The differential effects of self-view in virtual meetings when speaking vs. listening

ABSTRACT

With the surging reliance on videoconferencing tools, users may find themselves staring at their reflections for hours a day. We refer to this phenomenon as self-referential information (SRI) consumption and examine its consequences and the mechanism behind them. Building on self-awareness research and the strength model of self-control, we argue that SRI consumption heightens the state of self-awareness and thereby depletes participants’ mental resources, eventually undermining virtual meeting (VM) outcomes. Our findings from a European employee sample revealed contrary effects of SRI consumption across speaker vs listener roles. Engagement with self-view is positively associated with self-awareness, which, in turn, is negatively related to satisfaction with VM process, perceived productivity, and enjoyment. Looking at the self while listening to others exhibits adverse direct and indirect (via self-awareness) effects on VM outcomes. However, looking at the self when speaking exhibits positive direct effects on satisfaction with VM process and enjoyment.

KEYWORDS:

• Self-view
• virtual meetings
• self-awareness
• sender-receiver framework, zoom
• sender-receiver framework
• Zoom

1. Introduction

Virtual meetings (VMs)¹ enable physically distanced people to gather in a virtual space. Its growing use transforms organisations and communication practices (Waizenegger et al., 2020) in contexts like business, education (Toney et al., 2021), court hearings, and healthcare, to name a few. With a share of VMs in all meetings across industries skyrocketing to nearly 100% during the pandemic, VMs are anticipated to stay in the post-COVID-19 era (Standaert et al., 2022). Experts predict that 75% of meetings will be held online in 2024 (Gartner, 2020). Nonetheless, the implications of this transformation appear ambiguous. Along with the promises to revolutionise team collaboration (Waizenegger et al., 2020) and allow for better work-life balance (Riemer et al., 2009; Standaert et al., 2022), virtual work poses new challenges related to employee readiness (Eckhardt et al., 2019), trust formation (Robert et al., 2009), and team management (Beranek et al., 2005; Sundermeier, 2022). Besides, for many employees, days in the home office turn into a never-ending chain of Zoom² calls, with little break or time to recover (Fossilien & Duffy, 2020).

Beyond organisational practices, VMs transform how users perceive themselves. Most videoconferencing tools, like Zoom or Microsoft Teams, include a self-view feature that allows tracking “how one looks to others” in a video call (Support.Zoom, 2021). Consider Charlie, a remote product manager at a FinTech start-up. Every time he joins another Zoom call, he sees himself again thumb-sized in the top right corner. Sometimes the person looking at him from the screen does not match his ideal. For example, it makes Charlie think he needs to get a new haircut, his glasses could have a nicer frame, or perhaps he should switch to wearing contact lenses so that the screen reflection does not cover the eyes. Even after using quick fixes like filters (e.g., “Touch Up My Appearance” in Zoom) and setting up a camera angle, the temptation to monitor his live-streamed portrait persists. So does the tension between the need to engage with the content of an online work meeting and the urge to monitor one’s digital look.

Previously, looking at oneself has been limited to morning and evening rituals in front of a mirror. Now, users of videoconferencing tools may find themselves staring at their reflections for hours a day (Pfund et al., 2020; Pikoos et al., 2020). Since February 2020, VMs have occupied about 21.5 hours in a working week, whereas the pre-pandemic average was 14.2 hours (Reclaim.ai, 2021). Indeed, never before have people consumed that much self-referential information (SRI). Referred to as information about oneself (Lipitz et al., 2018), SRI—e.g., one’s name and portrait – is known for its stark potential to attract the attention of its owner (Morin, 2011). Hence, the self-view window accessible at all times during VMs naturally glues users’ attention. Existing eye-tracking (George et al., 2022) and self-report data (Pikoos et al., 2020) converge on users actually watching themselves during a VM. The phenomenon of becoming an object of one’s own attention has been previously studied outside the domain of technology use and was coined self-awareness (Morin, 2011, p. 807). Against this background, we explore:

RQ1:

How is self-referential information consumption related to videoconference outcomes at work?

RQ2:

What is the role of self-awareness in this relationship?

To answer these questions, we adopt the objective self-awareness theory (Duval & Wicklund, 1972) and the strength model of self-control (Baumeister & Vohs, 2007; Muraven & Baumeister, 2000). Furthermore, integrating the sender-receiver framework (Lin et al., 2005) offers a VM-specific understanding of SRI consumption. The responses of 311 European employees about their recent VM at work, along with the technical details of their participation (audio vs. video mode; self-view on vs. self-view off), show that the effects of SRI consumption vary for speakers as information senders and listeners as information receivers.

Together, our findings on the effects of SRI consumption during a VM make multiple contributions. So far, scholars have scrutinised the phenomena of VM stress (Riedl, 2022) and Zoom fatigue (Bailenson, 2021; for review, see Wang & Prester, 2022), while other meeting outcomes and especially the mechanisms behind VM-specific effects remained overlooked. First, the current paper directly contributes to the understudied research area by examining self-awareness as a mechanism behind employees’ satisfaction with VM process, perceived productivity, and enjoyment. Second, our results enrich the information systems (IS) literature on e-collaboration (Ahuja et al., 2020; Riemer et al., 2009; Waizenegger et al., 2020). The differentiation between consuming SRI as a speaker vs. listener unveils the self-view feature’s contrary – positive and negative – potentials. The dichotomy observed and described in this paper yields a more nuanced understanding of hybrid (Haubrich & Hafermalz, 2022; Weritz et al., 2022) and fully remote work (Hacker et al., 2020; Hafermalz & Riemer, 2021). Third, our insights are relevant for guiding digital transformation in the workplace (Eckhardt et al., 2019) and especially in times of crisis (Sundermeier, 2022). By uncovering when and how SRI consumption can be harmful and beneficial, we lay the groundwork for designing interventions that mitigate adverse effects and accentuate the positive potential of this technology.

The article is organised as follows. The next section builds a theoretical foundation, summarising past knowledge on VMs, how videoconferencing primes the state of self-awareness, sources of self-awareness, and its consequences. Section 3 develops the hypotheses and research model. Section 4 elaborates on the results. Section 5 discusses the findings and offers implications for scholars and practitioners. Details on methodology and additional analyses are in (online) appendices.

2. Theoretical background

In this section, we summarise previous research on VMs, which informs our theorising efforts and provides a basis for discussing and interpreting our findings. We then introduce our view of VMs as self-awareness-inducing environments and describe how SRI consumption translates into VM outcomes.

2.1. Virtual meetings research in information systems

VMs allow users in different physical locations to hold meetings through transmitted audio and video signals (Hacker et al., 2020). IS scholars have early recognised the potential of videoconferencing technology in enabling the synergy of otherwise distributed human resources (Dennis et al., 1988; Riemer, 2009; for review, see Ahuja et al., 2020). Early studies underscored, however, that employee performance in a VM setting was lower than in an offline, face-to-face setting (Potter & Balthazard, 2002). Technology-mediated communication reduced a team member’s trust intention and trust behaviour (Robert et al., 2009). To perform effectively in a digital workplace and to keep employees satisfied, virtual competence – the combination of virtual self-efficacy, virtual media skills, and virtual social skills – appeared to be the key (Wang & Haggerty, 2011), along with careful planning and execution of VMs. With an optimal transition to a virtual workforce in focus, gradual stage-by-stage models were proposed (see Eckhardt et al., 2019).

However, the sudden onset of the pandemic did not allow for the well-conceived phased plan to become a reality. Over a few months, remote work, healthcare, and education, enabled by VMs, have rapidly become the new norm (Standaert et al., 2022). As the VM adoption progressed, along with successful cases of recreating togetherness through VMs (Hacker et al., 2022; Waizenegger et al., 2020) and sustaining productivity scholars have documented challenges in managing virtual teams and engaging employees (Sundermeier, 2022) and students (Toney et al., 2021). In particular, a growing body of research has linked participation in VMs to such adverse outcomes as fatigue (Fauville et al., 2021, 2023; Bennett et al., 2021), stress (Riedl, 2022), nonverbal overload (Bailenson, 2021) and lower team cohesion (Beranek et al., 2005; Standaert et al., 2016).

In sum, mixed results of previous research hint that face-to-face gatherings differ more substantially from the VMs than it appears at first sight. Responding to the growing importance of VMs for organisations, the daily intensified, multi-hour use of VMs illuminated a fascinating peculiarity of VMs for users, upon which we expound next.

2.2. Videoconferencing as a self-awareness-priming environment

VM environments represent a constrained version of social interactions, limiting non-verbal cues like body language, eye gaze (e.g., Hacker et al., 2022; Standaert et al., 2022), and even extreme voice frequencies (Davis et al., 2015, pp. 5–6). At the same time, the VM environment allows users what is not possible in a physical meeting – an opportunity to see themselves in the self-view window (Bailenson, 2021). The self-video, absent in the physical world and attention-grabbing by nature, introduces another level of social complexity in that the VM attendees interact not only with other users through the interface of a VM tool but also with themselves by consuming and processing the self-referential information (SRI).

People’s interaction with SRI has been previously described in the self-awareness literature and can best be explained using the objective theory of self-awareness (Duval & Wicklund, 1972). This theory is premised on the idea that, at any point in time, individual attention can be headed either outward to the external environment (e.g., to a physical object or social setting) or inward to various attributes of the self (Duval & Wicklund, 1972). The latter is coined self-awareness and refers to the “capacity of becoming the object of one’s own attention” (Morin, 2011, p. 807). Self-awareness is predominantly conceptualised as a state (Duval & Wicklund, 1972; Fenigstein et al., 1975), as it reflects the responsiveness to “situational variables, chronic dispositions, or both” (Fenigstein et al., 1975, p. 522). Less often, self-awareness is theorised as a stable trait referring to a chronic tendency to fix attention on the self (Fenigstein et al., 1975). Because we examine the use and associated effects of the self-view feature – a situational stimulus – this study focuses on the state of self-awareness as susceptible to fluctuations and manipulation (Govern & Marsch, 2001).

2.3. Inducing self-awareness

Self-awareness can be evoked by stimuli that cause people to focus attention on themselves. The triggers of self-awareness, as comprehensively summarised by Morin (2004, 2011), can be clustered into (1) the social world, (2) the physical world, and (3) the self. First, the triggers embedded in (1) one’s social world, including face-to-face interactions, self-relevant feedback from others, and social comparisons, as well as a mere presence of an audience, can heighten one’s state of self-awareness (Carver & Scheier, 1978). Second, the self-awareness triggers exist in (2) one’s physical world and represent “objects and structures that produce bodily awareness and self-world differentiation” (Morin, 2011, p. 815). For example, things like mirrors but also photos, video, and audio recordings of a person provide relevant information on one’s self-aspects: physical appearance, posture, and tone of voice (Innes & Young, 1975; Morin, 2011). Finally, (3) the self can be a source of self-awareness, drawing attention through proprioception, inner speech, and autoscopic imagery (Morin, 2011).

In the context of VMs, Morin’s (2004, 2011) (1) “social world” and (2) “physical world” stimuli are particularly salient. Specifically, the “social world” is represented by attendees in a VM, whose presence may initiate the processes of self-awareness. Indeed, robust evidence from the offline context informs that facing an audience creates self-focus (e.g., Carver & Scheier, 1978; Daly et al., 1989), with even one observer being enough to trigger self-awareness (Buss, 1980).

Further, the design of VM tools introduces the (2) “physical world” triggers of self-awareness (Morin, 2004, 2011) by offering the self-view window (sometimes called video feedback (M. K. Miller et al., 2017). The self-view window works as a digital mirror, i.e., it displays to a person how they look in real-time (Support.Zoom, 2021) and thereby conveys self-referential information (SRI).

SRI that flows via self-view is of particular interest. First, this stimulus is absent in offline meetings and thus represents a distinctive VM characteristic, which nature and effects only start to surface (Pfund et al., 2020). Second, compared to other self-awareness stimuli present in the VM environment, SRI has stark attention-grabbing properties. Indeed, research into static images shows that among familiar faces, people look at their own one longer (Devue et al., 2009), as it holds emotional value and familiarity (Brédart et al., 2006; Devue & Brédart, 2008). Raising the bar, videos animate static pictures with movement and sound, making them a richer and more emotionally engaging medium than images. In a recent survey of 323 Australian adults regarding their recent video call, 27% reported paying most visual attention to their own face/body; 53% – to both their own face and others’ faces; and only 16% were mainly focused on the appearance of others (Pikoos et al., 2020). Likewise, the U.S. female respondents in the study of Pfund et al. (2020) admitted looking at themselves nearly one-third of the time during a video call. As a result, SRI broadcasted in the self-view window can have a strong effect in inducing self-awareness.

To sum up, with two prominent triggers of self-awareness in one place, VMs constitutes an intriguing research object. This study puts a spotlight on SRI consumption via a self-view window (a “physical world” stimulus) as a unique driver of VM users’ self-awareness, absent in offline meetings. Yet, we also acknowledge the potential influence of the presence of others (a “social world” stimulus) and control for it. The work-related VM use, i.e., tied to an individual’s regular and purposeful activity to earn a livelihood, raises the stakes of understanding the implications of being in a self-awareness state. Informed by the strength model of self-control and objective self-awareness theory, we describe how SRI consumption translates into VM outcomes next.

2.4. Effects of SRI consumption in a virtual meeting

In general, SRI broadcasted through the self-view window competes with work-related information for an employee’s limited attention. Put differently, focusing on a VM, be it an update on progress or collective decision-making, requires that users proactively inhibit their urges to engage with the SRI in the self-view window. The established strength model of self-control (Baumeister & Vohs, 2007; Muraven & Baumeister, 2000) addresses similar taxing contexts. It emphasises that acts of self-control (e.g., managing own behavioural and emotional states, persistence) are underpinned by a single, metaphorical strength with a finite capacity. Like muscles get tired during an exercise (Muraven & Baumeister, 2000), available resources get depleted by the efforts to prioritise one goal over competing urges. This, in turn, can impair performance (Baumeister & Vohs, 2007). The meta-analysis of 83 studies found a medium-to-large effect of depletion condition (when self-control was required) on task performance and related outcomes (Hagger et al., 2010). Against this backdrop, we posit that users may scrutinise themselves at the cost of productivity and the overall success of a VM.

Complementing and refining this general line of reasoning, objective self-awareness (OSA) theory (Duval & Wicklund, 1972) suggests that consuming SRI puts VM users in a state of self-awareness in which they perceive themselves as an object. “When attention is directed inward, and the individual’s consciousness is focused on himself, he is the object of his own consciousness – hence ‘objective’ self-awareness” (Duval & Wicklund, 1972, p. 2). Once self-aware, we automatically contrast what we see at the very moment (the “real” self) with who we eventually aspire to be (the “ideal” self) (Duval & Wicklund, 1972). Such self-evaluation is likely to reveal a discrepancy between the two representations. A negative real-ideal discrepancy is naturally unpleasant and generates (1) self-criticism, followed by escapism behaviours, or (2) attempts to reduce the gap by directly modifying the real self or the standard (Morin, 2011). The less common positive discrepancy, e.g., after seeing an unusually nice reflection of oneself, will likely make a person seek more self-awareness and increase self-focus (Morin, 2011). Either way, dealing with self-awareness will occupy mental resources, initially supposed to be dedicated to a work-related videoconference. Previously in other contexts, excessive self-awareness has been shown to impair self-control in sports (Baumeister & Steinhilber, 1984) and the education domain (Steele & Aronson, 1995).

Taken together, the strength model of self-control (Baumeister & Vohs, 2007; Muraven & Baumeister, 2000) assumes that any extra stimulus that imposes a necessity for self-control along with the primary task will deplete the limited resources and eventually hamper the activity outcomes. The objective self-awareness theory (Duval & Wicklund, 1972) details what exactly happens under SRI consumption and highlights the phenomenon of self-awareness as a natural result of self-focus. Hence, combining these theoretical lenses seems useful in examining the effects of SRI consumption on VM outcomes. In the next section, we define the constructs and explain the proposed relationships illustrated in our research model (Figure 1).

Figure 1. Research model.

3. Research model: constructs and hypotheses

3.1. Direct effects of SRI consumption on VM outcomes

At the start of any VM, enabling the video also automatically broadcasts the self-view to the participant, if not intentionally disabled. Key VM tools on the market, such as Zoom, Google Meets, Microsoft Teams, and Skype, stream self-view by default once the video mode is activated (e.g., Support Zoom, 2021). To describe the extent of attention to one’s own video in the course of VM at work, we use the term self-referential information consumption (SRI).

Given that videoconferences at work are purposeful business-related interactions (Rogelberg et al., 2006), VM users are expected to be primarily concerned with a work-related conversation, whether it is information exchange, a problem-solving task, or a discussion of some current issue. Exposure to SRI in the self-view window adds another stimulus to be processed, putting an individual into a situation with two competing urges: to follow a VM vs. their own look. To engage with the topic of a VM and to override the temptation to observe oneself continuously, one must consciously exert self-control. In light of the strength model of self-control, the latter is draining in terms of cognitive resources (Baumeister & Vohs, 2007) and will eventually take a toll on performance. In fact, neuroscientists agree that humans are not effective multitaskers (Bregman, 2010) and process the stimuli not simultaneously but sequentially at high speed (E. K. Miller & Buschman, 2015). Therefore, chances are that SRI consumption during a videoconference ultimately affects the outcomes.

In this study, we inspect satisfaction with VM process, perceived productivity, and enjoyment as relevant and widely examined outcomes in meeting science (Briggs et al., 2006; Rogelberg et al., 2006; Standaert et al., 2022). Defined as the attendee’s positive affective arousal about the procedures and tools used in a VM (Briggs et al., 2006, p. 588), satisfaction with the VM process represents one of the central outcomes in our research model. This is because individuals care not only about the decisions made in the course of a VM but also about the process of reaching the decisions (Korsgaard et al., 1995). Further, attendees’ perceptions of the productivity of a VM indicate what individual or work-related goals have been achieved – a critical metric of VM success. That is, perceived productivity assesses a VM’s utilitarian value. Lastly, perceived enjoyment reflects the pleasure that participants derive from a VM, capturing VM’s hedonic value.

We deem that resource-depleting SRI consumption results in attendees’ lower satisfaction with VM procedures. Devoting insufficient attention to the online meeting, users may miss some task-relevant information, thus being less productive. Similarly, on the emotional level, users may experience frustration due to their inability to attend effectively to all stimuli at once. For example, in a series of experiments conducted by M. K. Miller et al. (2017), self-view presence in a VM was linked to increased use of anxiety-expressing words, e.g., “worried”, “fearful”, and decreased the use of certainty-expressing, e.g., “always” or “never” (M. K. Miller et al., 2017). Altogether, we assume self-referential information consumption to interfere with VM outcomes negatively. Thus, we hypothesise:

H1:

Self-referential information consumption is negatively associated with VM outcomes, i.e., satisfaction with the VM process (H1a), perceived productivity (H1b), and enjoyment (H1c).

3.2. Effects of communication role

In this study, we further account for the communication roles. Information exchange is at the core of any online meeting, with participants sending (via speaking, sharing, and showing) and receiving communication messages (via listening and viewing). That said, VM participants continuously switch between the speaker and listener roles, as the sender-receiver framework suggests (Lin et al., 2005).

The implications of SRI consumption through the self-view window may vary across communication roles. Sending information (e.g., presenting at a meeting) is an active role (Lin et al., 2005). In contrast, receiving information is generally considered passive (mere listening to others), albeit listening can involve active components, such as attentive processing, analysis, and interpretation (Crespo, 2010; Goyer, 1954).

In a VM, a listener’s “silent” role as an information receiver allows for longer stretches of appearance examination and monitoring. When in this role, self-viewers can dive into the evaluation of mimics and gestures, possibly finding drawbacks in their movements, hairstyle, make-up, or experimenting with posing (Cristel et al., 2020). Research has shown that people listen at about 25% efficiency, on average (Nichols & Stevens, 1957). Moreover, when a message recipient follows his or her self-view instead of a VM flow, attentive, active listening (Figl & Bauer, 2008) will naturally be undermined.

In contrast, for a speaker, it is not as easy to settle into such a self-occupied thinking mode while simultaneously talking or presenting. Besides, knowing that they are the centre of attention, a VM speaker may use SRI strategically, i.e., to enhance the presentation and feel more control over the impression produced. Following this logic, we argue that the adverse effect of SRI consumption in a VM will likely be more pronounced while listening than while speaking. Hence, we hypothesise:

H2:

The negative effect of self-referential information consumption on VM outcomes, i.e., satisfaction with VM process (H2a), perceived productivity (H2b), and enjoyment (H2c), is stronger while listening (i.e., receiving information) than while speaking (i.e., sending information).

3.3. Mediating effects of self-awareness

Another essential argument in this paper is that the negative effect of SRI consumption on VM outcomes takes place through a heightened state of self-awareness. Namely, attending to one’s self-view is expected to induce a state of self-awareness that drains cognitive resources and thereby impedes work processes.

3.3.1. The Role of Self-referential Information Consumption in Inducing Self-awareness

Similar to observing oneself in the mirror, one’s own dynamic reflection on the screen in a VM prompts the state of self-directed attention (Morin, 2004) and acts as a self-awareness trigger (e.g., M. K. Miller et al., 2017). Indeed, by observing themselves via instant video feedback, information senders can see themselves through the eyes of others. This allows them to quickly assess and anticipate how well the content they communicate in a VM comes across, whether their articulation is strong enough and whether their appearance emanates confidence (Duval & Wicklund, 1972). Hence, continuous exposure to the self-view window is likely to heighten attention to self (i.e., self-awareness), especially when the public is present (Duval & Wicklund, 1972), as is usually the case in an online meeting. Therefore, we hypothesise:

H3a:

Self-referential information consumption in a VM while speaking (i.e., sending information) is positively associated with self-awareness.

Even more likely, listeners in an online meeting find themselves occupied with monitoring themselves through self-view, evaluating how they appear in the eyes of others (Duval & Wicklund, 1972). Such self-referential information informs listeners how others in a meeting perceive their appearance, movements, and reactions and whether their ongoing performance corresponds to the ideal they strive to. Hence, we hypothesise:

H3b:

Self-referential information consumption in a VM while listening (i.e., receiving information) is positively associated with self-awareness.

3.3.2. Self-awareness and videoconference outcomes

Induced by SRI consumption, the state of self-awareness presents a cognitive task that competes with the primary task of an online meeting, i.e., information exchange and communication among participants. The initiated process of self-evaluation (“real vs ideal self” comparison) and the attempts to act on the evaluation results, e.g., by fixing hair or a camera angle, may leave VM users with a limited self-control capacity. As suggested by the strength model of self-control, the longer the state of two competing urges lasts, the more fatigued the person becomes (Baumeister & Vohs, 2007). Eventually, the tension of dealing with self-awareness may lead a user to temporarily neglect the flow of an online meeting, if not switching to a different, work-unrelated task. This represents a crucial difference between shallow glances over the self-view window (SRI consumption) and the self-awareness state, suggesting stronger negative effects of self-awareness on VM outcomes than direct effects of SRI consumption.

Past empirical studies outside the work meeting context support the assumed negative potential of self-awareness: Self-awareness most commonly gives rise to self-criticism (Morin, 2011), which is naturally unpleasant. Indeed, a meta-analysis across 79 studies conducted by Fejfar and Hoyle (2000) has uncovered a consistent effect of self-awareness on negative affect. Similarly, a longitudinal four-wave survey over seven years reports a significant association between self-awareness and depressive moods (H. Chen et al., 1998). Following the same line of reasoning, we hypothesise:

H4:

Self-awareness is negatively associated with VM outcomes, i.e., satisfaction with VM process (H4a), perceived productivity (H4b), and enjoyment (H4c).

3.3.3. The mediating role of self-awareness

Together, H3 and H4 suggest a chain of relationships linking SRI consumption with a state of self-awareness (H3), which in turn is hypothesised to be negatively related to VM outcomes, as experienced by participants (H4). Considering (i) the prominent role of self-awareness in social settings (Gonzales & Hancock, 2011), to which VMs belong, and (ii) the reported adverse effects brought up by the state of self-awareness, e.g., resource-draining self-evaluations (Baumeister & Steinhilber, 1984), the unwanted outcomes of self-viewing may partly be attributed to self-awareness processes. In other words, we propose that it is not the SRI consumption per se that leads to adverse VM-related outcomes, such as reduced satisfaction with VM process, lower productivity, and enjoyment. Rather, the self-awareness state evoked by SRI consumption is responsible for the observed unpleasant outcomes. Hence, self-awareness is hypothesised to be a mediating mechanism that gets initiated by self-referential information consumption and may undermine attendees’ experiences.

H5:

Self-awareness mediates the relationship between self-referential information consumption and VM outcomes, i.e., satisfaction with VM process (H5a), perceived productivity (H5b), and enjoyment (H5c).

3.3.4. Control variables

We control for variables that may confound the proposed relationships in our model. VM characteristics, such as the number of participants (e.g., Geimer et al., 2015; Standaert et al., 2022) and duration (a.k.a. length) (Luong & Rogelberg, 2005, Rogelberg et al., 2006; Standaert et al., 2016), may affect an individual’s perceptions of a given VM and its outcomes. Furthermore, organisational behaviour literature found a positive link between job level and overall morale (for review, see Anderson et al. (2015). Thus, since employees higher in their organisation’s hierarchy report higher well-being (Anderson et al., 2015), we expect superiors to be more satisfied with VM processes. Additionally, the perception of the number of VMs as too high – the overabundance of VMs – is expected to negatively affect employees’ evaluations of the meeting (Luong & Rogelberg, 2005). As for personality characteristics, dispositional self-awareness – a chronic individual tendency – is likely to affect how susceptible a person is to becoming self-aware in the presence of the self-view window (e.g., Fenigstein et al., 1975; Kuhn, 2022). Moreover, demographic characteristics, such as age and gender, may influence participants’ technology use and perceptions. Since women are generally more inclined to put effort into impression management, e.g., on social media (Rui & Stefanone, 2013), we expect them to consume more self-referential information during a VM. Besides, older people are expected to be more susceptible to the effects of self-awareness (Beaman et al., 1979).

Figure 1 summarises the hypothesised relationships, including the control variables incorporated in our study.

4. Methodology

To test the proposed relationships, we followed the common approach for technology usage research (e.g., Carillo et al., 2021; Cutler et al., 2021) and surveyed professionals who regularly attend VMs at work. Data collection took place in the Autumn of 2020, during the 2^nd wave of the COVID-19 pandemic, when most European countries introduced the lockdown policy. Given these external constraints, the chosen method represented the most feasible and safest solution.

Measurements from prior research were used to operationalise the research model in Figure 1, with slight modifications to fit the VM context. The exact items’ formulations and sources of origin are provided in Appendix A. To immerse themselves in the context, participants were asked to recall their last VM/online meeting at work and describe it. Because self-referential information consumption may differ for each communication role, we asked to assess separately: “How often did you look at the self-view (video of yourself) window when you were speaking?” and “How often did you look at the self-view (video of yourself) window when you were listening?” on the scale 1= (almost) never, 2=very rarely, 3=rarely, 4=about half the time, 5=often, 6= very often, 7= (almost) always.

To measure the state of self-awareness, we adapted the established scale by Fenigstein et al. (1975),³ explicitly tailoring it to a state. Likert scale response anchors were: 1=strongly disagree, 2=disagree, 3= rather disagree, 4= neither agree nor disagree, 5=rather agree, 6= agree, 7=strongly agree.

To ensure high data quality, we incorporated an attention check into our instrument. Precisely, one item on a 7-point Likert scale asked participants: “please answer ‘often’ for this question”. Further, our study design followed the recommendations to mitigate common method bias (CMB). Specifically, we included the marker variable “blue attitude” (B. K. Miller & Chiodo, 2008, Simmering et al. 2014): “‘I prefer blue to other colours’,” “‘I like the colour blue’,” and “‘I like blue clothes’.” In the final step, the measurement instrument was pretested in a survey with 20 employees. All constructs were modelled as reflective.

An online questionnaire was published on the Prolific Academic platform (Prolific, 2021) in October 2020 and advertised as the “Online meetings at work” poll. We recruited at Prolific Academic because multiple checks have evidenced a broad, naïve and diverse participant pool, reasonable response rate and honest and attentive responses compared to alternatives like Amazon MTurk or Crowdflower (Adams et al., 2020; Peer et al., 2017). Five pre-selection criteria were applied to recruit the sample: 1) fluency in the English language, 2) currently employed full-time or part-time, 3) the participant’s residence is West European (UK, Ireland, Germany, France, Austria, Belgium, Luxembourg, the Netherlands, and Switzerland), 4) at least 50% of the past participations on Prolific Academic platform has been approved, 5) at least 10 previously completed surveys. The estimated duration was 15 minutes; participation was compensated with £1.88 (hourly rate of £7.52).

5. Data analysis and results

5.1. Sample and descriptive statistics

350 respondents filled in the survey, 322 delivered complete answers. 5 records from respondents who reported zero VMs in a typical working week were disregarded (thus, N = 317). After eliminating six participants who failed the attention check, we obtained a net sample of 311 observations for further analysis. Regarding communication type, 65 respondents attended the last VM audio-only, 243 participated with audio and video, 3 people answered “other”. Of those 243 employees, who were in an online meeting with audio and video, 64 had self-view off, and 179 individuals had self-view switched on. Since our research question concerns the self-view feature available only in videoconferencing, for the initial model estimation, we focused on 179 respondents who were involved with audio and video and had their self-view turned on. Later, in the triangulation stage, we also included 64 respondents who were in a videoconference but had their self-view off.

In terms of demographics, 120 (67%) of respondents were female, and the age ranged from 18 to 69 with a mean of 33.25 (median = 30; SD = 10.11). The average meeting size was eight participants (M = 7.62; median = 5; SD = 8.41). A typical meeting lasted about 1 hour (M = 59.93 min; median = 60; SD = 41.24). Our sample mainly consists of middle-level employees (mean job level within the organisation = 2.74; median = 3; SD = 0.92). Regarding the experience with VM tools, 127 (71%) of respondents met regularly in Zoom, 104 (58%) in Microsoft Teams, 42 (23%) in Skype, 12 (7%) in Google Meet, 11 (6%) in Cisco Webex Meeting. Less popular VM tools were Google Hangouts (7 respondents), Bluejeans (3 respondents), GoToMeeting (3 respondents) or others (6 respondents). As for the VM layout, 123 (68.7%) used gallery/grid view, 31 (17.3%) used speaker view, and 19 (10.6%) switched between the layouts during the meeting. The demographics of study participants is elaborated on in Appendix B.

Our respondents reported having been looking at themselves while speaking nearly half of the VM time, on average (M = 3.81, SD = 1.68, median = 4.0), given that we measured SRI consumption on a 7-point Likert scale (1= (almost) never, 4=about half the time, 7= (almost) always). The frequency of looking at self-view while listening was slightly lower (M = 3.20, SD = 1.58, median = 3.0). The difference is statistically significant (t-test: t (354) = 3.54, p = 0.0005 | Mann-Whitney U test: U = 12499.5,

z = −3.44, p = 0.0006). The SRI consumption distribution is examined in online Appendix C.

We checked whether SRI consumption varies across VM layouts: the “active speaker view” (n = 31), “gallery/grid view” (n_S = 122, n_L = 123),⁴ “floating thumbnail window” (n = 4), “pinned own video” (n = 0), “I switched between the layouts during the meeting” (n_S = 18, n_L = 19) or “other”, where participants mentioned screen sharing in an open field (n = 2). Kruskal-Wallis H test suggests mean ranks were insignificantly different between groups, χ²(4) = 2.105, p = 0.717 for SRIC when speaking and χ²(4) = 3.035, p = 0.552 for SRIC when listening.

5.2. Research model estimation

Our research model was tested using the Partial Least Squares (PLS) approach to Structural Equation Modelling in the SmartPLS 3.3 software (Ringle et al., 2014). First, we have chosen the PLS method because it works well with non-normal distributions (Hair et al., 2017), which is often the case in social science research, especially when respondents report their negative perceptions (Maier, Laumer, Eckhardt, 2015, Maier, Laumer, Wirth, 2019). Second, it tolerates using ordinal and binary scales, which is relevant to our study. Finally, our research is explorative in nature, aiming at early-stage theory development and testing. Given the above data and model characteristics, PLS-SEM is appropriate as an estimation method (Hair et al., 2017). We triangulate the PLS-SEM analysis with ANCOVAs, showing robustness to analytical choices.

Model estimation included three stages: exploratory factor analysis of the items, evaluation of the measurement model and the structural model. First, to account for both the new constructs in the research model, obtained from an extensive literature review, and the adaptation of the measurement scales to the online meeting context, we ran exploratory factor analysis for all constructs. Considering the ordinal and non-normal nature of the data, a principal axis factor estimator was used for variance extraction. Due to the fact that in social studies, factors are likely to correlate with each other, promax rotation (an oblique rotation) was chosen (Knekta et al., 2019). Items for satisfaction with the online meeting process, active use, self-awareness as a state, and dispositional self-awareness loaded separately, as theorised.

To examine the factor structure of the self-awareness construct, exploratory factor analysis was conducted for items measuring public self-awareness, private self-awareness, dispositional private self-awareness, and dispositional public self-awareness (online Appendix D). Parallel analysis based on eigenvalues from the principal components and factor analysis, together with theoretical considerations, were used to decide on the appropriate number of factors to retain (Knekta et al., 2019). Consequently, a 2-factor solution was selected: In our sample, we observed a clear distinction between self-awareness as a state and as a trait (i.e., dispositional self-awareness). Within a trait or a state, further distinction between private and public self-awareness appears to be inessential. Hence, in our model, private and public aspects of self-awareness are conceived not as different constructs but as facets of one “self-awareness” construct (Sutton, 2016).

Further, the measurement model quality was evaluated. We assessed indicator reliability, internal consistency, convergent validity and discriminant validity (online Appendix E). Composite reliability, average variance extracted, Cronbach’s α, correlations between reflective constructs, square root of AVE and descriptive statistics are reported in Table 1, and correlations are reported in online Appendix F. Overall, the measurement model evidenced good specification. To check for the common method bias (CMB), a marker-variable analysis was performed, suggesting that this problem is unlikely to be present in the data (online Appendix G).

Table 1. Measurement model. Note: self-awareness (SA), satisfaction with VM process (SAT), dispositional self-awareness (DSA), active use (AU), mean (M), standard deviation (SD), composite reliability (CR), average variance extracted (AVE), Cronbach’s α (CA), correlation values among constructs. Diagonal elements indicate the square root of AVE.

	CA	CR (rho_c)	CR (rho_c)	AVE	SA	SAT	DSA	AU
SA	0.920	0.935	0.924	0.643	0.802
SAT	0.929	0.955	0.930	0.876	−0.241	0.936
DSA	0.827	0.874	0.864	0.542	0.519	−0.085	0.736
AU	0.865	0.902	0.865	0.649	−0.199	0.392	−0.136	0.805
M					4.210	5.580	5.085	4.706
SD					1.689	1.470	1.425	1.328

In the next step, the structural model was assessed using the coefficient of determination (R²), the significance levels of each path coefficient, and the standardised root mean square residual (SRMR). Our model explains 31.8% of the variance in the dependent variable “satisfaction with VM process”, 17.3% in “perceived productivity”, 17.3% in “enjoyment”, and 38.6% in the “self-awareness” construct. As the SRMR reaches a value of 0.085, which is lower than the recommended value of 0.10 (Hu & Bentler, 1999), a good fit can be inferred for our sample. To test the assumed relationships, a bootstrapping with 5,000 iterations was performed. Figure 2 summarises the coefficients and their significance.

Figure 2. Structural Model results (path coefficients and significance levels). Note: in controls 1, the coefficients are displayed for SRIC when speaking (S), SRIC when listening (L) in the respective order. In controls 3, the coefficients are displayed for satisfaction with VM process (SAT), perceived productivity (P), and enjoyment (E) in the respective order. ^† p < 0.1, * p < 0.05, **p < 0.01, ***p < 0.001.

Evaluation of the direct paths “self-referential information consumption (SRIC) → VM outcomes” reveals a nuanced picture. SRIC when listening, is negatively associated with satisfaction with VM process (ß=-0.181, p = 0.036). The associations with perceived productivity (ß=-0.010, p = 0.912) and enjoyment (ß=-0.052, p = 0.552) are insignificant (for listener role, H1b and H1c rejected). Contrary to our expectations, SRIC when speaking is positively linked to satisfaction with VM process (ß = 0.188, p = 0.026) and enjoyment (ß = 0.205, p = 0.010); the association with perceived productivity is insignificant (ß = 0.112, p = 0.190) (for speaker role, H1a, H1b, H1c rejected). Since the negative effect of SRIC on VM outcomes is registered for the listener role only, H2a and H2c is supported. In line with H3a and H3b, SRIC when speaking (ß = 0.162, p = 0.041) and when listening (ß = 0.216, p = 0.002) is positively associated with self-awareness. In turn, higher self-awareness is associated with lower satisfaction with VM process (ß=-0.159, p = 0.014), lower perceived productivity (ß=-0.177, p = 0.015), and lower enjoyment (ß=-0.200, p = 0.010) (H4 supported).

Effect sizes (f²) for the path “SRIC → self-awareness” were small when sending (f² (SRIC_S) = 0.029) and receiving (f² (SRIC_L) = 0.052) a message. We detected a medium effect size for the path “dispositional self-awareness → the state of self-awareness” (f² (DSA) = 0.206), and a small effect size for the path “dispositional self-awareness → self-view engagement when speaking” (f² (DSA) = 0.046) and “dispositional self-awareness → self-view engagement when listening (f² (DSA) = 0.126). Self-view engagement exhibits a small effect size on satisfaction with VM process (f² (SRIC_S) = 0.034 | f² (SRIC_L) = 0.031). Likewise, the effect of self-view engagement when speaking on meeting enjoyment is small (f² (SRIC_S) = 0.034).

To test the hypothesised mediation effect of self-awareness (H5), we followed the approach of Zhao et al. (2010) and used the bootstrapping mediation procedure with 5 000 samples and a 95% bias‐corrected confidence interval (CI) in model 4 of the PROCESS macros by Hayes (2018) to establish the significance of the indirect effect. The results (Table 2) suggest a statistically significant mediation effect of SRIC when speaking on VM outcomes via self-awareness: for satisfaction with VM process, indirect effect = −0.605; standard error = 0.0231; bias‐corrected CI [−0.1116;-0.0201] | for perceived productivity, indirect effect =-0.0362; standard error = 0.0177; bias‐corrected CI [−0.0746;-0.0043] | for enjoyment indirect effect = −0.0594; standard error = 0.0213; bias‐corrected CI [−0.1083;-0.0208]. Mediation of self-awareness is also confirmed for the link between SRIC when listening and VM outcomes: for satisfaction with VM process, indirect effect = −0.0643; standard error = 0.0305; bias‐corrected CI [−0.1271;-0.0083]| for perceived productivity, indirect effect = −0.0482; standard error = 0.0211; bias‐corrected CI [−0.0922;-0.0098]| for enjoyment, indirect effect = −0.0593; standard error = 0.0257; bias‐corrected CI [−0.1125;-0.0100]. Thus, H5 is supported.

Table 2. Mediation analysis results.

Communication role	SRIC →self-awareness → …	Indirect effect	Direct effect	Type of mediation (Zhao et al. 2010)
SRIC during speaking, a.k.a. sending information	→satisfaction	significant	significant	partial (competitive)
	→productivity	significant	insignificant	full
	→enjoyment	significant	significant	partial (competitive)
SRIC during listening, a.k.a. receiving information	→satisfaction	significant	significant	partial (complementary)
	→productivity	significant	insignificant	full
	→enjoyment	significant	insignificant	full

We found that for both communication roles (i.e., speaker and listener) SRI consumption was significantly associated with self-awareness (SAT: b_{SRIC_S→SA} = 0.2995; p < 0.001 | P: b_{SRIC_S→SA} = 0.3131; p < 0.001 | E: b_{SRIC_S→SA} = 0.3131; p < 0.001 || SAT: b_{SRIC_L→SA} = 0.3684; p < 0.001 | P: b_{SRIC_L→SA} = 0.3770; p < 0.001| E: b_{SRIC_L→SA} = 0.3770; p < 0.001), while self-awareness significantly decreased VM outcomes (S: b_SA→SAT = −0.2020; p = 0.0058 | b_SA→P = −0.1157; p = 0.0305 | b_SA→E = −0.1898; p = 0.0029 || R: b_{SA →SAT} = −0.1744; p = 0.0234 | b_{SA →P} = −0.1277; p = 0.0204| b_SA→E = −0.1574; p = 0.0153). In total, our results show that SRI consumption in a VM has a negative indirect effect on VM outcomes in terms of satisfaction with VM process, perceived productivity, and enjoyment via self-awareness.

Finally, we examined the impact of control variables. Females are more frequently engaged with SRI consumption when sending information (ß = 0.143, p = 0.045). There are no gender differences in SRI consumption when receiving information (ß = 0.083, p = 0.248). Younger people tend to engage with SRI consumption when sending information more often (ß=-0.197, p = 0.006). Naturally, the higher the dispositional self-awareness, the more frequently a person engages with self-view when speaking (ß = 0.207, p = 0.005) and listening (ß = 0.342, p < 0.000), and the higher the state of self-awareness experienced during the last VM (ß = 0.397, p < 0.000). A higher number of participants in a VM is significantly positively associated with self-awareness (ß = 0.106, p = 0.026) but is insignificantly related to meeting outcomes in terms of satisfaction with VM process (ß=-0.135, p = 0.164), perceived productivity (ß=-0.142, p = 0.100) and enjoyment (ß = 0.031, p = 0.695). Active use is significantly positively linked to satisfaction with VM process (ß = 0.332, p < 0.000), perceived productivity (ß = 0.309, p < 0.000), and enjoyment (ß = 0.313, p < 0.000); in contrast, for job level and length of a meeting, the respective links are insignificant. Perceived overabundance of VMs is associated with lower satisfaction with VM process (ß=-0.212, p = 0.006).

5.3. Triangulation

In our analysis above, SRI consumption was measured through self-report. In view of potential biases, we triangulated our findings by checking if the results are robust to alternative approaches to measurement. We contrasted users with the turned-off self-view to users who reported high engagement with the self-view, i.e., scoring at least 5 on the 7-point Likert scale, i.e., looking at themselves often, very often, or (almost) always during a VM. One-way ANCOVAs with Bonferroni correction revealed that there was a significant difference in mean self-awareness for high vs. zero SRI consumption groups while speaking (F(1,88) = 14.37, p = 0.008, η²_p = 0.08) and listening (F(1,58) = 6.84, p = 0.01, η²_p = 0.11). In turn, self-awareness exhibited a significant negative association with satisfaction with VM process (ρ= −0.21, p = 0.001), productivity (ρ= −0.16, p = 0.012), and enjoyment (ρ= −0.17, p = 0.008). As for direct effects of SRI consumption on VM outcomes, the group with high SRI consumption while speaking reported better outcomes than the group with zero SRI consumption, and this difference was statistically significant (satisfaction: F(1,87) = 5.26, p = 0.02 | productivity: F(1,87) = 3.17, p = 0.08 | enjoyment: F(1,87) = 6.73, p = 0.01). Altogether, comparing users with zero SRI consumption (i.e., the turned-off self-view) to users with high SRI consumption generally corroborated the identified patterns of the main study (see online Appendix H for details).

5.4. Additional analyses: dispositions and VM type

Beyond assessing the primary research model and hypotheses (Figure 1), we performed additional analyses. First, to better understand the role of individual differences in SRI consumption in a VM, we examined dispositional self-awareness. We observe that VM participants who reported higher C speaking, as assessed by visual inspection of boxplots (Figure I1) and statistically significant positive partial correlation between age and SRI consumption (r = 0.19, p = 0.013), controlling for age and gender. For the listener role, the association between dispositional self-awareness and SRI consumption is even stronger (r = 0.32, p < 0.001), given the same controls (see online Appendix I for details).

Second, in the business world, it is common to differentiate between types of meetings to make them purposeful (MeetingSift, 2020). We checked the influence of VM type post-hoc based on the activities performed in a VM such as status update, information sharing, decision-making, innovation, and team building (MeetingSift, 2020). In VMs, involving innovation, SRI consumption was significantly higher (SRIC speaking: t(177)=-2.45, p = 0.02 | SRIC listening: t(177), p = 0.06). In VMs, involving information sharing, participants consumed slightly less SRI when listening (t(177) = 2.40, p = 0.02). The relationships among all primary constructs remained unchanged, hinting that core model results remain unaffected by particular VM activities (see online Appendix J for details).

Table 3 presents a summary of our findings.

Table 3. Summary of findings.

Focus	Major findings
Describing the phenomenon SRI consumption and its salience in a VM	SRI consumption (a.k.a. looking at self-view) when speaking occupies nearly half of the VM time, on average. SRI consumption when listening is slightly lower.
Direct effects of SRI consumption on VM outcomes	Direct positive effect of SRI consumption when speaking on satisfaction with VM process and enjoyment. Direct negative effect of SRI consumption when listening on satisfaction with VM process. Insignificant direct effect of SRI consumption both when speaking and listening on perceived productivity.
Distinguishing speaker vs listener role	Communication role matters. The direct effects of SRI consumption are reversed from negative to positive with the role change. The association between SRI consumption and self-awareness is stronger for the listener than for the speaker role.
Indirect effect (mediation) of SRI consumption on VM outcomes through the increased state of self-awareness	For both communication roles, self-awareness is a mediator in the relationships:
	Supplementary findings from additional analyses
Individual differences in SRI consumption	Females consume more SRI than males only when speaking. For the listener role, gender differences are insignificant. Dispositional self-awareness increases SRI consumption for both speaker and listener roles.
Testing the influence of VM layout on SRI consumption	When the self-view is on, a VM layout (e.g., active speaker view, gallery/grid view, floating thumbnail window, and switching between the layouts) is associated with insignificant differences in SRI consumption.
Testing the influence of VM type on SRI consumption	SRI consumption does not vary significantly between VMs, involving status update, decision making, and team building and VMs not involving these activities. In information-sharing VMs (e.g., presentations, panel debates, keynotes, lectures), SRI when listening is slightly lower compared to VMs, not focused on information sharing. In VMs, involving innovation (e.g., thinking outside the box by brainstorming, associating, and sharing ideas in a broad scope), SRI consumption is higher for both roles, as compared to VMs without innovation activities.

6. Discussion, contributions, and future research

The present study takes a granular look at the distinctive feature of VMs – the self-view – and examines the implications of SRI consumption for VM outcomes. We hypothesised that higher SRI consumption is linked to worse VM outcomes (H1), with effects varying across communication roles (Lin et al., 2005), that is, for speakers vs. listeners (H2). On top of that, we speculated that SRI consumption heightens self-awareness (H3), which, in turn, inhibits VM outcomes (H4). Thus, self-awareness was proposed as a mediator (H5). Conceptual propositions (Figure 1) were tested in the sample of West European employees who regularly attended VMs at work.

6.1. Consequences of SRI consumption in a VM at work

Responding to the first research question (RQ1: How is self-referential information consumption related to videoconference outcomes at work?), we show that the exact answer depends on the communication role. Looking at self while listening to others hampers work-related outcomes (see Figure 2 and Table 2). We observe a direct negative effect of SRI consumption when listening on satisfaction with VM process (H1a) and indirect negative effects on all VM outcomes tested, namely satisfaction with the VM process, perceived productivity, and enjoyment. Contrary to H1b and H1c, the direct effects of SRI consumption while listening on perceived productivity and enjoyment were insignificant.

Noteworthy, looking at self while speaking produced contrary effects. Contrary to H1, the direct effects of SRI consumption while speaking on satisfaction with the VM process and enjoyment are positive (see Figure 2). Simultaneously, the indirect effect of SRI consumption while speaking, via heightened self-awareness, undermines the VM outcomes, as per H3-H5 (see Table 2). The alternative analytical strategy of contrasting VM attendees with zero and high SRI consumption did not change the results, pointing to their robustness. Overall, the direct effects of SRI consumption vary with the communication roles.

6.2. Mediation through self-awareness

Our investigation of the second research question about self-awareness (RQ2: What is the role of self-awareness in this relationship?) provides evidence that SRI consumption induces a state of heightened self-awareness (as per H3) for both communication roles (speaker and listener). However, the effect’s magnitude is higher for the listener role. The state of heightened self-awareness, in turn, is negatively associated with all three examined outcomes regardless of the communication role, as we hypothesised in H4. Hence, the state of heightened self-awareness appears to be the “thief” of joy and productivity in VMs at work, not self-viewing per se. Speakers look at themselves more. Yet, listeners’ SRI consumption is cognitively more depleting due to its stronger association with the self-awareness state, which is destructive for VM outcomes.

6.3. Individual differences, VM layout and type

The additional analysis revealed that differences in VM layouts that determine the size and position of the self-view window are neglectable. In other words, when self-view is turned on, participants attend to it. As for VM types, the status update, decision-making, and team-building meeting types are insignificant factors for attendees’ SRI consumption. In contrast to them, in VMs with an innovation component – e.g., thinking outside the box by brainstorming, associating, and generally sharing ideas – attendees show higher SRI consumption regardless of the role. In information-sharing VMs (e.g., presentations, panel debates, keynotes, lectures), SRI consumption when listening was slightly higher than in other VMs.

Turning to individual characteristics, gender differences are pronounced only for the speaker role: Female employees look at themselves more often than male employees when speaking but not when listening. Furthermore, individuals with higher dispositional self-awareness report higher SRI consumption. A tendency to scrutinise own feelings and image makes people more prone to checking how one appears to others in a VM.

6.4. Limitations and opportunities for further triangulation

Before discussing our findings for the academic community and practitioners, we acknowledge study limitations and list triangulation opportunities for our research model. First, our survey was administered to West European employees, which leaves room for checking the implications of SRI consumption in other cultures. Prior research on the mirror effects informs, for example, that a mirror does not affect Japanese participants, while North American participants show higher self-criticism and lower cheating rates in the presence of a mirror than those without a mirror (Heine et al., 2008). Second, we recruited participants on a survey platform, thus having people initially willing to be surveyed for a small compensation. This approach, however, allowed us to reach a heterogeneous sample, not restricted to a particular organisation. Third, we captured subjective employees’ experiences of VMs. Future research can use the advantage of a laboratory setting to test objective performance measurements like speed and decision accuracy on a selected task. Besides, following George et al. (2022) exploration, SRI consumption can be precisely captured with eye-tracking equipment. Lastly, our data was collected at a single point in time. Hence, we test associations in this paper, and our causal explanations are rooted in our theorising. Longitudinal studies (Shockley et al., 2021) may test if the revealed effects hold in repeated observations and follow the expected temporal precedence. On top of that, rich qualitative data can help elucidate the human experience of being continuously exposed to their own image through technology.

6.5. Research contributions and future research outlook

Below we outline the specific paper’s contributions and envision further research on VMs and beyond.

SRI Consumption Model

We extend IS research on VMs (e.g., Toney et al., 2021, Regenbrecht & Langlotz, 2015; Riedl, 2022) by conceptualising SRI consumption. Diving into the underexplored VM peculiarities, this paper introduces the notion of SRI consumption as behaviour primed by VM technology. External circumstances, namely the pandemic, triggered an unprecedented transition to remote collaboration and intensified SRI consumption, which resulted in new experiences and significant outcomes for unprepared users. We propose fitting conceptual lenses (i.e., by combining the strength model of self-control, objective self-awareness theory, and sender-receiver framework) to model SRI consumption implications in the work-related context. In doing so, we explicate and empirically validate Bailenson’s (2021, p. 4) “all-day mirror” theoretical argument that the Zoom interface will likely lead to psychological consequences.

Moreover, drawing on meeting science research (Briggs et al., 2006; Rogelberg et al., 2006; Standaert et al., 2022), our proposed research model explains relevant and understudied outcomes such as satisfaction with VM process, perceived productivity, and enjoyment. Thus, we broaden the extant body of knowledge, which previously revolved mainly around Zoom fatigue (Fauville et al., 2021; for review, see Wang & Prester, 2022). Furthermore, in the face of arguments that view self-view as largely detrimental (Fauville et al., 2021; Riedl, 2022), our study advocates for a nuanced picture and is the first to uncover the complexity of its effects. Following our results, the effects’ valence (positive, negative, or neutral) depends on whether one’s engagement with the self-view happens when listening or when speaking in a call. The mechanism of self-awareness activation sheds additional light on how the distinct effects unfold.

Future works can extend our model by theorising the self-view window as a source of not only techno-distress but also techno-eustress, the type of stress that leads to positive outcomes (Tarafdar et al., 2017). In doing so, the objective self-awareness theory might serve as a theoretical basis explaining when interactions with the self-view lead to their appraisals as “bad” stress and when – as challenging “good” stress (Tarafdar et al., 2017). More specifically, when in the state of heightened self-awareness, a VM attendee perceives the “real” self as close to the “ideal” self, the self-view interaction might more likely be appraised as “good” stress.

E-Collaboration

This paper contributes to the long-standing IS literature on e-collaboration (Ahuja et al., 2020; Dennis et al., 1988; Riemer et al., 2009) and specifically to the thematic clusters “group’s production function” and “group and member well-being” according to the Ahuja et al. (2020)’s classification. In doing so, we join the current debate on digital transformation in the workplace (Eckhardt et al., 2019) and the future of work (Standaert et al., 2022; Weritz et al., 2022). While VMs support workers’ feelings of togetherness (Hacker et al., 2022; Waizenegger et al., 2020) and help sustain productivity (Hafermalz & Riemer, 2021), they also give rise to adverse states like heightened self-awareness. Following our findings, heightened self-awareness – a by-product of intensive VM use by uninformed staff – hits productivity, satisfaction with VM process, and enjoyment. Our data collected during the early pandemic vividly snapshots a forced, unprepared switch to virtualness – a scenario opposite to the anticipated and desired gradual transition (Eckhardt et al., 2019).

Looking forward, the possible habituation effects are intriguing. VMs have been serving organisations for several years since the COVID-19 outbreak and lockdowns. Hence, one could speculate that most employees have gotten used to constant mirrors in front of their eyes by now. Such adaptation could reduce the triggering capacity of SRI consumption regarding the self-awareness state. We surmise, however, that while such an effect may vary in strength, it is unlikely to disappear completely. Post-pandemic data collection and its benchmarking to our results are needed to illustrate if and how users evolve concerning the SRI consumption in a VM phenomenon. In other words, future studies could ask: Do employees care less about how they look after staying tuned in VMs for over 2 years? Do they become less self-aware being exposed to constant streaming of their own video? Or do they consciously turn the self-view off to avoid a heightened self-awareness state?

Furthermore, examining the spill over effects of self-viewing from work to private life represents another fruitful direction for research. In our sample collected in October 2020 in Europe, only 10% (32 of 311) worked outside their home, while 47% (146 of 311) worked from home all the time and 43% (133 of 311) some days a week. Given a high share of people joining VMs from home, it is worth exploring whether (and if so, how) the self-view-driven effects stretch to the nonwork domain, as previously documented by A. Chen and Karahanna (2018) and Benlian (2020).

User Response to Technology & Boundary Conditions

On a broader level, we add to the knowledge base about the cognitive and affective responses to the technology (Agogo & Hess, 2018; Maier et al., 2019). We show that VMs and interactions with the self-view are not problematic technology‐induced stressors per se. Our results highlight the importance of identifying boundary conditions that ultimately determine the effects’ magnitude and valence (positive or negative). In particular, we demonstrate that in VMs, the scales are ultimately tipped in favour of speakers/presenters: we found positive direct effects of SRI consumption when speaking on satisfaction with the VM process and enjoyment. Possibly, watching oneself while speaking compensates for the lack of audience cues (e.g., silence in the room, mutual direct gaze as typical signals of focus and receptivity) and gives speakers a sense of control over their appearance. As such, the positive feelings even out or outweigh the adverseness caused by heightened self-awareness.

As the next step, we envision studying and bringing up the positive outcomes of technology use. Now that we have brought communication roles to the forefront of the discussion about the VM outcomes, follow-up investigations might integrate role switching as a dynamic process. While the present study captures VM attendees’ experience during speaking and listening separately, future research might strive for higher external validity because participants often transfer between roles in a VM.

Beyond IS

We extend research on the effects of self-awareness in the work context. Self-awareness is considered a positive characteristic of managers (Eurich, 2017): seeing oneself and one’s own aspirations, reactions, and fitting to the environment is related to higher satisfaction with the job and relationships, as well as overall happiness (Eurich, 2017). However, while the trait of self-awareness is assumed to be beneficial, our results show that the state of heightened self-awareness can be unpleasant and even impede short-term outcomes, like in the case of VMs. Hence, we illuminate the differences between the state and trait self-awareness.

6.6. Practical contributions

Our findings pass a cautionary message for videoconferencing technology users at the workplace. Especially the participants in a listener role, who stay quiet and do not get the spotlight most of the VM time, are at risk of negative consequences. Being a passive audience makes them prone to continuous SRI consumption and a heightened state of self-awareness. We advise differentiating between glancing at oneself and being in a heightened self-awareness state. The revealed duality of SRI effects suggests that while SRI consumption can bring up positive effects, the heightened self-awareness state is a “thief” of joy and productivity. That is VM attendees – especially speakers – can and should use self-view to their advantage but avoid the state of self-awareness.

Meeting hosts and project managers may be puzzled or even upset after reading that watching their own appearance occupied nearly half of the VM time among participants in our sample (online Appendix C). For example, the longer a team leader talks, the longer others remain in the listener role, and the more probable it is to be deeply self-monitoring, fizzling out to “present absence” in a VM. Consequently, a significant portion of work-related content, even though communicated with great effort, may not reach the message addressee, vanishing into thin air. Our results stress the importance of promoting active participation in a VM. By asking seemingly inactive participants to share an opinion or answer a question, a VM host can take them back to the VM agenda, thereby breaking the self-scrutinising process and fostering their inclusion. This resonates with the earlier recommendations for team leaders to execute VMs in a more explicit manner than offline meetings (Beranek et al., 2005).

On a general note, we hope our results nudge team leaders towards a wiser organisation of VMs (Standaert et al., 2022). Since entering a virtual space corresponds to nearly zero transaction costs, the number of VMs has been increasing in recent years. Not surprisingly, the feeling of VM overabundance is common: 31% (97 out of 311) of our respondents rated the number of online work meetings that they participated in as “barely too many”, “too many”, or “much too many”. For passive listeners – a likely role for employees appraising a VM as redundant – the disadvantages stretch to becoming highly self-aware. Prosocially oriented managers may help to avoid this pain by limiting the VM quantity (“Do we need this VM?”) and by carefully selecting attendees (“Whom should I invite to this VM?”). Besides, learning from our analysis of VM types (online Appendix J), team leaders might consider integrating team building (e.g., kick-off, team-building outings, and corporate events) to make VMs more enjoyable. Lastly, when onboarding less experienced staff, managers may consider advising their new colleagues about the options: (a) to work around the default layout (e.g., decreasing the self-view window size, placing the self-view window in the corner of the screen, etc.) or (b) to use the feature in the respective VM tool (Zoom, Google Meets, Skype, etc.) mainly when speaking.

For providers of videoconferencing tools, our paper outlines a knotty situation: our empirical findings rather pose challenges for good solution design than give straight answers. The default settings now include the self-view window in VMs with the video on. It exposes VM attendees to a trade-off: besides the benefit of informing how one appears to conversational partners, the self-view feature can also impair users’ focus and, thereby, work-related outcomes. Disabling the self-view feature – an intervention that lies on the surface – will eliminate all its effects at once. This measure means getting rid of a control tool (over one’s background, camera angle, and overall visual presentation). Some users may choose to leave the self-view active, as they are mostly satisfied with the trade-off of the afforded benefits and challenges. Informed by our findings, VM software providers can reconsider the default settings or update the design so that the self-view window can be easily customised or pop up for speakers only.

7. Conclusion

This paper focuses on users’ engagement with the self-view feature – a phenomenon we refer to as “self-referential information consumption” – and its effects on users’ perceptions of a diverse set of meeting outcomes. Facing the real-time self-video, online meeting attendees are occupied with self-referential information coming from a self-view window and, consequently, get into a state of heightened self-awareness. The individual‘s self-regulation efforts towards keeping attention on business issues in the presence of such a powerful distractor as own reflection on the screen are high. Such efforts deplete cognitive capacity and ultimately undermine work-related outcomes, i.e., satisfaction with the meeting process, perceived productivity, and meeting enjoyment. This is especially salient for the listener role. However, looking at self when speaking can be neutral (in terms of perceived productivity) or even advantageous for a participant (in terms of higher satisfaction with the online meeting process and enjoyment). Shedding light on when and how self-referential information consumption can be advantageous or disadvantageous serves as an impetus to designing and using technology solutions that make the best of users’ limited attentional resources.

Acknowledgement

We thank three anonymous reviewers, the AE and the SE, Iris Junglas, for their guidance in improving the manuscript. We also acknowledge the feedback provided by Jason Thatcher during his stay at the Weizenbaum Institute in summer 2022. An earlier version of this paper was presented at the International Conference on Information Systems (ICIS) in December 2021, Austin, Texas, USA (Abramova, Gladkaya, Krasnova, 2021). The authors are grateful to the ICIS reviewers, editors, and session attendees, as well as to the participants of the AMJ paper development workshop in Amsterdam, the Netherlands, 30 June-1 July 2022, for their comments.

Disclosure statement

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

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/0960085X.2024.2325350.

Additional information

Funding

This work was supported by the Federal Ministry of Education and Research of Germany (BMBF) [grant no. 16DII127, 16DII131 (“Deutsches Internet-Institut”)].

Notes

1. The terms “virtual meeting” and “online meeting” are used interchangeably in this paper.

2. The term “Zoom” is used here to account for the product’s popularity. Our narrative is related to all videoconferencing tools that support a self-view feature, including products like Google Meets, Microsoft Teams, Skype, Webex, etc., and in-house solutions.

3. The original instrument was designed for the trait of self-consciousness and additionally distinguished between private self-consciousness (attention to one’s inner thoughts and feelings) and public self-consciousness (general awareness of the self as a social object that has an effect on others). For a state of self-awareness, the private-public distinction is inconclusive. In our theorising, we follow Sutton (2016), who argues that the attributes “private” and “public” are not distinct constructs, but dimensions of a “self-awareness” construct. Our measurement instrument includes items from both dimensions to allow for empirical clarification of the relationship during factor analysis.

4. Due to two missing values in SRI when speaking, we report the sample sizes separately.