Strengthening perceptions of virtual team cohesiveness and effectiveness in new normal: A hyperpersonal communication theory perspective

Abstract

In the post-pandemic era, many team management challenges arise in multinational organisations along with the pervasion of virtual teamwork collaboration, such as workplace isolation (an unsatisfied need for belongingness). However, little is known about the psychological mechanisms required for solving this context-specific challenge by strengthening virtual team cohesiveness and effectiveness. Accordingly, this study adopts the hyperpersonal communication theory derived from Walther’s hyperpersonal model of computer-mediated communication (CMC) to provide an alternative theoretical explanation to understand better how individuals perceive their team cohesiveness and effectiveness in the specific CMC setting. Supported by 549 valid responses, I illuminate how mutual assistance as a typical interaction among CMC team members cultivates benevolent norms and facilitates members’ perceptions of team cohesiveness and effectiveness. More importantly, this study opens the door for future research on management interventions that can help enhance team cohesiveness and effectiveness at the individual perception level.

Supplementary Information

The online version contains supplementary material available at 10.1057/s41291-023-00225-6.

Introduction

Nowadays, team cohesiveness has become a pivotal factor for the success of multinational organisations and educational institutions in the rapidly changing environment (Cho et al., 2018; Franz et al., 2017; Picazo et al., 2015). Empowering by mature communication technology, organisations are no longer provided with local talents but around the world. Team management targets are expanded from local, physical and mono-cultural teams to global, virtual and multi-cultural teams. While virtual team brings over many benefits to the organisation (see Cincera et al., 2018; Franz et al., 2017; Kagan & Kirchberg, 2016; Kao, 2019), managers are in front of new challenges in virtual team management too, namely workplace isolation (Gao & Sai, 2020). This challenge refers to employees working remotely or at home feeling lonely and isolated from their organisation and the rest of their colleagues (Moss, 2020). Social isolation can make employees anxious (Baumeister & Tice, 1990) and even turn them antisocial (Twenge et al., 2001) because the need to belong is a fundamental need for humans (Baumeister & Leary, 2017). Indeed, the lack of nonverbal cue systems for self-presentation, impression management, interpersonal regard signalisation, and emotion and status expression (Burgoon et al., 2002) is a cardinal issue in the online teamwork context. Supported by ample empirical evidence (Bandara & Senanayaka, 2020; Denstadli et al., 2012), virtual teams are less effective in communication and interpersonal relationships than physical teams, especially in conveying socioemotional warmth and outstanding salience of conversational partners to individual interactants (Short et al., 1976). In other words, workplace isolation is a paramount and inevitable barrier to achieving team cohesiveness and team effectiveness, attracting both practitioners’ and scholars’ attention.

Given that, this paper focuses on the psychological mechanism to resolve the workplace isolation issue. I argued that how employees are satisfied with the need for belongingness in the virtual team can be measured by their perceived team cohesiveness: a member perceiving a higher team cohesion is less likely to have the workplace isolation issue. Accordingly, the present study draws on the hyperpersonal communication theory (Schouten et al., 2007) to answer this crucial research question: what can and how to improve virtual team members' perceived team cohesiveness. I deem that hyperpersonal communication theory is germane and apposite in the virtual teamwork context. This theory partially constitutes the novelty of this paper as it has been little studied in this field. The hyperpersonal communication theory (Walther, 1996) proposes that the level of affection and emotion individuals with computer-mediated communication (CMC) achieve can sometimes outstrip parallel face-to-face communication. Individuals with CMC will selectively and consistently present themselves and behave favourably and desirably, meeting social aspirations (personal impression management) and social expectations (interpersonal relationship management). Specifically, I selected mutual assistance as a representative behaviour of hyperpersonal communication theory. The findings showed that team members would perceive a higher team cohesiveness and effectiveness along with more mutual assistance among members. Further, team members’ perception of team cohesiveness neither significantly impacts the perception of team effectiveness nor moderates the relationship between mutual assistance and perceived team effectiveness in the CMC settings.

This study has twofold contributions. First, this study is among the first to adopt the hyperpersonal communication theory contributing to the team cohesion literature. Contrary to the existing research that focuses on team-level and organisational-level cohesiveness and performance (e.g. Chang & Bordia, 2001; Mach et al., 2010; Whillans et al., 2021), this paper advances our knowledge by shedding light on the perception-level factor (please refer to the “why individual analysis” subsection). This effort is vital in addressing implicit team management issues (such as workplace isolation) when they are not presented and are not easy to discern in the organisational- and team-level data but significantly influence the actual performance of employees and the entire team. Second, the present study also contributes to the mutual assistance literature. Supported by the hyperpersonal communication theory, this study achieves a more nuanced understanding of the boundary effect of mutual assistance so as to reconcile the inconsistent findings about its effect on perceived team effectiveness (Appelbaum et al., 2020; Kozlowski & Ilgen, 2006; Yukelson, 1997). Specifically, this study is uniquely situated in a context-specific CMC teamwork setting in which individuals’ perceptions of team cohesiveness and effectiveness are evaluated. This paper enables us to gain insight into the unexplored grounds by revealing a condition that could strengthen employees’ belongingness and commitment to the team. The contributions of this paper will be further elaborated in the theoretical contribution section in detail.

The present paper is structured as follows. The next section contains an introduction to hyperpersonal communication theory and a discussion of mutual assistance on perceived team cohesiveness and effectiveness. The development of the research hypotheses, methodology and empirical findings come after. Finally, the discussion, implication, limitation, future study and conclusion are presented.

Literature review

Hyperpersonal communication theory

Hyperpersonal communication theory is drawn on the social information processing theory (Walther, 1996), suggesting that people can develop and manipulate their impressions and reinforce desired relationships better and more strategically throughout CMC over traditional face-to-face communications. Specifically, hyperpersonal communication theory extends the assumptions of social information processing theory by repudiating the predominant argument that the absence of nonverbal cues in CMC will engender a comparative reduction in the quality of social interaction (Walther et al., 2015). Instead, hyperpersonal communication theory advocates that the language and nonverbal cues through different communication media and channels are functionally interchangeable. Social information processing theory additionally assumes the different rates of social information exchange to justify the conceptualisation of hyperpersonal communication theory by acknowledging that CMC may require more time to develop the impression and modification of a relationship typical in face-to-face interaction (Walther, 1992).

Hyperpersonal communication theory elaborates how CMC champions the relationship with relatively greater desirability and intimacy from four concurrent dimensions, including “senders,” “receivers,” “channel,” and “feedback.” Firstly, supported by the social identity model of de-individualisation effects theory, Walther et al. (2015) argued that CMC receivers who miss some physical or other cues about the interactants (sender) are more inclined to conjecture the missing information instead of abandoning the impression formation process. Suppose contextual cues about the conversational partners are favourable and share some salient social identity with the receiver under the condition of visual anonymity. In that case, such a conjecture process may exaggerate, becoming a form of “idealisation.” On the other hand, CMC senders can selectively present themselves in desirable ways and contribute to the idealised perceptions receivers develop of them via the cues they need and want to reveal. Besides, hyperpersonal communication theory suggests that CMC is a deliberate channel that facilitates users to “exploit the ability to take time to contemplate and construct messages mindfully” and “create optimally desirable messages without interfering with conversation flow” by discretional editing, recalling, rewriting the messages to the purposive effects (Walther et al., 2015, p.13). More importantly, this idealisation, selective self-representation and channel effects can develop a feedback system to intensify and amplify the “hyperpersonal” effect based on the behavioural confirmation perspective (which underscores the impact of social expectations on people’s behaviour; Snyder et al., 1977). In other words, when CMC users receive a selectively self-presented message, they will idealise them and respond with biased impressions, in turn bolstering the modified personae and motivating CMC senders to behave expectedly in turn.

Why individual-level analysis

The present study aims to answer a research question: what can and how to improve virtual team members' perceived team cohesiveness. This research question is important because, firstly, it is less studied since much contemporary literature assumes a concrete bonding relationship between how people believe and behave (Xiao et al., 2016). For instance, Whillans et al. (2021) collected group-level data to examine the influences pandemic on team cohesiveness and effectiveness. Nevertheless, locus of control theory (Spector, 1982) and cognitive dissonance theory (Festinger, 1962) suggests that perception is not commonly consistent with behaviour. The results on team outcomes do not equate to individual perception. High team cohesion does not mean every team member perceives equivalently high cohesiveness with other team members, nor higher team effectiveness will lead to a higher perception of team effectiveness. Secondly, most scholars attempt to assess group-level constructs by aggregating group means. It will cause a potentially severe problem, namely aggregation bias, as scholars assume individual responses are homogeneous (Gully et al., 1995, 2012). Thirdly, though I agree that group-level analysis is an effective approach to understanding human and group living behaviour, I also value cognition and perceptual process in intergroup relations and socially functional behaviour as well as their irreplaceability, such as social identity (Xiao et al., 2016) as well as goal attainment and survival by informing and shaping action (Clark, 2013). The individual perception-level analysis is essential as it reflects the dynamic integration of the human mind (Cunningham et al., 2007; Freeman & Ambady, 2011; Xiao et al., 2016), providing additional insights for both scientists and practitioners.

Mutual assistance and hyperpersonal communication theory

Mutual assistance in the team refers to the extent that team members are willing to give and receive support to and from team others (Levine, 1988). The related keywords to mutual assistance subsume reciprocity, caregiving/help giving and help receiving (Becker et al., 2003; Bengtson et al., 1996). It is important to clarify that mutual assistance, unlike reciprocity and mutual dependency, refers to an amalgam of an individual’s emotional, psychological and behavioural state of alacrity in giving and receiving support to and from the team members (Levine, 1988). Ikeda and Garces-Ozanne (2019) have interpreted that mutual assistance is based on social capital derived from the trust, norms of reciprocity, and networks. Harada (2012) asserted that regular information sharing and collaborative relationships underlie the formation and maintenance of such tight networks. The joint effort in different fields is tendentious to the contrast between mutual assistance and self-help (Fafchamps & LaFerrara, 2012; Karashima & Ohgai, 2021; Levine, 1988).

It is believed that mutual assistance will be a commonly existing behaviour among team members in CMC settings (Barrett & Murphy, 2021), given the fundamental assumption of hyperpersonal communication theory (i.e. CMC can develop the impression and modification of a relationship commensurate with face-to-face interaction because the language and nonverbal cues through different communication media and channels are functionally interchangeable). Firstly, from the social exchange perspective, employees are more motivated, or at least less reluctant, to assist each other if they want to have close relationships with colleagues and seek their help in the future. Secondly, from the job design perspective, virtual team members are empowered with autonomy, independence and freedom, which are positively associated with employees’ work engagement and mutual assistance (Zhang et al., 2020). Thirdly, from the social interpersonal and instrumental networking perspective, mutual assistance is the most accessible (either actively or passively) manifestation of cordiality and friendliness for virtual team members to manage their impressions and interpersonal relationship in the workplace. Indeed, empirical findings reveal that help seekers will make more polite requests (Duthler, 2006), and the help providers will modestly offer a higher-quality solution (Rains et al., 2019) in CMC settings compared to face-to-face settings. Therefore, mutual assistance is selected as the representative behaviour of hyperpersonal communication theory to explain the variety of employees’ perceptions of team cohesiveness and effectiveness in the following sections.

Mutual assistance and perceived team cohesiveness

Team cohesiveness is the bonding relationship between team members (Ghosh et al., 2019). The higher members’ perceived team cohesiveness, the stronger the bonding is. Some scholars described team cohesiveness as the “therapeutic alliance” (Budman et al., 1993; Marziali et al., 1997) or “sticks together” (Weick, 1976) or “strong ties within the team” (Owen, 1985). Two components constitute perceived team cohesiveness: perceived task cohesiveness and perceived social cohesiveness (Carless & de Paola, 2000; van Vianen & de Dreu, 2001). Perceived task cohesion refers to the extent of shared commitment to the team’s task, representing the goal-oriented nature of cohesion, whereas the perceived social cohesion refers to the extent of positive relationships with other team members, representing the interpersonal nature of the construct (Acton et al., 2020). Internal factors influencing perceived team cohesiveness include members’ similarity, team size, self-disclosure, acceptance, intra-team trust, empathy, respect and linking, and normative conduct (Brewer, 1981; Carron & Spink, 1995; Hogg, 1993; Rempel & Fisher, 1997; Tajfel, 2010). In comparison, the external factors promoting team members’ sense of belonging and identity as intra-team members subsume, based on the uncertainty identity theory (Hogg, 2007) and the positive distinctiveness concept (Hogg et al., 1984), team differentiation, external competition, external threats and team achievements (Ghosh et al., 2019).

Only a few prior studies have investigated the relationship between perceived team cohesiveness and mutual assistance (e.g. Cramton, 2001; Salas et al., 2015). Nonetheless, these studies were either outdated (e.g. conducted forty years ago; Schriesheim, 1980) or consideringly less constructive for understanding team collaboration and establishing team cohesiveness in this new era (Oh et al., 2004; Sparrowe & Liden, 1997). It is essential to clarify the boundary conditions for the relationship between mutual assistance and perceived team cohesiveness regarding the premise and consequence of the effects. For instance, it is intuitive to propose that mutual assistance should positively affect perceived team cohesiveness from the perspective of organisational commitment and affective commitment in the organisational citizenship literature (e.g. Grego-Planer, 2019). Likewise, in the shared leadership literature, members will perceive a higher team cohesion through mutual assistance regarding frequent knowledge sharing, interactions, decision-making participation and pride in collective goal achievement (Imam & Zaheer, 2021). However, it is not always the case that mutual assistance among team members can boost members’ perception of team cohesiveness and effectiveness, such as when help givers offer help and assistance condescendingly and patronisingly. In other words, if the help is not provided favourably and desirably, it may exasperate or annoy help receivers. For instance, help receivers may not appreciate the help and assistance and feel powerless and disappointed in their abilities. Wang et al. (2015) highlighted solicitation as an essential premise for providing help. Even if the help is not explicitly patronising, unsolicited and imposed help can generate a “suspicion of inferiority” and threat to the self-esteem of help recipients (Turner & Pratkanis, 1994). Other adverse consequences related to inappropriate help providing include low group identity (Halabi et al., 2011), negative mood (Bolger & Amarel, 2007), stress (Bolger et al., 2000), sense of inadequacy (Barnes et al., 2008), uncomfortable interdependencies between providers and seekers (Bamberger, 2009), and low self-worth (Lennard & van Dyne, 2018).

Some existing studies have explained the effects and correlations of mutual attraction, mutual knowledge and mutual awareness of belongingness concerning increasing team cohesion (Cramton, 2001; Hogg & Hardie, 1992; Langfred, 1998). For instance, Schriesheim (1980) studied that specific mutual supportive interactions can improve team cohesion. However, the constructs in these studies were still differentiated from mutual assistance, and they are relatively outdated, providing less value to enhance our understanding of the current situation. On the other hand, giving help and assistance means that team members must prioritise team interests over self-interests (LAM et al., 2002). However, this prioritising rationale and psychological mechanism are still unclear. Similarly, scholars agree that perceived team cohesiveness is developed by mutual assistance when a task requires mutual dependency among team members (Huang, 2009). Nevertheless, it merely indicates the willingness to receive help but not necessarily to help one another. Therefore, more alternative explanations from a psychological perspective are demanded in this field. It is critical to understand how mutual assistance is potentially performed as a virtual team practice in context-specific CMC settings and how it establishes the perception of team cohesiveness among team members.

Mutual assistance and perceived team effectiveness

For managerial contribution, the construct of perceived team effectiveness is included in this research design. According to past studies, team effectiveness is mainly influenced by three factors: potency, task interdependency and outcome interdependency (Barrick et al., 2007; Gibson, 1999; Kiggundu, 1981; Sbea & Guzzo, 1987). While potency refers to team members’ belief that their team would be effective, task interdependency means the degree of closeness and intimacy between team members (Lester et al., 2002). Moreover, outcome interdependency connects with the likelihood of the reward provided for high team performance. Since this study investigates the individual level and the reward is not being tested, it focuses on exploring the perception levels of potency and task interdependency alongside the existence of mutual assistance.

Based on previous studies, whether and how mutual assistance can positively influence perceived team effectiveness is an interesting question. It is important to reiterate that perceived team effectiveness is at the individual perception level, significantly different from group-level or team-level effectiveness. In other words, perceived team effectiveness is not outcome oriented but emotion and perception oriented. For instance, a low-quality idea or solution within a team cannot contribute to a favourable consequence (Forgas, 2007; Huang, 2009; Milliken & Martins, 1996), yet the individual may still perceive high team effectiveness because they can come up with a quick solution or effective decision-making process. On the other hand, it is essential to mention that team effectiveness is commonly measured by the team leader, managers or investors with some objective indicators related to key performance. However, the team members' perceived effectiveness is measured with subjective gauges, such as whether employees can speak out their opinions and whether the weekly team meeting is occupied by the boss's bombasts and apple polishers’ adulations.

Past literature has studied the relationship between mutual assistance and perceived team effectiveness from two primary perspectives. Firstly, within the team support model (Anderson & West, 1994), appraisal support is considered a common mutual assistance practice to foster individual team members and aid the team in solving the team problem. Its effect was even postulated to intensify mutual assistance and perceived team effectiveness (Drach‐Zahavy & Somech, 2002): seeing the team problem as one target to solve fortifies the need to provide help to team members within a team. The bond of mutual support and team consensus can increase the perceived effectiveness and quality of work accomplishments (Kozlowski & Ilgen, 2006; Yukelson, 1997). The second perspective centres on the deindividuation effect of social identity theory. The deindividuation phenomenon developed through virtual learning teams facilitates individuals to see themselves as a part of the team other than as individuals (Guegan et al., 2019; Spears & Postmes, 2015), along with the extension of the collaboration period and proximity among team members (Walsh, 2015). When team members treat the team problem as a common target, they have higher alacrity and stronger belief to trust each other and to battle against the issue effectively (Kets De Vries, 2005; Kozlowski & Ilgen, 2006). Hence, it partially explains the effect of mutual assistance on potency. In sum, a more direct and unequivocal illustration is needed to increase our understanding of the relationship between mutual assistance and perceived team effectiveness. The following section will elaborate on how social information processing theory supports the hypotheses.

Hypothesis development

As mentioned in the last section, mutual assistance is theoretically a commonly seen behaviour among team members in the CMC settings based on the assumption of hyperpersonal communication theory. Thus, I propose that mutual assistance in CMC settings is positively associated with perceived team cohesiveness. The increase in mutual assistance between CMC senders and receivers is expected to strengthen the perception of team cohesiveness among team members. The current inconsistency in the literature findings can be mainly attributed to solicitation and job creep issues. Firstly, the core nature of mutual assistance is mutuality, where both help-seeking and providing behaviour should be voluntary and considerate. Any unsolicited and imposed help without asking is not perfectly welcome, but help recipients will feel inferior, vulnerable and shame of themselves instead (Turner & Pratkanis, 1994; Wang et al., 2015). Specifically, such patronising help commonly happens in real team settings where the need for help is explicit and observable from others’ expressions of anxiety and worries and their emotions of frustration, confusion and disappointment. Herewith, out of benevolent intentions or probably lacking sophisticated social skills, help providers may be too eager to offer assistance in either the wrong way or at the wrong time, resulting in a “backfire” effect and reduced perceived team cohesiveness. In contrast, such a phenomenon is less pervasive in CMC settings because the need for help is implicit and needs an explicit request and clarification. For instance, new team members no longer need to be bewildered and embarrassed due to the unfamiliarity with the office printing machine; neither experience the awkwardness of accepting overly “passionate” assistance. Any requests for help can be made through emails that are “carefully edited, formal and linguistically complex” (Herring, 2002, p.115), at least not looking “stupid” or “naïve”. Meanwhile, other team members can provide help more effectively and with higher quality (Rains et al., 2019), and such help is expectedly more desirable, appreciative and beneficial to perceived team cohesiveness.

In addition to the solicitation that focuses on the feeling of the help recipient, the job creep issue, on the other hand, highlights the feeling of help providers. Job creep (also known as role creep, role ambiguity and role overload) occurs when team members feel pressure to continue performing more than their explicit job requirements, such as providing help to other team members over time (van Dyne & Ellis, 2004). In other words, they are afraid others are gradually used to such prosocial helping behaviour and expect them to perform more and more extra-role behaviours in the future. Undoubtedly, helpers need to spend their extra or private time to provide help, such as by working late and coming early, especially for many women workers who are more concerned with work–family conflict (Bolino and Turnley 2005). However, it seems job creep may be a very unlikely case in the CMC settings. First, remote workers are working at home so they can arrange the time on their personal and working plans more flexibly. At the very least, they can save commuting time and spend it with their families. Second, one key differentiation of CMC settings from face-to-face settings is the timeliness of responding. In a face-to-face setting, team members are expected to accept or reject the requests at the moment when other team members seek help. On the one hand, team members may be stressed about rejecting the request because they are coworkers. On the other hand, team members may be reluctant to accept because of the increased workload. However, team members will not have that pressure in the CMC settings: they do not need to respond to the help seekers immediately; they are provided with more autonomy to make decisions. Even if they want to reject the request, they still have more time to compose and edit their rejection to make it look more considerate, respectful and without the threat of damaging interpersonal relationships. Lastly, from the impression management perspective, team members in a CMC setting will be more active in help-providing behaviours but passive in help-seeking behaviours since the latter, to somewhat extent, represents the self-image of incompetence, inferiority and dependence (Lee, 1997; Lennard & van Dyne, 2018). Conversely, team members are exposed to fewer situations that make them feel helpless. For example, they meet each other online. They feel less anxious and stressed presenting in front of all colleagues. They can answer questions more unperturbedly by retrieving the data and information from the computer and showing it on screen.

Additionally, the effect of mutual assistance in CMC settings on improving the perceived team cohesiveness can be illustrated from the social identity perspective. According to social identity theory (Hogg, 2016), when team members share the same characteristics and identity, they will perceive they are in the same “group” and with a higher sense of belonging to that “group”. In CMC settings, team members can behave in a more polite, favourable, desirable, respected, concerned and like-minded way. Likewise, team members have the opportunity and sufficient time to adjust their behaviours consistent with the expectations they presented initially and the expectations from others over time, predisposing other team members to perceive each other as more like-minded, similar, congruent and sharing the same values, characteristics, knowledge and identities. Even though someone misses presenting the desirable cues, others will still imagine and idealise that person as long as the first impression is positive. More importantly, these processes and affinity will be amplified and intensified by increasing interactions and social information/cues exchanges over time (Walther et al., 2015). Eventually, team members will be gratified with other team members and more willing to stay with the team, escalating the team cohesiveness to the optimal extent. In other words, when team members in CMC settings experience more mutual assistance, they will perceive a higher cohesiveness in the team.

Hypothesis 1

In CMC settings, mutual assistance among virtual team members is positively related to members’ perceived team cohesiveness.

Moreover, the relationship between perceived team cohesiveness and perceived team effectiveness is significantly subject to the nature of the task if the task necessitates coordination, communication and mutual supervision among team members (Gully et al., 1995). In other words, if everyone has a clear division of jobs and is responsible for their own tasks, they are given fewer chances to interact with each other. In that case, team members may not have a high perception of team effectiveness even though they are united. Accordingly, perceived task interdependence can determine the effect of perceived team cohesiveness on perceived team effectiveness, and I argue that it is, more importantly, evident under the CMC settings from two perspectives. Firstly, employees in virtual teams can harder acquire and share new information and implicit knowledge across their network, such as organisational and team culture, the leadership style of the manager and supervisor, and the information related to promotion (Yang et al., 2022). Thus, team members rely more on interactions with each other (such as senior team members) to learn extra unwritten information in a way complementary to task interdependences. Given that, mutual assistance is expected to be the “best” and most chosen approach for virtual team members to strengthen their intimacy with others as the interactions starting with tasks-related affairs are most formal, recognised and authorised. With this approach team members can lower others’ vigilance and facilitate trust. This benevolent, competent and upright approach is expected to accelerate information disclosure and sharing (Butler Jr, 1991; Mayer et al., 1995). In short, when team members perceive the team with higher cohesiveness and other team members more trustworthiness, they are more interdependent in information sharing across the network, leading to a higher perception of team effectiveness.

Besides, virtual teams are characterised by higher demands in task interdependence from the other perspectives of labour mobility and mentoring. On the one hand, a virtual team is more diverse and creative as the organisation can recruit talents from around the world. On the other hand, talents are provided with more opportunities and are less restricted to a specific company. Smet et al. (2021) indicated that “The Great Attrition is happening. A record number of employees have or plan to quit…. Voluntary attrition increased by almost 800,000 in the past year.” Yang et al. (2022) found that remote work can result in a more static and soiled collaboration network among team members and fewer bridges between disparate parts. Given that, mentoring and supervising from senior colleagues and peers to newcomers are more pervasive and regular in virtual teams under CMC settings, creating a higher “task interdependence” situation. It is not only because team leaders and other members need to ensure that newcomers can get up to their duties and tasks but also predispose newcomers with a sense of belongingness to the team. Although organisations can indeed reduce the demands of task interdependence by division of work, it seems not a good idea for some intellectual tasks that require complex forms and fields of knowledge and creativity in an era of knowledge economy nowadays. Thus, I argue that virtual teams of diverse backgrounds and globally operated in CMC settings are inherently more task interdependent so that members with higher perceived team cohesiveness will perceive higher team effectiveness.

Hypothesis 2

In CMC settings, members’ perceived team cohesiveness is positively related to members’ perceived team effectiveness.

Furthermore, I expect mutual assistance to foster the perception of team effectiveness among team members in the CMC settings. Firstly, there should be a positive association between mutual assistance and perceived team effectiveness regarding the perceptions of potency and task interdependence. As mentioned above, perceived team potency refers to the extent that team members believe their team is in effect, while perceived task interdependence corresponds to the perceived intimacy and closeness between team members in the cooperation and accomplishment of their tasks during the production phase (Lester et al., 2002). According to van der Vegt et al. (2001), task interdependence can be improved through the frequency of interaction of mutual support. Team members will be more confident to confront any challenges in the workplace as they get support from others, especially receiving a higher-quality solution in CMC settings (Rains et al., 2019).

Hypothesis 3

In CMC settings, mutual assistance among virtual team members is positively related to members’ perceived team effectiveness.

Furthermore, the perceived team cohesiveness is posited to positively moderate the relationship between mutual assistance and perceived team effectiveness. Zaccaro (1991) found that under greater interdependence working on specific tasks, communication increases along with reducing intra-team conflict and enhancing perceived team effectiveness, facilitating individuals’ task accomplishments. When team members perceive a higher level of team cohesiveness with each other, they should have a stronger sense of belongingness to the team, higher trust, empathy, respect and normative conduct toward other team members (Brewer, 1981; Carron & Spink, 1995; Hogg, 1993). Based on the proximity perspective, team members with higher perceived team cohesiveness are more likely to offer help and less reluctant to seek help when needed (Khusanova et al., 2019), causing higher perceived task interdependence and perceived team potency. In contrast, an incohesive and inharmonious team will incur more dissension and friction, preventing the team from effectively completing the tasks (Alsharo et al., 2017; Chung, 2019), such as making decisions ineffectively. Apart from that, compared to face-to-face settings, a virtual team is informationally insular to the outer team due to reduced interaction channels and increased obstacles to communicating with outer team members. So intra-team members are more focused on each other, and they are more willing to spend time, attention, and effort on the affairs of their team and provide help and responses to other team members more immediately. Therefore, when team members perceive a higher level of team cohesiveness, a more frequent mutual assistance practice should precipitate team members to perceive a higher level of team effectiveness.

Hypothesis 4

In CMC settings, members’ perceived team cohesiveness positively moderates the relationship between mutual assistance and perceived team effectiveness.

Methodology

Sampling and survey administration

Online survey is the cardinal method for data collection with the boons of isolating researchers in the study (Bell et al., 2022), and the collection convenience and flexibility are unrestricted by location and time during COVID-19. I select the structural equation modelling (SEM) approach to analyse the hypotheses and the expected moderating effects. Besides, it is more appropriate to use an online survey as a data collection method ontologically, epistemologically and methodologically (Tarhini et al., 2017). This study selects random and snowball methods to send the online questionnaire to young adults in Hong Kong and fourteen cities in China as the main participants. The demographic information includes the education level, age, gender and online teamwork experience. The entire questionnaire conduction period, including the pilot test, was 45 days. 876 questionnaires were issued, and 610 responses were received. With a specific screening question to filter the invalid and unscrupulous responses, 549 valid responses are eventually received, corresponding to a valid response rate of 90 per cent.

Common method variance

Common method variance is a significant issue when conducting business-related research. I accounted for the common method variance by the nature of the questionnaire design and data collection procedure, including providing reversed measurement items and temporal separation to prevent us from erroneously attributing the variance from the measurement method bias to the constructs and minimise the common method variance issue (Williams et al., 2010). Followed with the instructions provided by Williams et al., (2010), a marker variable, which is theoretically unrelated to the independent variable, dependent variable and moderating variables, is provided to test the common method variance with the common marker variable approach. Specifically, there are two models (i.e. the default and controlled models). I assimilate the value of the coefficient of all paths, then set the variance of variables as “1”. The default model is assumed as no common method variance where the factor loads are “0”, and the factor loads in the controlled model are “a.” Eventually, with the chi-square difference test, the result of the default model is as CMID = 1199.384, DF = 113, while the result of the controlled model is as CMID = 1199.724, degree of freedom (DF) = 112. Therefore, the difference in CMID between the default model and the controlled model is 0.34, and the difference of DF is 1 (p = 0.560, > 0.05), indicating that the difference is not statistically significant. Therefore, the common method variance is unlikely an issue in the data of this study (Williams et al., 2010).

Measures

Mutual assistance

Since there is relatively little literature conducting quantitative analysis methodology for mutual assistance, I acknowledge the related keywords for mutual assistance are mutual support and mutual aid. Taking (Gitterman, 2005) research as a reference, I developed four items to measure respondents’ perception of mutual assistance in their online team working experience. The sample item is: “I feel that my team members (including me) can understand other team members’ psychological needs, concerns, and cultural issues.”

Perceived team cohesiveness

Based on (Langfred, 1998) research, three items were developed to determine respondents’ perception of team cohesiveness. Sample item includes “I feel that my team members (including me) are willing to give and receive suggestions to and from other team members to complete the tasks.”

Perceived team effectiveness

I deem that the perceived team effectiveness is the capacity of a team to accomplish the goals or objectives administered by authorised personnel or the organisation, constituted with the achievement of goals, operational efficiency and quality, and satisfaction with teamwork. Three items are selected from (Carson et al., 2007) research and adjusted to measure students’ perception of team effectiveness. A sample item is: “I feel that my team members (including me) can complete the team project effectively and efficiently.”

Most of the measurement items in the questionnaire are with a 5-point Likert scale from “1” to “5”, where “1” refers to strongly disagree and “5” refers to strongly agree. The control variables include gender, age, educational level, the number of past online team project experiences and the team size on a nominal scale.

Data analysis methods

The structural equation modelling (SEM) method is adopted to answer the above research questions. Compared with traditional multivariate techniques, SEM is appropriate for this study because it can (MacCallum & Austin, 2000): (1) test the model, (2) estimate the unobserved variables and (3) explicitly assess the measurement errors. Besides, I target generations Y and Z as the primary research respondents because they represent the growing population in the job market that witnesses and experiences the development of remote working and virtual collaboration in this new era (Bencsik et al., 2016; Marshall & Wolanskyj-Spinner, 2020; Schroth, 2019). Statistic Package for the Social Science (SPSS) software with version 26 and Analysis of Moment Structures (AMOS) with version 26 was used for measuring validation and examining the proposed structural model. In specific, there are two phases in this section where the first phase is to examine the factor structure of a set of the studied variables, including the construct reliability and validity of the proposed research model with the confirmatory factor analysis, while the second phase is about the structural model analysis and hypothesis testing.

Results

Demographic data description

Table 1 shows that the respondents consist of approximately 36 per cent males and 64 per cent females. The majority of the respondents are 19 to 21 years old (69 per cent), with a bachelor’s degree (87 per cent), and have 7 to 9 past online teamwork experience (25 per cent). There were no significant differences between males and females in mutual assistance (F(1,547) = 0.510, p = 0.249, Cohen’s d = 0.3406), nor in perceived team cohesiveness (F(1,547) = 0.857, p = 0.199, d = 0.3401) and perceived team effectiveness (F(1,547) = 0.802, p = 0.413, d = 0.3431). Likewise, no significant differences were found on mutual assistance (F(3,545) = 0.990, p = 0.397, ${\eta }^2$ = 0.005), perceived team cohesiveness (F(3,545) = 1.374, p = 0.250, ${\eta }^2$ = 0.008), and perceived team effectiveness (F(3,545) = 0.664, p = 0.574, ${\eta }^2$ = 0.004) between educational levels. This means that the participants did not attribute mutual assistance, perceived team cohesiveness and effectiveness based on their gender and educational levels. I found a significant difference between online teamwork experience on perceived team effectiveness (F(6,542) = 2.397, p = 0.027, ${\eta }^2$ = 0.026), but not on mutual assistance (F(6,542) = 2.017, p = 0.062, ${\eta }^2$ = 0.022), nor perceived team cohesiveness (F(6,542) = 1.210, p = 0.300, ${\eta }^2$ = 0.013). However, a follow-up post hoc analysis based on the Turkey Honest Significant Difference test revealed that compared to the online teamwork experience of 4 to 6 projects, team members with 19 projects or above perceive lower team effectiveness (MD =  − 0.2093, p = 0.024, 95% CI [− 0.4026, − 0.0160]). It is probably because experienced people may have higher standards and expectations to their team members. Especially based on the negativity bias theory, people are more attentive and “vulnerable” to negative stimuli (Cacioppo et al., 1999; Vaish et al., 2008), which may play an important effect on how team members construe teamwork performance and experience.

Table 1.

Demographic information of respondents

		Frequency	Valid percent
Gender	Male	197	35.9
	Female	352	64.1
Age	Below 18	32	5.8
	19–21	376	68.5
	22–24	96	17.5
	25–27	24	4.4
	Above 28	21	3.8
Education level	College or below	33	6
	Bachelor's degree	478	87.1
	Master's degree	27	4.9
	Above master's degree	11	2
Past online working experience (years)	1–3 projects	70	12.8
	4–6 projects	135	24.6
	7–9 projects	138	25.1
	10–12 projects	102	18.6
	13–15 projects	43	7.8
	16–18 projects	27	4.9
	Above 19 projects	34	6.2
Total		549	100%

Normality test

Normality test to the observed variables is a recommended procedure before confirmatory factor analysis (CFA) (Bentler & Wu, 2005; Gao et al., 2008; Sharma et al., 2013), since the multivariate distribution whatever the observed variables are univariate normal can be multivariate non-normal (Byrne, 2004; Mahmoud & Khalifa, 2015; West et al., 1995). Following the instruction by (Tabachnick & Fidell, 1996), the log transformation (logarithms) and square root transformation are tried respectively to transform variables due to non-normal distribution. Finally, the log transformation procedure is selected as it presents a more satisfactory result. Then, I measure the multivariate normality with Mardia’s multivariate kurtosis (Gao et al., 2008). The results shown in Table 2 indicate that most of the critical ratios (Z-score) accessing skewness and kurtosis are smaller than 5, so the sample is deemed normal (Bentler & Wu, 2005; Sharma et al., 2013), and it is concluded that the condition of this sample being multivariate normal is satisfied.

Table 2.

Descriptive statistic of normality results

	Skewness	Std. Error	Z-score	Kurtosis	Std. Error	Z-score
MA1	− 0.220	0.104	− 2.107	− 0.512	0.208	− 2.4577
MA2	− 0.480	0.104	− 4.605	− 0.328	0.208	− 1.5764
MA3	− 0.468	0.104	− 4.487	− 0.351	0.208	− 1.6880
MA4	− 0.445	0.104	− 4.267	− 0.153	0.208	− 0.7344
PTC1	− 0.364	0.104	− 3.487	− 0.389	0.208	− 1.8689
PTC2	− 0.302	0.104	− 2.894	− 0.679	0.208	− 3.2621
PTC3	− 0.285	0.104	− 2.732	− 0.633	0.208	− 3.0405
PTE1	− 0.305	0.104	− 2.929	− 0.358	0.208	− 1.7217
PTE2	− 0.413	0.104	− 3.964	− 0.433	0.208	− 2.0819
PTE3	− 0.445	0.104	− 4.272	− 0.301	0.208	− 1.4453

PTC perceived team cohesiveness, PTE perceived team effectiveness

Item reliability, construct reliability and construct validity

Notwithstanding that the question items used in the study were from the past literature, which should be reliable, I still checked the item reliability by evaluating the loading of each item and examining the correlation between each question item and the corresponding construct. If the items’ loading and item-construct correlation are 0.6, then these items are reliable (Hair et al., 2010). In this study, almost all items have standardised loading factors over 0.7.

The construct reliability can be determined by the value of Cronbach’s alpha, which tests the accuracy of a set of items to a single unidirectional latent construct (Tarhini et al., 2017). Since the universally accepted reliability value is at least 0.7 (Keysor et al., 2003; Robinson, 1991), and the Cronbach’s alpha of all constructs shown in Table 3 is over 0.7, the constructs in this study have adequate reliability.

Table 3.

Results of validity and reliability

Variable	Indicator	Mean	SD	Standardized loading factor	Cronbach’s alpha	Construct reliability (CR)	Average variance extract (AVE)
Mutual assistance	MA1	3.98	0.80	0.815***	0.858	0.836	0.563
	MA2	3.83	0.78	0.786***
	MA3	3.59	0.91	0.646***
	MA4	3.87	0.77	0.743***
Perceived team cohesiveness	PTC1	4.08	0.67	0.819***	0.820	0.828	0.616
	PTC2	4.08	0.73	0.795***
	PTC3	3.95	0.79	0.738***
Perceived team effectiveness	PTE1	4.00	0.75	0.801***	0.853	0.854	0.661
	PTE2	3.89	0.79	0.815***
	PTE3	3.94	0.75	0.822***

Cα ≥ 0.7; Composite reliability ≥ 0.7; AVE ≥ 0.5

***p < 0.001; **p < 0.01; *p < 0.05

Composite reliability (CR) and average variance extracted (AVE) are the standard measures to assess the convergent validity and discriminant validity (Hair et al., 2010), indicating that an established convergent validity should be with the CR value of above 0.7 and AVE value of above 0.5 in the standard practice. The discriminant validity occurs only when a higher total AVE value of the average value of all variables than their correlation value (Hair et al., 2010). The results shown in Table 3 show that the AVE for all constructs is higher than 0.5. Additionally, the square root of AVE is all over 0.7, higher than their correlation values shown in Table 4. Thus, the constructs have sufficient convergent validity and discriminant validity.

Table 4.

Descriptive statistics and average variance extracted from constructs

	Gender	Age	Educational Level	No. of Virtual Teamwork Experience	Team Size	MA	PTC	PTE
Gender	1
Age	− 0.114**	1
Educational Level	− 0.037	0.457**	1
No. of Virtual Teamwork Experience	0.121**	− 0.105*	− 0.069	1
Team Size	0.041	− 0.060	− 0.035	0.046	1
Mutual Assistance (MA)	0.044	− 0.011	− 0.033	0.057	0.074	0.750
Perceived Team Cohesiveness (PTC)	0.048	− 0.071	− 0.022	0.072	0.027	0.715**	0.785
Perceived Team Effectiveness (PTE)	0.037	− 0.028	− 0.002	0.114**	0.053	0.708**	0.704**	0.813

Diagonal elements (in italic) are square root of the AVE; off-diagonal elements are the correlations among constructs

**Correlation is significant at the 0.01 level (2-tailed), *Correlation is significant at the 0.05 level (2-tailed)

Measurement model and structural model index

Utilising the approach of structural equation modelling with SPSS-Amos v26, I adopt several models fit indexes for the measurement model in this study, including CMID/DF, comparative fit index (CFI), normed fit index (NFI), the root means the square error of approximation (RMSEA), Tucker-Lewis index (TLI), and standardised root means square residual (SRMR) (Hu & Bentler, 1999; Mahmoud & Khalifa, 2015). In testing the measurement model, the fitness indices were not satisfactory (CMID/DF = 6.626; IFI = 0.948; TLI = 0.926; CFI = 0.948; RMSEA = 0.101). The modification indices indicated that the error terms corresponding to MA2 and MA3 and MA3 and MA4 were correlated. Therefore, these error terms were allowed to covary freely in the revised model. The fit of the revised model was satisfactorily meeting the requirements, and the indices for the measurement model and structural model are the same (shown in Table 5: CMIN/DF = 4.884; NFI = 0.958; CFI = 0.966; TLI = 0.949; RMSEA = 0.084; SRMR = 0.0344; (Hu & Bentler, 1999; Kline, 2015; Maroufizadeh et al., 2014; Norberg et al., 2007). Figures 1 and 2 demonstrate the measurement and structural models with standardised estimate values and the model fit summary.

Table 5.

Model fit summary for measurement and structural model

Fit index	Recommended value	Structural model
Absolute fit measures
RMSEA	< 0.1; preferably .05 to .08	0.084
SRMR	< 0.1; preferably < .05	0.0344
Incremental fit measures
NFI	> .90	0.958
CFI	> .90	0.966
TLI	> 0.90	0.949
Parsimony fit measures
CMIN/DF	< 5 preferable < 3	4.884

Fig. 1.

Measurement model

Fig. 2.

Structural model

Hypothesis test

The path significance levels were predicted by using bootstrap confidence intervals. The results of multiple regression analyses are shown in Table 6. On the other hand, I also use simple slope analysis to check if the results are robust and replicable.

Table 6.

The direct effects and moderation effects

Variable	Perceived team cohesiveness (Model 1)	Perceived team effectiveness (Model 2)	Perceived team cohesiveness (Model 3)	Perceived team effectiveness (Model 4)
Independent variable
Mutual assistance (MA)	0.832*** (0.731***)		0.741*** (0.755***)	0.726*** (0.470***)
Perceived team cohesiveness		0.126 (0.731)
Moderator
Perceived team cohesiveness				0.132 (0.374***)
Interaction variable
Mutual assistance*Perceived team cohesiveness				− 0.085 (− 0.124*)
R2	0.732	0.725	0.755	0.798
Adjusted R2	0.535	0.525	0.569	0.634
F	632.743	607.062	723.050	317.922

***p < 0.001; **p < 0.01; *p < 0.05

Hypothesis 1 posits a positive relationship between mutual assistance and perceived team cohesiveness in the CMC settings. In the model of Table 6, the result pinpoints a statistically significant facilitating effect of mutual assistance on perceived team cohesiveness (β = 0.832; p < 0.001), interpreting that when mutual assistance increases with one unit, the perceived team cohesiveness is predicted to increase with 0.832 unit. Likewise, the simple slope analysis (Aiken et al., 1991) also found a positive relationship between mutual assistance and perceived team cohesiveness (β = 0.731, p < 0.001). Therefore, hypothesis 1 is supported.

Hypothesis 2 proposes that virtual team members’ perceived team cohesiveness is positively related to members’ perceived team effectiveness in CMC settings. In model 2 of Table 6, the result repudiates this hypothesis by highlighting that perceived team cohesiveness is not significantly related to perceived team effectiveness (β = 0.126, p > 0.05). The simple slope analysis also found a statistically insignificant relationship between perceived team cohesiveness and perceived team effectiveness (β = 0.731, p > 0.05). Accordingly, hypothesis 2 is unsupported.

In hypothesis 3, the mutual assistance practice in CMC settings is anticipated to bolster virtual team members’ perception of team effectiveness. This hypothesis is corroborated by the results in model 3 of Table 6 that mutual assistance is positively related to perceived team effectiveness (β = 0.741, p < 0.001). In other words, when mutual assistance increases with one unit, perceived team effectiveness will increase by 0.741 units accordingly. This simple slope analysis result also buttresses this finding (β = 0.755, p < 0.001). Therefore, hypothesis 3 is supported.

Hypothesis 4 predicts that members’ perceived team cohesiveness can positively moderate the relationship between mutual assistance and perceived team effectiveness. However, in model 4 of Table 6, we can observe a statistically insignificant moderation effect of interaction in terms of mutual assistance and perceived team cohesiveness on perceived team effectiveness (β =  − 0.085, p > 0.05). Despite the simple slop analysis revealing that the interaction term is statistically significant at the 0.05 significance level, the interaction term's moderation effect is negative on perceived team effectiveness (β =  − 0.124, p < 0.05). Therefore, we can conclude that perceived team cohesiveness cannot positively moderate the relationship between mutual assistance and perceived team cohesiveness in the CMC settings, and hypothesis 4 is not supported.

Discussion

This study aims to shed light on the paucity of studying psychological mechanisms to strengthen team cohesiveness and effectiveness in virtual teamwork. Drawing on the hyperpersonal communication theory (i.e. hyperpersonal model of computer-mediated communication; Walther, 1996), I advocate the existence and importance of mutual assistance as one of the benevolent practices via CMC among team members and how mutual assistance is expectedly linked with members’ perception of team cohesiveness and team effectiveness. The present study found that team members in CMC settings will perceive a higher cohesiveness and effectiveness of the team, along with the increasing frequency of mutual assistance. Further, team members’ perception of team cohesiveness neither significantly impacts the perception of team effectiveness nor moderates the relationship between mutual assistance and perceived team effectiveness.

The findings on the relationship between mutual assistance, perceived team cohesiveness and perceived team effectiveness in CMC settings are consistent with previous literature (e.g. Chen, 2018; Christensen et al., 2006; SOPA, 2014; Yang & Lin, 2022). Specifically, past literature examines the effect/association of/between these variables in different contexts, such as software development and product development management, hospital management, craft beer industry sector, team learning of emerging technology startup context, as well as from various perspectives such as proximity, organisational justice, reciprocal tolerance, shared belief, collective identity, the pursuit to a common goal and shared knowledge (Chen et al., 2022; Dayan & di Benedetto, 2008; Hoegl & Proserpio, 2004; Mathias et al., 2018; Padgett, 2013; Saqr et al., 2020). Specific to the virtual team and CMC context, the study of Yang and Lin (2022) drew on the social penetration theory and found that virtual team cohesion can be acquired in CMC settings in addition to face-to-face interaction, from the perspective of reciprocal information disclosure behaviour about work-related knowledge and opinions to workplace improvements. Empirical results from the study of Kahlow et al. (2020) also supported that virtual teams in CMC settings can have solid norms and leadership behaviours, indirectly facilitating trust, cohesion and decision making. The present study fills in the research gap by centring on the CMC settings with the adoption of hyperpersonal communication theory to provide an alternative explanation of the relationship between mutual assistance and perceived team cohesiveness. The managerial outcome of perceived team effectiveness is additionally investigated.

I argue that symmetry and mutual obligation are two critical natures in all types of interpersonal interactions to improve members’ perceived team cohesiveness, resulting in consistency between the present findings and past literature. First, symmetry refers to the exchange of specific resources within a relationship where “benefits received come hand in hand with benefits given” (Casciaro et al., 2014, p.709). Team members are expected to do the same for others else when they receive help in terms of affection and technical and socioemotional support (Carley & Krackhardt, 1996). For example, the present study argued the mutual willingness to provide and receive help, while Chen et al. (2022) valued sharing knowledge reciprocally in their research. Second, mutual obligation underscores the equalisation of the relationship, which is complementary to symmetry that only concerns the social exchange perspective in the interaction. Specifically, team members are obliged to care for the welfare of other team members, even if doing so provides neither present nor future material rewards (Clark & Mills, 1979); instead of providing benefits (regardless of knowledge or assistance) with an expectation of receiving comparable benefits in return (which is direct reciprocity; Casciaro et al., 2014). Indeed, the intimateness between team members can be beyond work relationships and be much more intimate, along with the increase in positive interactions, based on hyperpersonal communication theory (Walther et al., 2015). Future studies related to these two factors in the team cohesion literature are recommended.

Furthermore, the results reveal that perceived team cohesiveness neither affects perceived team effectiveness nor moderates the relationship between mutual assistance and perceived team effectiveness, though these constructs are highly correlated. I consider three possible explanations. The first possible reason behind that is the possibility of the curvilinear relationship between perceived team cohesiveness and perceived team effectiveness. Considering the nature of virtual teams where the team members are from different backgrounds, it is reasonable that the relationship between the two constructs and the moderating effect of perceived team cohesiveness is not linear (Lau & Murnighan, 1998; Webber & Donahue, 2001). Indeed, the study of Park et al. (2017) indicated that team cohesion could be a contextual factor causing a reduction in team effectiveness when the team efficacy is either excessively high or low. Groupthink bias is probably a typical example to illustrate such a nonlinear relationship and too-much-of-a-good-thing effect: the level of perceived team effectiveness regarding decision quality may be indifferent to both non-cohesive and cohesive teams when teams are under groupthink conditions and provided with an excuse for potentially poor performance (Fung, 2014; Turner et al., 1992).

Additionally, the second possible explanation for the nullification of perceived team cohesiveness’s predictive and moderative effect is the aspects of context, such as the team culture and leadership style, which may promote individual differences and inconsistency of team members’ perception in team processes. For example, perceived team cohesiveness can strengthen perceived team effectiveness when there is transformational leadership in the team and team members are empowered (Jung & Sosik, 2002). The final explanation relates to data. Specifically, though it is out of the focus in this study, it is probably better to examine the team's longevity and cooperation experience with team members, such as how long team members took to finish their project. Team members are expected to know each other more as time passes. Particularly the “hyperpersonal effect” will be intensified and amplified by the feedback system and behavioural confirmation, which is one of the assumptions of hyperpersonal communication theory (Walther, 1996). On the other hand, some teams where team members had past coworking experience in other teams or already made acquaintances may influence perceived team cohesiveness, thus, resulting in different outcomes of perceived team effectiveness. Unfortunately, the present study did not include these variables during the analysis, and related future studies are encouraged.

Theoretical contributions

The findings on the current study move forward the cumulative theoretical knowledge relating to mutual assistance, perceived team cohesiveness and perceived team effectiveness. First, this study contributes to the team management literature by being the first to introduce hyperpersonal communication theory in explaining how team members are expected to perceive higher team cohesiveness and effectiveness through mutual assistance in CMC settings. Specifically, unlike past work on team cohesion that focused on group-level team cohesion and team outcome as measurement (e.g. Chang & Bordia, 2001; Mach et al., 2010; Whillans et al., 2021), this study emphasised the importance of individual-level analysis. The present study also responded to the call from the studies with the growing interest and attention placed on the psychological mechanism of how perceived team cohesiveness and effectiveness are established (Salas et al., 2015). I reconcile the inconsistent relationships between mutual assistance and perceived team cohesiveness, and perceived team effectiveness (Kahlow et al., 2020; Kozlowski & Ilgen, 2006; Morrison-Smith & Ruiz, 2020; Mysirlaki & Paraskeva, 2019), by providing a context-specific boundary condition and alternative explanation. Additionally, the findings suggest that the prediction effect of team cohesion on team effectiveness cannot be assumed to be ever-present (e.g. Barrick et al., 1998; Manata, 2020; Paul et al., 2016; Tekleab et al., 2009); instead, perceived team cohesiveness may only associate with perceived team effectiveness under a complex situation of virtual team processes. This study opens doors for future research about psychological intervention in team management fields that can enhance team cohesiveness and effectiveness at the individual perception level.

Second, this study contributes to the mutual assistance literature from several perspectives. Specifically, most current literature focused on conducting qualitative analysis for mutual assistance (e.g. Salas et al., 2015), where quantitative and self-reporting measures for mutual assistance in perception level are in demand. The present study developed four items with reasonable reliability and validity to measure the team members perceived mutual assistance in the CMC settings, establishing the replicability of not only the theorised framework but also the mutual assistance construct. Additionally, this study extends the mutual assistance literature to a specific context of CMC settings, achieving a more nuanced understanding of the boundary effect mutual assistance plays in team cohesion and team effectiveness at the perception level and reconciling the inconsistent findings in this domain (Kozlowski & Ilgen, 2006; Yukelson, 1997). Lastly, this study also contributes to the hyperpersonal communication theory by extending this theory into team cohesion and team effectiveness literature. It identifies and theorises the context-specific antecedents and consequences of computer-mediated communication in a virtual teamwork setting, facilitating the generalizability of this theory in this domain.

Practical implications

This study provides several implications for organisations. Firstly, it suggests that a critical factor, namely, mutual assistance in the CMC setting, can strengthen team members' perceptions of team cohesiveness and effectiveness. This is an essential and beneficial means for organisations to resolve some implicit team management issues, such as workplace isolation which will weaken employees’ productivity and belongingness to the organisation. The practice of mutual assistance is spontaneously activated among team members to manage their impressions and interpersonal relationships in the teamwork and asynchronous CMC context. Organisations that want to improve employee loyalty are recommended to design policies encouraging employees to help needed colleagues and cultivate an amicable norm in the workplace. If the employees are introverted or aloof, more time and effort should be put into implementing training programmes and practices for employees to become mutually helpful. Fehr et al. (2017) proposed that the expression of gratitude (e.g. thank you notes) can facilitate the beneficiary’s future intention to provoke a communal exchange norm when the benefactor requires support. In this light, the components related to mutual assistance are recommended for orientation and appreciation training programmes.

Related implications fall in the performance management section of the HRM function such that appraisal programmes should be promotive towards the expression of gratitude, i.e. rewards given as recognition by peers. For example, performance management managers can set up a “Most Popular Award” to encourage employees to interact and help each other more. As mutual assistance consists of alacrity in giving and receiving help, the programme, which may encourage the expression of gratitude that could motivate the willingness to provide help to others. Nevertheless, organisations should be made aware of different communication channels/methods. Only the asynchronous CMC method is theoretically supposed to leave team members enough time to mindfully organise their responses and selectively present them in a favourable and desirable way.

Limitations and future research

Although this study contributes essential understandings and values to the literature, there are still some considerable limitations. Firstly, this is a cross-sectional design collecting the data at one given point across a sample population from different educational backgrounds, which cannot eliminate the possibility that the simple correlations may change in the long term due to the psychological or behavioural changes over time, so a longitudinal study is recommended for future studies to consolidate the results. Secondly, this study does not consider other possible contextual factors that may drive the variant practices and applications in mutual assistance. To fully understand the effect of mutual assistance on perceived team cohesiveness and effectiveness, future works are recommended to conduct an experimental study on a specific sample population with the close or same background. Finally, the present study does not test the effect of mutual assistance within the core nomological network of perceived team cohesiveness and effectiveness. Future studies are encouraged to examine other factors, such as personality traits, dispositions and cultures potentially giving rise to mutual assistance in a virtual teamwork context.

Conclusion

Under the circumstance of globalisation and post-pandemic, virtual teamwork collaboration has become a pervasive practice for multinationals. Understanding related psychological mechanisms is a significant prerequisite to strengthening the actual team cohesiveness and effectiveness. The present study contributes to this effort by providing new and alternative theoretical explanations of the hyperpersonal communication theory and identifying the strengthening effect of mutual assistance on perceptions of team cohesiveness and effectiveness among virtual team members. Finally, this study contributes to designing and deploying effective mutual assistance interactions in team management practices.

Supplementary Information

Below is the link to the electronic supplementary material.

Yuejun Lawrance Cai

is currently a full-time research assistant at the department of Management of the Hong Kong University of Science and Technology. He received his bachelor’s degree in business management with first class honours from the Hong Kong Polytechnic University. His research interests are interpersonal and group relationships as well as conflict management and solutions.

Data availability

The corresponding author's data supporting this study's findings are available upon reasonable request.

Declarations

Conflict of interest

The authors declare that they have no conflicts of interest.

Informed consent

Informed consent was obtained from all individual participants involved in the study.