Moreover, teams with low diversity in agreeableness, consisting of members who both had low levels of agreeableness, experienced the most substantial performance improvement from virtual collaboration, surpassing teams whose members both had high levels of agreeableness and teams with high diversity in agreeableness. Finally, at the individual level, a team member with a lower level of expertness who collaborated with a high-expertness partner demonstrated the most significant performance enhancement in virtual collaboration.
Theoretical implications
Through the lens of differentiated and integrated TMS structures, traditional TMS scholarship primarily examines how team expertise is dispersed across different knowledge domains and how such inter-domain expertise diversity affects team performance (Gupta, 2012; Gupta and Hollingshead, 2010; Iannone et al., 2017). However, the distribution of expertise within the same knowledge domain and the impact of intra-domain expertise diversity on team performance remain largely unexplored. Adopting the expertness diversity perspective (Van der Vegt et al., 2006), this study extends TMS research by examining how varying degrees of expertness diversity in a given knowledge domain could influence team performance change in virtual CPS settings. Contrary to previous research suggesting a negative correlation between expertness diversity and team performance (Lee et al., 2022), our findings reveal the positive effects of expertness diversity on team performance enhancement in virtual CPS environments. Moreover, addressing previous concerns that high-expertness team members might be unwilling to assist those with lower expertness levels (Van der Vegt et al., 2006), our study finds that pairing low-expertness members with high-expertness counterparts results in significant performance improvements in virtual collaboration. This finding suggests that high-expertness members may indeed be inclined to offer their expertise to help those with lower levels of knowledge when collaborating with each other to solve intellective tasks. Our study also extends prior work, which predominantly involved student participants who had existing collaborative relationships (Martins et al., 2013; Van der Vegt et al., 2006), by demonstrating that expertness diversity can enhance team performance in ad hoc CPS teams where members have not previously collaborated.
Although our research did not focus on the direct effects of differentiated and integrated TMS structures on virtual CPS performance improvement, prior studies on related topics provide valuable insights to support our findings. Previous TMS research has emphasized that both differentiated and integrated TMS structures can enhance team collaboration, albeit through distinct mechanisms and in varying task settings. For instance, teams operating with an integrated TMS structure tend to outperform those with a differentiated TMS when engaging in intellective tasks, whereas teams with a differentiated TMS structure tend to leverage the unique information possessed by individual members more effectively (Gupta and Hollingshead, 2010). Additionally, a differentiated TMS may facilitate centralized information-seeking, particularly when specialized expertise is required (Yan et al., 2021). These studies provide a relevant rationale for understanding the findings observed in our research. Indeed, all teams participating in the current study were required to complete an intellective CPS task, specifically to identify a single, correct answer to a series of quiz questions focused on a particular domain of science. Consequently, the differentiation in expertness levels between team members necessitated the low-expertness member’s active pursuit and utilization of the expertise held by the high-expertness member. In turn, the high-expertness member has the opportunity to solidify and refine her/his own understanding of the problem by explaining concepts and correcting misconceptions during the discussion and decision-making process. This reciprocal learning dynamic (Jokisch et al., 2020) likely contributed to the overall significant performance gains observed in their virtual collaboration.
As evidenced in our experimental design and analysis results, when team members collaborated with each other on a virtual platform without in-person, face-to-face interactions, they still engaged in expertise coordination and information exchange to achieve collaborative outcomes. This finding resonates with social information processing theory (Walther, 1992) that suggests individuals are able and motivated to compensate for the loss or reduction of non-verbal cues in virtual CPS settings by exchanging and processing remaining information cues such as language content and style characteristics. As shown in this study, although team members were deprived of rich communication channels such as face-to-face and audio/video-based communication (Fonner and Roloff, 2012), team members exhibited adeptness in adjusting their relational behaviors to effectively utilize textual cues to accomplish their collaborative tasks.
Finally, this research contributes to the Big Five personality scholarship by offering empirical evidence of the linkage between the diverse composition (or lack thereof) of the Big Five personality traits of team members and their performance gains from virtual collaboration. Echoing prior research findings that the agreeableness personality trait (Barrick and Mount, 1993) and the diversity of team agreeableness (Mohammed and Angell, 2003) could hinder team performance, our study revealed that teams whose members were both low in agreeableness benefited the most from virtual collaboration compared with teams whose members were both highly agreeable and teams whose members had diverse agreeableness levels. This finding may be explained by the prediction that low agreeableness can be associated with assertiveness and a strong desire to achieve goals (Fang et al., 2015). In a virtual CPS setting where social cues are limited, low agreeableness may lead to a more focused and efficient work style, with less time spent on pleasantries or social niceties. In addition, lower agreeableness can lead to more direct communication, which can be crucial in virtual settings to avoid misunderstandings or delays due to unclear communication (Huang et al., 2014). Moreover, when team members are uniformly and highly agreeable, they are less likely to confront and challenge each other. Also, in teams where members have diverse levels of agreeableness, although the less agreeable ones may take aggressive or argumentative actions, the more agreeable ones would strive for harmony by staying silent or compliant. Both situations can foster groupthink (McCauley, 1998), a phenomenon characterized by the suppression of critical inquiries and dissenting viewpoints. Consequently, there are limited opportunities for the exchange of creative ideas and critical thinking, which are essential for collaborative problem-solving (Hao et al., 2017).
Practical implications
Besides demonstrating the value of using both TMS theory and Big Five personality traits models to understand virtual team dynamics, this study also offers practical guidance for improving team performance and informing managerial practices. One of the primary implications of this research is that expertness diversity (i.e., intra-domain expertise differentiation) is equally advantageous for team performance improvement as expertise diversity (i.e., inter-domain expertise differentiation). Hence, for those organizations and teams that rely on a single or limited knowledge domain for CPS, they can and should harness the benefits of expertness diversity. When forming virtual teams to complete intellective collaborative tasks, the management should strategically compose teams by pairing individuals with high and low levels of expertness. For example, in a project team developing an AI application, a Senior AI Engineer leads the design of complex algorithms and system architecture while mentoring a Junior AI Developer. The Junior Developer assists with coding and testing, freeing the Senior Engineer to tackle more advanced problems. This dynamic benefits both: the Junior Developer gains experience, and the Senior Engineer reinforces her/his knowledge through mentorship. This mix of high and low expertness ensures efficient resource use, innovation, and project success.
Aligned with recommendations from Van der Vegt et al. (2006), we encourage organizations to establish structured mentorship programs in which high-expertness members are assigned as formal mentors to less-expertness members. Organizations should provide incentives for high-expertness mentors to actively share their expertise and offer guidance to mentees. This could involve recognition programs, training opportunities, or other forms of appreciation for their contribution to team development. Furthermore, this research suggests that teams with low agreeableness may benefit more from virtual collaboration. This highlights the potential advantages of strategically assigning less agreeable members to work in the virtual team, especially where groupthink is present and critical thinking is desired. However, the management should be mindful that individuals scoring low in agreeableness may be more prone to conflicts or disagreements. Thus, organization and team leaders should proactively implement conflict resolution strategies and foster an environment that encourages open communication and constructive feedback among team members. This approach can help alleviate potential conflicts among less agreeable members and bridge potential divides between highly agreeable and less agreeable colleagues.
Limitations and future directions
This research has several limitations that should be addressed in future research. First, the experimental design took place in a highly constrained context (i.e., ad hoc teams with two randomly assigned team members who communicated through text chat for a duration of ~50 min to complete an online simulation-based task). The dyadic composition of the team, text-only communication, short duration of the collaboration process, and simulation of the collaborative task are all very likely to limit the generalizability of this study to understanding how real-world teams work. For example, the effects of TMS development (such as expertise recognition and knowledge retrieval) and the Big Five personality traits might not have had sufficient time to unfold during the short span of virtual collaboration. Future research should examine more natural and mature collaboration settings for an extensive period of time to fully uncover the short- and long-term impacts of TMS development and personality traits on collaborative performance. Additionally, further studies are needed to go beyond dyadic teams and explore how these dynamics play out and influence CPS outcomes in larger, multi-person teams.
Second, this research measured task performance by grading participants’ responses to seven multiple-choice questions only. The small number and limited type of questions rendered less room for performance improvement among the teams participating in our study. Therefore, future research should develop more sophisticated and multidimensional instruments to assess individual and team performance with greater accuracy and the ability to capture subtle variations.
Third, this study focuses on the effects of two prominent individual attributes, namely expertness and personality traits, on virtual CPS outcomes. However, we excluded other personal attributes, such as demographic data, from our analysis. While prior research showed that TMS processes (particularly intra-team coordination) had a greater impact on team effectiveness than demographics such as gender and social-economic status (Michinov et al., 2008), other studies highlighted the importance of team members’ demographic properties, such as gender composition (Iannone et al., 2017), cultural backgrounds (Yoon and Hollingshead, 2010), and demographic similarity (e.g., racial and gender homophily) (Keith et al., 2017), on TMS development and team performance. Therefore, future research should continue to explore the influence of a wider range of personal attributes, including demographic, personality, psychological, and behavioral factors, on TMS development and work performance in virtual CPS settings.
Finally, the current study does not include the communication content (text-chat) data in the analysis. Though discourse analysis has been previously deployed to provide substantive insights into team members’ collaborative behaviors (Andrews et al., 2017), more research is needed to fully delineate how TMS development, expertness diversity, and personality traits influence collaborative outcomes by examining the content generated and shared throughout the collaboration processes.
Conclusion
The landscape of collaboration among team members has undergone significant transformation in recent years, driven largely by the global pandemic. This shift has prompted scholars to emphasize the importance of investigating the factors that influence virtual collaboration, spanning from its antecedents to its outcomes, across both experimental and real-world work environments (Jiang et al., 2023; O’Bryan et al., 2022). As evidenced in the present study, the interplay of team members’ expertness and Big Five personality traits plays a crucial role in shaping collaborative outcomes. Moving forward, enhancing our understanding of individual and team performance in CPS settings will require the integration of diverse theoretical frameworks and the adoption of innovative methodological approaches.
Acknowledgements
This work was supported in part by the National Key Research and Development Program of China (2021YFF0901600), the National Natural Science Foundation of China (62177044), and the USTC Research Funds of the Double First-Class Initiative (FSSF-A-240110).
Author contributions
MZ, CS, JH, LL, PK, and AvD conceived and designed the study. CS and MZ developed the theoretical framework of the study. MZ, JH, and LL ran the experiments and prepared the dataset. MZ analyzed the data. All discussed results and contributed to the manuscript. MZ and CS wrote the manuscript.
Data availability
The datasets generated during and/or analyzed during the current study are not publicly available because when participants signed the informed consent, they specifically agreed to the term that their responses will not be made publicly available.
Competing interests
The authors declare no competing interests.
Ethical approval
All procedures performed in this research involving human participants were in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. This study was approved by the Committee for Prior Review of Research (CPRR) at the Educational Testing Service (ETS) with the approval number 2014-05-05T090401.
Informed consent
Informed consent was obtained from all participants.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
Akgün, AE; Byrne, J; Keskin, H; Lynn, GS; Imamoglu, SZ. Knowledge networks in new product development projects: a transactive memory perspective. Inf Manag; 2005; 42,
Andrews, JJ; Kerr, D; Mislevy, RJ; von Davier, A; Hao, J; Liu, L. Modeling collaborative interaction patterns in a simulation‐based task. J Educ Meas; 2017; 54,
Aritz, J; Walker, R; Cardon, PW. Media use in virtual teams of varying levels of coordination. Bus Prof Commun Q; 2018; 81,
Austin, JR. Transactive memory in organizational groups: the effects of content, consensus, specialization, and accuracy on group performance. J Appl Psychol; 2003; 88,
Barrick, MR; Mount, MK. Autonomy as a moderator of the relationships between the Big Five personality dimensions and job performance. J Appl Psychol; 1993; 78,
Breuer, C; Huffmeier, J; Hertel, G. Does trust matter more in virtual teams? A meta-analysis of trust and team effectiveness considering virtuality and documentation as moderators. J Appl Psychol; 2016; 101,
Choi, OK; Cho, E. The mechanism of trust affecting collaboration in virtual teams and the moderating roles of the culture of autonomy and task complexity. Comput Hum Behav; 2019; 91, pp. 305-315. [DOI: https://dx.doi.org/10.1016/j.chb.2018.09.032]
Connelly, CE; Černe, M; Dysvik, A; Škerlavaj, M. Understanding knowledge hiding in organizations. J Organ Behav; 2019; 40,
Di Gangi, PM; Wasko, MM; Tang, X. Would you share? Examining knowledge type and communication channel for knowledge sharing within and across the organizational boundary. Int J Knowl Manag; 2012; 8,
Fang, R; Landis, B; Zhang, Z; Anderson, MH; Shaw, JD; Kilduff, M. Integrating personality and social networks: a meta-analysis of personality, network position, and work outcomes in organizations. Organ Sci; 2015; 26,
Fiore S, Graesser A, Greiff S, Griffin P, Gong B, Kyllonen P, Massey C, O’neil H, Pellegrino J, Rothman R, Soulé H, von Davier A (2017) Collaborative problem solving: considerations for the national assessment of educational progress. National Center for Education Statistics
Fonner, KL; Roloff, ME. Testing the connectivity paradox: linking teleworkers’ communication media use to social presence, stress from interruptions, and organizational identification. Commun Monogr; 2012; 79,
Goldberg, LR. An alternative “description of personality”: the Big-Five factor structure. J. Personal. Soc. Psychol.; 1990; 59,
Gosling, SD; Rentfrow, PJ; Swann, WB. A very brief measure of the Big-Five personality domains. J Res Personal; 2003; 37,
Gupta, N. Team responses to noncontributing members: the effects of attribution and knowledge overlap. Group Dyn-Theory Res Pract; 2012; 16,
Gupta, N; Hollingshead, AB. Differentiated versus integrated transactive memory effectiveness: It depends on the task. Group Dyn; 2010; 14,
Haas, MR; Hansen, MT. Different knowledge, different benefits: toward a productivity perspective on knowledge sharing in organizations. Strateg Manag J; 2007; 28,
Hao J, Liu L, von Davier AA, Kyllonen PC (2017) Initial steps towards a standardized assessment for collaborative problem solving (cps): Practical challenges and strategies. In: von Davier AA, Zhu M, Kyllonen PC (eds) Innovative assessment of collaboration. Springer International Publishing, pp. 135–156
Hernaus, T; Cerne, M; Connelly, C; Poloski Vokic, N; Škerlavaj, M. Evasive knowledge hiding in academia: when competitive individuals are asked to collaborate. J Knowl Manag; 2019; 23,
Hollingshead, A. Communication, learning, and retrieval in transactive memory systems. J Exp Soc Psychol; 1998; 34,
Hollingshead, A. Retrieval processes in transactive memory systems. J. Personal Soc Psychol; 1998; 74,
Hollingshead, A. Perceptions of expertise and transactive memory in work relationships. Group Process Intergroup Relat; 2000; 3,
Hollingshead, A. Cognitive interdependence and convergent expectations in transactive memory. J Personal Soc Psychol; 2001; 81,
Hollingshead A, Poole MS (2011) Research methods for studying groups and teams: a guide to approaches, tools, and technologies. Routledge
Hsu, JS-C; Shih, S-P; Chiang, JC; Liu, JY-C. The impact of transactive memory systems on IS development teams’ coordination, communication, and performance. Int J Proj Manag; 2012; 30,
Huang, JL; Ryan, AM; Zabel, KL; Palmer, A. Personality and adaptive performance at work: a meta-analytic investigation. J Appl Psychol; 2014; 99,
Iannone, NE; McCarty, MK; Kelly, JR. With a little help from your friend: transactive memory in best friendships. J Soc Personal Relatsh; 2017; 34,
Jackson, M; Moreland, RL. Transactive memory in the classroom. Small Group Res; 2009; 40,
Janis, I. Groupthink. Psychol Today; 1971; 5,
Jiang Y, Martín-Raugh M, Yang Z, Hao J, Liu L, Kyllonen P (2023) Do you know your partner’s personality through virtual collaboration or negotiation? Investigating perceptions of personality and their impacts on performance. Comput Hum Behav 141. https://doi.org/10.1016/j.chb.2022.107608
Jokisch, MR; Schmidt, LI; Doh, M; Marquard, M; Wahl, H-W. The role of Internet self-efficacy, innovativeness and technology avoidance in breadth of internet use: comparing older technology experts and non-experts. Comput Hum Behav; 2020; 111, [DOI: https://dx.doi.org/10.1016/j.chb.2020.106408]
Judge, TA; Bono, JE; Ilies, R; Gerhardt, MW. Personality and leadership: a qualitative and quantitative review. J Appl Psychol; 2002; 87,
Kanawattanachai, P; Yoo, Y. Dynamic nature of trust in virtual teams. J Strateg Inf Syst; 2002; 11, pp. 187-213. [DOI: https://dx.doi.org/10.1016/S0963-8687(02)00019-7]
Kanawattanachai, P; Yoo, Y. The impact of knowledge coordination on virtual team performance over time. MIS Q; 2007; 31,
Karl, KA; Peluchette, JV; Aghakhani, N. Virtual work meetings during the Covid-19 pandemic: the good, bad, and ugly. Small Group Res; 2022; 53,
Keith, M; Demirkan, H; Goul, M. The role of task uncertainty in IT project team advice networks. Decis Sci; 2017; 48,
Lee, E; Chung, W; Hong, W. Task conflict and team performance: roles of expertise disparity and functional background diversity. Int J Confl Manag; 2022; 33,
Lewis, K. Measuring transactive memory systems in the field: scale development and validation. J Appl Psychol; 2003; 88,
Lewis, K; Herndon, B. Transactive memory systems: current issues and future research directions. Organ Sci; 2011; 22,
Liang, D; Moreland, R; Argote, L. Group versus individual training and group performance: the mediating role of transactive memory. Personal Soc Psychol Bull; 1995; 21,
Linton, RK; Critch, S; Kehoe, EJ. Role-specific versus cross-role preparation for decision-making teams. Group Dyn-Theory Res Pract; 2018; 22,
Littlepage, GE; Hollingshead, AB; Drake, LR; Littlepage, AM. Transactive memory and performance in work groups: specificity, communication, ability differences, and work allocation. Group Dyn; 2008; 12,
Littlepage, GE; Mueller, AL. Recognition and utilization of expertise in problem-solving groups: expert characteristics and behavior. Group Dyn: Theory Res Pract; 1997; 1,
Littlepage, GE; Silbiger, H. Recognition of expertise in decision-making groups: effects of group size and participation patterns. Small Group Res; 1992; 23,
Luebstorf, S; Allen, JA; Eden, E; Kramer, WS; Reiter-Palmon, R; Lehmann-Willenbrock, N. Digging into “Zoom Fatigue”: a qualitative exploration of remote work challenges and virtual meeting stressors. Merits; 2023; 3,
Martins, L; Schilpzand, M; Kirkman, B; Ivanaj, S; Ivanaj, V. A contingency view of the effects of cognitive diversity on team performance: the moderating roles of team psychological safety and relationship conflict. Small Group Res; 2013; 44,
Mathieu, JE; Maynard, MT; Rapp, TL; Gilson, LL. Team effectiveness 1997–2007: a review of recent advancements and a glimpse into the future. J Manag; 2008; 34, pp. 410-476.
Maurer, M; Bach, N; Oertel, S. Forced to go virtual. Working-from-home arrangements and their effect on team communication during COVID-19 lockdown. Ger J Hum Resour Manag-Z Personalforsch; 2022; 36,
McCauley, C. Group dynamics in Janis’s theory of groupthink: backward and forward. Organ Behav Hum Decis Process; 1998; 73,
McCord, MA; Joseph, DL; Grijalva, E. Blinded by the light: the dark side of traditionally desirable personality traits. Ind Organ Psychol; 2014; 7,
McCrae RR, Costa PT (2008) The five-factor theory of personality. In: John OP, Robins RW, Pervin LA (eds) Handbook of personality: theory and research, 3rd ed., The Guilford Press, pp 159–181
Michinov, E; Olivier-Chiron, E; Rusch, E; Chiron, B. Influence of transactive memory on perceived performance, job satisfaction and identification in anaesthesia teams. Br J Anaesth; 2008; 100,
Mohammed, S; Angell, LC. Personality heterogeneity in teams: which differences make a difference for team performance?. Small Group Res; 2003; 34,
Mohammed, S; Nadkarni, S. Are we all on the same temporal page? The moderating effects of temporal team cognition on the polychronicity diversity-team performance relationship. J Appl Psychol; 2014; 99,
Moreland, R; Myaskovsky, L. Exploring the performance benefits of group training: transactive memory or improved communication?. Organ Behav Hum Decis Process; 2000; 82,
Navick N, Gibbs J (2023) Exhausting work–life challenges through boundary management: an investigation of work–life boundary management among college students during remote work and COVID-19. Inf Commun Soc. https://doi.org/10.1080/1369118X.2023.2247049
Neff, JJ; Fulk, J; Yuan, YC. Not in the mood? Affective state and transactive communication. J Commun; 2014; 64,
O’Bryan L, Oxendahl T, Chen X, McDuff D, Segarra S, Wettergreen M, Beier ME, Sabharwal A (2022) Objective communication patterns associated with team member effectiveness in real-world virtual teams. Hum Factors 17. https://doi.org/10.1177/00187208221147341
Oviedo, VY; Fox Tree, JE. Meeting by text or video-chat: effects on confidence and performance. Comput Hum Behav Rep; 2021; 3, [DOI: https://dx.doi.org/10.1016/j.chbr.2021.100054]
Pearsall, MJ; Ellis, APJ. The effects of critical team member assertiveness on team performance and satisfaction. J Manag; 2006; 32,
Poropat, AE. A meta-analysis of the five-factor model of personality and academic performance. Psychol Bull; 2009; 135,
Pramila-Savukoski, S; Kärnä, R; Kuivila, H; Oikarainen, A; Törmänen, T; Juntunen, J; Järvelä, S; Mikkonen, K. Competence development in collaborative hybrid learning among health sciences students: a quasi-experimental mixed-method study. J Comput Assist Learn; 2023; 39,
Price, KH; Harrison, DA; Gavin, JH. Withholding inputs in team contexts: member composition, interaction processes, evaluation structure, and social loafing. J Appl Psychol; 2006; 91,
Robert, C; Cheung, YH. An examination of the relationship between conscientiousness and group performance on a creative task. J Res Personal; 2010; 44,
Sackett, PR; Walmsley, PT. Which personality attributes are most important in the workplace?. Perspect Psychol Sci; 2014; 9,
Shi, W; Weber, MS. Rethinking the complexity of virtual work and knowledge sharing. J Assoc Inf Sci Technol; 2018; 69,
Shoss, MK; Callison, K; Witt, LA. The effects of other-oriented perfectionism and conscientiousness on helping at work. Appl Psychol; 2015; 64,
Silvester, J; Wyatt, M; Randall, R. Politician personality, Machiavellianism, and political skill as predictors of performance ratings in political roles. J Occup Organ Psychol; 2014; 87,
Sivunen, A; Gibbs, J; Leppäkumpu, J. Managing collapsed boundaries in global work. J Comput-Mediat Commun; 2023; 28,
Soboleva, E; Suvorova, T; Grinshkun, A; Nimatulaev, M. Formation of group creative thinking when working with virtual walls. Eur J Contemp Educ; 2021; 10,
Stadler M, Herborn K, Mustafic M, Greiff S (2019). Computer-based collaborative problem solving in PISA 2015 and the role of personality. J Intell 7(3). https://doi.org/10.3390/jintelligence7030015
Su, C. Who knows who knows what in the group? The effects of communication network centralities, use of digital knowledge repositories and work remoteness on organizational members’ accuracy in expertise recognition. Commun Res; 2012; 39,
Su, C; Huang, M; Contractor, N. Understanding the structures, antecedents and outcomes of organisational learning and knowledge transfer: a multi-theoretical and multilevel network analysis. Eur J Int Manag; 2010; 4,
Tang, F; Mu, J; Thomas, E. Who knows what in NPD teams: Communication context, mode, and task contingencies. J Prod Innov Manag; 2014; 32,
Treem, J; Leonardi, P; van den Hooff, B. Computer-mediated communication in the age of communication visibility. J Comput-Mediat Commun; 2020; 25,
Treem, J; van Zoonen, W; Sivunen, A. Examining communication visibility and social technology platform use in organizations. N Meia Soc; 2024; 26,
Tschan, F; Semmer, N; Gurtner, A; Bizzari, L; Spychiger, M; Breuer, M; Marsch, S. Explicit reasoning, confirmation bias, and illusory transactive memory: a simulation study of group medical decision making. Small Group Res; 2009; 40,
Van der Vegt, G; Bunderson, J; Oosterhof, A. Expertness diversity and interpersonal helping in teams: why those who need the most help end up getting the least. Acad Manag J; 2006; 49,
Walther, JB. Interpersonal effects in computer-mediated interaction: a relational perspective. Commun Res; 1992; 19,
Walther, JB (2011). Theories of computer-mediated communication and interpersonal relations. In: Knapp ML, Daly JA (eds) The handbook of interpersonal communication. Sage, pp. 443–479
Wegner D (1986) Transactive memory: a contemporary analysis of the group mind. In: Mullen B, Goethals GR (eds) Theories of group behavior. Springer-Verlag, pp. 185–208
Wegner, D. A computer network model of human transactive memory. Soc Cogn; 1995; 13,
Wegner D, Giuliano T, Hertel P (1985) Cognitive interdependence in close relationships. In: Ickes WJ (ed) Compatible and incompatible relationships. Springer-Verlag, pp. 253–276
Wilmot, MP; Wanberg, CR; Kammeyer-Mueller, JD; Ones, DS. Extraversion advantages at work: a quantitative review and synthesis of the meta-analytic evidence. J Appl Psychol; 2019; 104,
Xu E, Wang W, Wang Q (2023) The effectiveness of collaborative problem solving in promoting students’ critical thinking: a meta-analysis based on empirical literature. Humanit Soc Sci Commun 10(1). https://doi.org/10.1057/s41599-023-01508-1
Yan, B; Hollingshead, AB; Alexander, KS; Cruz, I; Shaikh, SJ. Communication in transactive memory systems: a review and multidimensional network perspective. Small Group Res; 2021; 52,
Yang, L; Holtz, D; Jaffe, S; Suri, S; Sinha, S; Weston, J; Joyce, C; Shah, N; Sherman, K; Hecht, B; Teevan, J. The effects of remote work on collaboration among information workers. Nat Hum Behav; 2022; 6,
Yin,K, Li D, Zhang X, Dong N, Sheldon O (2023) The influence of the Big Five and Dark Triad personality constructs on knowledge sharing: a meta-analysis. Pers Individ Differ 214. https://doi.org/10.1016/j.paid.2023.112353
Yoon, K; Hollingshead, AB. Cultural stereotyping, convergent expectations, and performance in cross-cultural collaborations. Soc Psychol Personal Sci; 2010; 1,
Yoon, K; Zhu, Y. Social media affordances and transactive memory systems in virtual teams. Manag Commun Q; 2022; 36,
Yuan, Y; Carboni, I; Ehrlich, K. The impact of awareness, social accessibility and media multiplexity on expertise retrieval. J Am Soc Inf Sci Technol; 2010; 61,
Yuan, Y; Fulk, J; Monge, P. Access to information in connective and communal transactive memory systems. Commun Res; 2007; 34,
Yuan, Y; Fulk, J; Monge, P; Contractor, N. Expertise directory development, shared task-interdependence, and strength of communication network ties as multilevel predictors of expertise exchange in transactive memory work group systems. Commun Res; 2010; 37,
Zhang, X; Chen, H; de Pablos, PO; Lytras, MD; Sun, Y. Coordinated implicitly? An empirical study on the role of social media in collaborative learning. Int Rev Res Open Distrib Learn; 2016; 17, pp. 121-144. [DOI: https://dx.doi.org/10.19173/irrodl.v17i6.2622]
Zheng, S; Zeng, X; Zhang, C. The effects of role variety and ability disparity on virtual group performance. J Bus Res; 2016; 69,
Zhong S, Lin Y, Chang C, Liu Y, Lee C, Int Speech Commun Assoc (2019) Predicting group performances using a personality composite-network architecture during collaborative task. National Tsing Hua University, pp. 1676–1680
© The Author(s) 2024. This work is published under http://creativecommons.org/licenses/by-nc-nd/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.