Why people engage in supplemental work: The role of technology, response expectations, and communication persistence

Summary Supported by various collaboration technologies that allow communication from any place or time, employees increasingly engage in technology-assisted supplemental work (TASW). Challenges associated with managing work and nonwork time have been further complicated by a global pandemic that has altered traditional work patterns and locations. To date, studies applying a TASW framework have focused mainly on individual uses of technology or connectivity behaviors and not considered the potential team and social pressures underlying these processes. This study provides clarity on the differences between technology use and TASW and sheds light on the drivers of TASW in a work environment characterized by high connectivity and diverse team structures. Specifically, we demonstrate how individual, social, and material pressures concomitantly impact individual work practices in a team context. Drawing on multisource and multilevel data provided by 443 employees nested in 122 teams, this study shows that individual collaboration technology use and team-level response expectations are independently contributing to TASW. Though the persistence of communication afforded by collaboration technologies mitigates the impact of collaboration technology use on TASW, this persistence is not found to impact the relationship between team-level response expectations and TASW. We discuss how these findings inform our understanding of TASW.


Summary
Supported by various collaboration technologies that allow communication from any place or time, employees increasingly engage in technology-assisted supplemental work (TASW). Challenges associated with managing work and nonwork time have been further complicated by a global pandemic that has altered traditional work patterns and locations. To date, studies applying a TASW framework have focused mainly on individual uses of technology or connectivity behaviors and not considered the potential team and social pressures underlying these processes. This study provides clarity on the differences between technology use and TASW and sheds light on the drivers of TASW in a work environment characterized by high connectivity and diverse team structures. Specifically, we demonstrate how individual, social, and material pressures concomitantly impact individual work practices in a team context. Drawing on multisource and multilevel data provided by 443 employees nested in 122 teams, this study shows that individual collaboration technology use and team-level response expectations are independently contributing to TASW. Though the persistence of communication afforded by collaboration technologies mitigates the impact of collaboration technology use on TASW, this persistence is not found to impact the relationship between team-level response expectations and TASW. We discuss how these findings inform our understanding of TASW.

K E Y W O R D S
collaboration technologies, communication persistence, response expectations, team structure, technology-assisted supplemental work 1 | INTRODUCTION Employees around the world experienced an abrupt transition in their work roles due to the COVID-19 pandemic (Vaziri et al., 2020). For many workers, this involved a reconfiguration of the boundaries between work and nonwork (Fisher et al., 2020) as workers began teleworking almost overnight, in some cases for the first time (Kramer & Kramer, 2020). These shifts likely increased challenges for individual workers to manage work and nonwork time (Allen et al., 2021) whereas pressures to work during evenings, nights, or on weekends intensified. For instance, as work time becomes more porous and individuals may attend to house chores, home schooling, or other social activities between work meetings, individuals may choose to sacrifice sleep hours, nights, or early mornings to meet work demands (Xiao et al., 2021). In addition, Chong et al. (2020) noted that the interdependency of work may generate greater task setbacks under conditions of forced telework during the pandemic, which may require workers to engage in supplemental work to maintain individual and team performance levels. However, technologyassisted supplemental work (TASW) has been steadily on the rise over the past decades and may contribute to a host of negative outcomes (Eichberger et al., 2020) the ways individual, social, and technological factors independently or conjointly affect these work practices are largely unknown. Hence, the current study builds on the TASW framework to examine how different factors present in contemporary work environments may exert pressure to engage in TASW.
Information and communication technologies (ICTs) differ from industrial or production technologies in that they not only facilitate new work practices themselves, but in doing so produce information that can be accessed by others and acted upon (Zuboff, 1988). Over the past several decades, scholars have noted that as ICTs alter the ways that information is presented and made available within organizations, they support new structures and ways of organizing (Barley, 1986;Volkoff & Strong, 2013;Zammuto et al., 2007). Specifically, the use of ICTs can facilitate a variety of organizational changes including altering advice networks (Leonardi, 2007), ways of learning (Garud & Kumaraswamy, 2005), patterns of knowledge production (Schultze & Boland, 2000) and individual work routines (Pentland & Feldman, 2008). A particularly notable and influential aspect of contemporary ICTs is that they afford workers greater forms of connectivity with individuals, teams, and organizations (e.g., Nurmi & Hinds, 2020;Wajcman & Rose, 2011). Through the use of mobile phones, personal computers, and wireless networks, workers can interact with each other without the constraints of time or need for co-location. This malleability and ubiquity of ICTs also means that employees increasingly work extended time outside the office-at home beyond regular working hours, at night, or on weekends (Dery & MacCormick, 2012;Fenner & Renn, 2010;Wajcman & Rose, 2011).
For the present study, we are concerned with TASW defined as distributed work practices performed after hours, often discretionary and not covered by a formal contract or compensation, and accomplished through ICTs such as laptops or other mobile devices (Arlinghaus & Nachreiner, 2014;Fenner & Renn, 2004Ojala, 2011). Though certain work characteristics may impose demands for extending one's connectivity to work or working during nontraditional time periods, less is known about the particular mechanisms-individual, social, and material-driving how and why employees engage in productive work behavior, collaborate, and complete substantial work tasks using these technologies outside of work hours. Investigating how differing work conditions operate as pressures underlying the performance of TASW makes several contributions to organizational scholarship.
First, this work contributes to an understanding of how the TASW framework operates in a work environment characterized by high connectivity and diverse team structures. Previous studies referencing the TASW framework typically operationalize TASW as technology use after hours (e.g., Arlinghaus & Nachreiner, 2014;Barber & Jenkins, 2014;Boswell & Olson-Buchanan, 2007;Chen & Karahanna, 2014;Day et al., 2012;Derks & Bakker, 2014;Diaz et al., 2012;Ohly & Latour, 2014;Olson-Buchanan & Boswell, 2006;Park et al., 2011;Wajcman et al., 2010;Wright et al., 2014). Notably, most of these studies implicitly reference TASW while actually capturing the frequency (e.g., Boswell & Olson-Buchanan, 2007;Park et al., 2011) extent (e.g., Diaz et al., 2012, duration (e.g., Wright et al., 2014), or timing (Richardson & Benbunan-Fich, 2011) of ICT use after hours, failing to address the extent to which these technologies are used to actually perform work and complete work-related tasks outside regular work hours (Fenner & Renn, 2010). Another concern is that previous work mostly covers extended availability and connectivity (Dery et al., 2014;Mazmanian et al., 2013;Thörel et al., 2020) rather than supplemental work practices. Though escalating connectivity to work is becoming increasingly common, contemporary work may present connectivity demands (Nurmi & Hinds, 2020) that require employees to go beyond merely signaling availability or monitoring email messages (Thörel et al., 2020) and engage in more substantial work tasks after hours (Gadeyne et al., 2018). Hence, although there is a large body of literature referencing the TASW framework, the links are primarily implicit, placing ICT use and TASW, or TASW and extended availability or connectivity on equal footing. To provide conceptual clarity, this study makes an important distinction between these concepts that can contribute to theory regarding why individuals engage in supplemental work and how particular aspects of ICT use contribute to these behaviors.
Second, our investigation of the various mechanisms present in the relationship between ICT use and TASW seeks to extend our understanding of how organizational dynamics operate in a context of interdependent work, and without the boundaries of time and location. Hence, we examine a specific branch of ICTs-namely, collaboration technologies. We use the term collaboration technology to refer to a specific set of cloud-based software platforms aimed at supporting collaboration and communication among participants as well as facilitating information processing and accessibility (Dennis et al., 2003;DeSanctis & Gallupe, 1987). Collaboration technologies differ from other organizational technologies that offer possibilities for ubiquitous connectivity among workers and teams (i.e., phones, email) in that the (a) use of the technology is potentially visible over time and accessible to third parties not initially relevant (Treem et al., 2020) and (b) use of the technology can be related to an individual task, be interdependent with the work of others, or move fluidly between these states. Scholars have recognized that the materiality of collaboration technologies that makes information visible in new ways can alter modes of working. For instance, Leonardi (2007) studied the use of a shared IT system by computer technicians and found that the ability to view the activity of others led to changes in who workers asked for task advice. Similarly, Dery et al. (2006) studied the use of enterprise resource planning systems by bank managers. The authors concluded that while the material nature of the technological artifact required users to have specialized knowledge of how to enter data, other nonmaterial factors including the time available to use and learn the system contributed to limited-use practices. Though scholars have considered ways that the materiality of collaboration technologies might facilitate new forms of work within traditional work settings and roles (e.g., Jasperson et al., 2005), less attention has been paid to the extent to which interdependent work facilitated by collaboration technologies might alter the boundaries of work itself and contribute to TASW.
Third, by examining distinct aspects associated with the utilization of collaboration technologies by workers embedded in teams, this work examines the relative extent to which those differences are driven by factors that are or are not under the control of the individual worker. Scholarship on the role of collaboration technologies in organizations has demonstrated that although they provide opportunities for continuous connectivity, they are utilized by workers to manage and regulate when and how they connect to work (Gibbs et al., 2013). When this work takes place in an interdependent team environment, workers may feel obligated to respond to communication from other team members, and these obligations may erode some of the control individuals have over work behaviors (Mazmanian et al., 2005). Because collaboration technologies make communication visible to other employees in ways that are different than other ICTs offering connectivity (i.e., email and phone) they may create different pressures regarding supplemental work (Leonardi & Vaast, 2017). By focusing on the individual, social, and material aspects of work as potentially competing or complementary in their relationship to TASW, this work is consistent with calls to examine the ways technology use in organizations presents possibilities for action, but is not deterministic in its effects (Cecez-Kecmanovic et al., 2014;Leonardi & Barley, 2008).

| Collaboration technology use and TASW
The notion that increased ICT use can escalate one's commitment or connectivity to work is well documented in the literature (Dery & MacCormick, 2012;Kolb et al., 2012;Matusik & Mickel, 2011;Richardson & Benbunan-Fich, 2011;Wajcman & Rose, 2011;Wright et al., 2014). Research suggests that workers whose activities require continual coordination with colleagues, clients, or supervisors take a predominantly positive attitude toward technology use, while acknowledging that technologies may be accompanied by work activities encroaching upon the private sphere (Cavazotte et al., 2014).
Employees seek to use technologies that will facilitate more efficient collaboration and information flows across spatial and temporal boundaries, but are wary of the communication that invariably accompanies this level of connectivity.
Organizations are increasingly operating across geographical borders (and time zones) and are implementing new collaboration technologies creating demands for TASW (Golden & Raghuram, 2010;Nurmi & Hinds, 2020;Piszczek, 2017;Thörel et al., 2020). Employees have been found to reciprocate the distribution of mobile technology by using these technologies to extend their connectivity (Richardson & Benbunan-Fich, 2011). Conversely, when these technologies are used frequently and intensely, employees will have more possibilities to engage in supplemental work, as there are no technological barriers preventing these practices (Venkatesh & Vitalari, 1992). The use of ICTs is sometimes referred to as the electronic leash, tethering employees to work (Büchler et al., 2020;Richardson & Thompson, 2012;Schlachter et al., 2018) and encroaching into individuals' personal lives and nonwork time (Schlachter et al., 2018).
Employees may feel the obligation to utilize the available options technologies offer and engage in more substantive supplemental work behaviors when collaboration technologies are used across spatial and temporal boundaries. Collaboration technologies, such as Google Workspace or Microsoft 365, include various applications that enable distributed coworkers to share files, edit documents individually or collectively, and collaborate synchronously through video and conference calls. These technologies are typically aimed at collaboration and productivity and, as such, require more attention or effort from their users (Robey et al., 2000) than technologies that are more focused on facilitating connectivity (Gadeyne et al., 2018). Focusing on collaboration technologies foregrounds the potential interdependence among organizational members when completing tasks. In work contexts mere connectivity may prove to be inadequate as collaboration technologies use may-implicitly or explicitly-require more commitment, contributions, or engagement from users, leading to TASW. Hence, we hypothesize the following: H1. Collaboration technology use is positively related to TASW.

| Response expectations and TASW
Many studies have argued that employers and employees may develop responsiveness expectations that shape how technologies are used and may escalate employees' connectivity to work (Derks et al., 2015;Mazmanian et al., 2013). The shared expectations about responsiveness within a team may mold a social norm regarding connectivity after hours (Derks et al., 2015). Hence, at a team-level response expectations may normalize into a social pressure to remain available and accountable to others after hours. More broadly, questions of if, when, and how to connect to work are situated in the age-old antinomy of individual and technical agency versus normative social pressures (Leonardi & Barley, 2010;Orlikowski, 1992). Thus, it can be argued that both the material features (casu quo; technical systems; Leung & Wang, 2015) as well as the social practices and expectations (casu quo; social systems; Leung & Wang, 2015) in workplaces play a role in employees' decisions to engage in TASW. These team-level response expectations, such as whether other team members or supervisors expect a response to work-related messages during nonwork hours could be an important driver in TASW (Fenner & Renn, 2010). Specifically, workers will look at the behaviors and communication of influential van ZOONEN ET AL. peers and organizational members to gain insights into appropriate ways that ICTs should be used within a context (Fulk, 1993;Schmitz & Fulk, 1991). Important referents that influence employees' behaviors and choices tend to be (a) people with whom employees frequently communicate, (b) those in similar roles, and (c) those who occupy a high(er)-status position (Boh & Wong, 2015;Friedkin, 1993;Shah et al., 2006). Prior research showed that managers and team members are two dominant social relationships that influence how employees perceive their work environment (Tierney, 1999). Hence, supervisors and team members are two key referent groups that can potentially influence expectations about responsiveness.
Team-level response expectations refer to the shared beliefs regarding appropriate levels of responsiveness within the team. Such shared expectations or norms may exert a powerful form of social control and direct individual behavior within social groups (Barker, 1993;Eby & Dobbins, 1997;Taggar & Ellis, 2007). Several scholars have directed attention to how individual-level cognitions about response expectations may contribute to an environment that expects workers to be constantly connected to work (e.g., Derks & Bakker, 2014;Dery & MacCormick, 2012;Mazmanian et al., 2013).
Others have noted that social norms and expectations serve as important predictors of higher levels of connectivity behaviors after hours (e.g., Adkins & Premeaux, 2014;Gadeyne et al., 2018;Mazmanian et al., 2013;Thörel et al., 2020). This previous work demonstrates the need to distinguish between social norms that contribute to states of connectivity and availability, and social norms around ICT use that results in substantive work practices that constitute TASW. We argue that shared expectations at a team level are particularly influential in predicting individuals' work behaviors within the team. This means that employees in teams with high shared response expectations may signal their proficiency within the team by responding to these expectations (Paczkowski & Kuruzovich, 2016) through technology-assisted supplemental work practices.
In addition, response expectations are particularly acute at the team level, where employees often have a shorter response cycle to other team members, and the pace of responsiveness is reciprocal (Tyler & Tang, 2003). Employees may strategically manage a certain type of image of being responsive, as this can enhance one's reputation as a caring and sensitive colleague (Barley et al., 2011) and proficient coworker (Paczkowski & Kuruzovich, 2016). Conversely, nonresponsiveness is often interpreted with negative attributes, such as incompetence and lack of commitment (Sarker & Sahay, 2004). Team-level response expectations may be so strong that team member's unavailability outside office hours is accepted only when it is collectively agreed on (Perlow & Porter, 2009).
Indeed, Renn (2004, 2010) suggested that organizations that promote quick responses may coerce employees to remain connected and engage in supplemental work after work hours. Hence, we hypothesize that team-level response expectations are associated with TASW.
H2. Team-level response expectations are positively related to TASW.

| Interplay between collaboration technology use and response expectations
In line with recent theorizing on connectivity demands, we acknowledge that workers have some agency in when and how to connect to work (Nurmi & Hinds, 2020). These choices are highly interwoven with the social (organizational) context (Schlachter et al., 2018)-here team-level response expectations. Studies have articulated that being responsive to others is a social practice that may escalate to a norm among coworkers in a team, leading to escalating engagement and cycles of increased responsiveness (Mazmanian et al., 2013). Hence, the interplay between individual technology use and response expectations leads to a redefinition of work practices and work roles. Building on these findings, one might argue that the individual use of collaboration technologies may escalate into TASW practices, especially when these uses are embedded in social contexts characterized by high shared response expectations.
Nurmi and Hinds (2020) argued that workers may face connectivity-related demands (i.e., pressures) but also retain some agency in their decisions about how to respond-for example, engage in more frequent communication, after-hour communication, and/or site visits to dispersed colleagues. Derks et al. (2015) found that smartphone use increases work-life interferences especially for employees who felt strong responsiveness norms. They articulate that employees not only learn from the behaviors of their colleagues that are reflected in these norms but they also mimic their colleagues' behaviors regarding smartphone use after work. In the context of collaborative tools, these norms may exacerbate the extent to which these collaboration technologies are used to engage in supplemental work. Gadeyne et al. (2018) demonstrated that laptop and PC use led to more time-and strain-based work-to-life conflict than smartphone use because connecting through these technologies required a higher level of attention and focus than connecting through a smartphone. In light of the discussion above, collaboration technology use may increase TASW, especially when such usage is embedded in a social context characterized by high shared response expectations among coworkers and supervisors. Hence, we hypothesize the following: H3. Team-level response expectations moderate the positive relationship between collaboration technology use and TASW such that the relationship becomes stronger as response expectations increase.

| Perceived persistence of communication and TASW
Collaboration technology may facilitate work in dispersed teams, where team members work across disparate locations and times (Ellison et al., 2015;Martins et al., 2004). These technologies often have distinct features that present opportunities for employees to take action, communicate, and access information in ways that would be difficult or impossible with other technologies. As such, we recognize that not only do team members act by using these collaboration technologies, these technologies also play an active role in keeping communication available to users. Specifically, organizational realities may be shaped by the interplay of human agency and the materiality of technical artifacts, as technologies may constrain or afford-but not determine-the possibility of achieving new goals and routines (Leonardi, 2011;Orlikowski, 2000). or that users of a technology will recognize or perceive possibilities for persistence. Instead, persistence can be viewed as a possibility of technology use, and not an inherent feature of a technology nor a binary outcome (Evans et al., 2016). In analyzing individuals' activities in online spaces, Mynatt et al. (1998) note that persistence can help "support a wide range of user interaction and collaborative activity" (p. 210). Workers may view persistence as supportive of knowledge sharing across space and time and as facilitating the growth of available content in organizations (Treem & Leonardi, 2013 Ultimately, the applications offered through Google Workspace are aimed at helping its users to keep their workflows organized.
In this study, we draw on multilevel data from multiple sources.
Individual and team-level responses were gathered from team members through an online survey, and the company's human resource department provided demographics and background data. Questionnaires were administered in English. Employees participated on a voluntary basis and did not receive any compensation. They were contacted through email, and questionnaires were collected during a 3-week time period.
In total, 443 participants nested in 122 teams completed the questionnaire. In total, there are 1075 teams with at least three team members; hence, the team-level response rate is 11.35%, the withinteam response rate of participating teams was 40%. Finally

| Dependent variables
Technology-assisted supplemental work (TASW) was measured using a four-item scale adopted from Fenner and Renn (2010 The across-team variance in perceived persistence is σ 2 u0 = .098 and the within-team variance is σ 2 e = .728, suggesting that the shared variance at group level (ICC[1]) is 11.9%. This indicates that perceived persistence did indeed differ among workers overall and within teams.

| Team-level measures
Response expectations were measured adopting the six-items published by Derks et al. (2015, p. 163)  other members of the team, the aggregate composes a construct at the team level (e.g., response norms). Hence, although the referent measured was "I" or "My supervisor," the referent of interest was the team (i.e., collective response expectations within the team). Thus, response expectations represent a social norm about responsiveness that are understood by members of a group and guide or constrain (social) behavior without the force of policy (Cialdini & Trost, 1998).
At the team-level response expectations include the shared expectations valued others-that is, referents, here supervisors and team colleagues-have of responsiveness within their team.
Empirically, a direct consensus method is particularly appropriate for the study of norms formation because it reflects a process of interaction, which is the basis of the norm construction (Taggar & Ellis, 2007). Individuals may have difficulties acting as reliable informants about the group as a whole (van Mierlo et al., 2009). However, within-group homogeneity is a valid assumption as expectations derived from interactions with dispersed supervisors and colleagues are often shared within teams. Third, as indicated below, coefficients of agreement support aggregation of individual-level data to the team level. As such, the direct consensus model is deemed an appropriate composition model for response expectations (Chan, 1998).
Respondents were asked to indicate their agreement or disagreement with regard to expectation about responsiveness in their role within their team. Respondents were prompted to consider the context of their work teams and direct supervisor; "Please consider the interactions with other members of your team and your team supervisor when responding to the following statements." Sample items were "My supervisor expects me to respond to work-related messages during my free time" and "When I send a message to colleagues during the weekend, most colleagues respond the same day." Response categories ranged from 1 (strongly disagree) to 5 (strongly agree). The scale statistics are α = .86, M = 2.20, SD = 0.94.
A prerequisite to conducting multilevel analysis is demonstrating that higher level predictors share substantial within-group variance.
Intraclass correlation (ICC[1]), is used to assess proportional consistency of the total variance that can be explained by team membership, indicating that the proportion of the variability in individual team member's ratings could be attributed to team membership. ICC(2) provides an estimate for reliability of group means, based on mean squares from one-way ANOVA. For this study, the ICC(1) for response expectations was 0.28. The shared variance among team members was significant (F = 2,532 p < .001). In addition, the ICC(2) for response expectations was 0.86, exceeding the 0.7 value recommended by Klein and Kozlowski (2000).

| Controls
Although not the main focus of our analysis, we believe that teamlevel parameters-for example, team size and team dispersion-and individual characteristics of work-for example, working hoursrelated to structural and spatio-temporal context may operate as confounding factors in the relationships underlying the extent to which workers engage in TASW. Human resource data allow us to control for quasi-material features of the team contexts (Barley et al., 2011;Olson & Olson, 2000) that can play an important role in how workers extend their efforts beyond regular work hours in global organizations.

| Individual-level controls
Team leader location. In (globally) dispersed teams, the location of the team leader may also play a role as team members working at a distance (compared with their team leader) may feel excluded and "out of the loop" (O'Leary & Cummings, 2007). Thus, members working remotely from team leaders may have a stronger need to show their presence and contribution by engaging in TASW. Subsequently, for each member, human resource data was used to determine whether the supervisor was co-located (0) or working in a remote location (1).
Finally, work hours and tenure are included as TASW is closely associated with employees' time management and ability to organize their work; actual work hours and organizational tenure may impact TASW (Fenner & Renn, 2010). Work hours and organizational tenure for all respondents was provided by human resources data in hours per week and total number of years, respectively.

| Team-level controls
Time-zone differences. When working across time zones, other team members or leaders may take remote employee's TASW for granted as synchronous collaboration within the team is sometimes needed, even though real-time problem solving in global teams often decreases as time differences increase (O'Leary & Cummings, 2007).
At the same time, remote members may experience higher pressure for responsiveness outside office hours as time differences exacerbate the lag in response time in distributed teams (Sarker & Sahay, 2004).
Based on the geographical location of each team member in a team, the maximum time difference within each team was calculated. The average time difference within dispersed teams was 3.64 h, ranging from 1 h (Helsinki, Finland and Milan, Italy) to 12.5 h (between team van ZOONEN ET AL. members in Chennai, India, and Oakland, USA). Co-location of team members, referring to spatial distance, can also play a role in how much employees engage in TASW. Spatial distance has been shown to have several effects on collaboration (see, e.g., Cummings et al., 2009). Teams of which the constituent members were completely co-located were coded 0, and teams with members that were not co-located were coded 1. Finally, team size can be of importance, as in larger teams coordination of work may become more complex and make extensions of work outside regular hours more likely.
Conversely, in smaller teams, it may be easier to accommodate individual needs, and awareness of others' overtime and schedules are more obvious (Bowers et al., 2000). Team size was measured simply as the number of employees that constitute the team. Correlations ranged from .00 to .43. All constructs demonstrate convergent validity. Discriminant validity was established by evaluating the maximum shared variance (MSV), ranging between .08 and .18, against the square root of the AVE, ranging between .72 and .89. All constructs demonstrate good discriminant validity as the MSV is lower than the AVE.

| Strategy of analysis
The nested data structure with individual responses at level 1 (N = 443) nested within teams at level 2 (N = 122) was analyzed using hierarchical linear modeling. Prior to hypotheses testing individual-level predictor, collaboration technology use, and the moderator, persistence of communication, were centered to the individual mean. The team-level predictor, response norms, was centered around the grand mean (Bauer & Curran, 2005). Curve estimations demonstrated that the relationships in the model were sufficiently linear. To test our hypotheses, we start with the null model and subsequently estimate a sequence of increasingly complex models adding level 1 and level 2 predictors as well as (cross-level) interactions and controls. Overall, the analysis focused on understanding the factors that explain three sources of variance at two levels (a) lower level direct effects, that is, individual-level factors; (b) cross-level direct effects, that is, team-level factors; and (c) cross-level interactions, that is, cross-level factors that explain variance across-group slopes.

| Descriptive statistics and control variables
Correlations and descriptive statistics are provided in Table 1. In order to examine the proportion of variance that is attributed to different levels of analysis, we tested the null model for TASW. The results indicate that the mean for TASW was 2.60 (γ 00 , t = 38.97, p < .001). The model also demonstrates significant variance component at the team level (σ 2 u0 = 0.264 χ 2 = 241.47, df = 121, p < .001). The across-team variance in individual TASW is .264 and the within-team variance is .983. As shown in Table 2, the ICC = .212, which means that team differences account for about 21.2% of the variability TASW. Hence, these results provide evidence for a nested data structure that requires a multilevel modeling analytical approach. Table 2      Subsequently, in Model 2, response expectations were added to the model again demonstrating significant model improvement (Δ À 2x log = 34.05, Δdf = 1 p < .001). The results showed (see Table 2  Note: L1 = Level 1; L2 = Level 2. Values in parentheses are standard errors; t statistics are computed as ratio of each regression coefficient divided by its standard error (reported in text).

| Hypothesis testing
Abbreviations: FIML, full information maximum likelihood estimation; ICC, intraclass correlation. ***Significance levels at p < .001. **Significance levels at p < .01. First, the findings give rise to reconsider the importance of TASW in the context of an increasingly boundary-less work environment.
They do so especially against the backdrop of a global health pandemic and the associated remote work mandates that have almost overnight retired the "nine-to-five commuter" in favor of the "24/7 always available" worker. It might be tempting to abandon the idea that employees have designated times to engage with work-related issues (i.e., work hours) and down-time in which workers recover from work (i.e., nonwork hours). However, in contrast, we believe that what constitutes work and nonwork time and by extension when and why workers engage in supplemental work practices is now more important than ever. Research has shown that employees increasingly experience greater difficulties in managing work and nonwork times (Wang et al., 2021), for instance as (forced) remote work leads to greater task setbacks (Chong et al., 2020), or as they, willingly or compulsory, sacrifice nonwork time to engage in supplemental work (Xiao et al., 2021). Hence, it is important to highlight the need for time off, recovery from work, sleep, and the overall benefits of time in daily life that is distinctly recognized as nonwork. In a working paper, DeFilippis et al. (2020) report that the average workday span increased by 8.2% during a single lockdown period. The authors concluded that the average workday span of an employee was higher in every week following the lockdown than any week in the 8 weeks prior to the lockdown.
Our findings contribute to addressing contemporary challenges in organizational behavior by identifying important mechanisms that contribute to this increase in workday span, and in identifying drivers of these processes also suggest potential interventions. Specifically, we demonstrate that interdependent work in teams supported by col-  (Becker et al., 2018).
Second, the findings confirm the assumption that the use of collaboration technologies is associated with higher levels of TASW, but this relationship is less strong when persistence of communication is high rather than low. Hence, the findings suggest that especially at high levels of persistence, collaboration technology use may not increase individual TASW. For organizations implementing these technologies and developers who build them, these findings are interesting as they may inform design decisions. This means that organizations and developers should pay specific attention to how their technologies may afford a continuous and persistent stream of information and communication that employees may use at their discretion.

| Limitations and future directions
As with any study, several limitations need to be acknowledged in light of the findings presented here. First, this study relies on crosssectional survey data collected from 433 employees spread over 122 teams. Although the study included secondary data sources, the lack of longitudinal data limits our ability to make strong (causal) claims about the hypothesized relationships in our model, or include temporality in our modeling. Second, this study considered the effects of team structure and response expectations. Although team structure largely failed to uniquely contribute to explain variance in individuallevel TASW, response expectations were found to have positive effects. This suggests that other social dynamics at a group level may also contribute to explain variance in TASW. For instance, psychological and broader organizational climates characterized by promoting workaholism and competitiveness may also contribute to TASW (Fenner & Renn, 2004, as well as team and organizational level work-life boundary preferences and initiatives (Kossek et al., 2010).
Hence, future studies may consider a broader model that also • Collaboration technologies used at [organization] keep a record of communication that can last long after the initial communication.
• Communication through collaboration technologies exists long after the initial interaction is finished.
Response expectations (Derks et al., 2015) • My supervisor expects me to respond to work-related messages during my free time after work.
• When I don't answer to messages in my free time, my supervisor clearly shows that he/she does not appreciate it.
• I feel that I have to respond to messages from my supervisor immediately also during leisure time.
• I often receive messages from my colleagues during the weekend.
• When I send a message to colleagues during the weekend, most colleagues react the same day.
• If I do not respond to messages from my colleagues, my position in the group is threatened.
Technology-assisted supplemental work (Fenner & Renn, 2010) • When I fall behind in my work during the day, I work hard at home at night or on weekends to get caught up by using my smartphone or computer.
• I perform job-related tasks at home at night or on weekends using my smartphone or computer.
• I feel my smartphone or computer is helpful in enabling me to work at home at night or on weekend.
• When there is an urgent issue or deadline at work, I tend to perform work-related tasks at home during the night or on weekends using my smartphone or computer.