Dissertation Assignment

Supporting Interdependent Telework Employees: A Moderated-Mediation Model Linking Daily COVID-19 Task Setbacks to Next-Day

Work Withdrawal

SinHui Chong and Yi Huang Nanyang Technological University

Chu-Hsiang (Daisy) Chang Michigan State University

The COVID-19 crisis has compelled many organizations to implement full-time telework for their employees in a bid to prevent a transmission of the virus. At the same time, the volatile COVID-19 situation presents unique, unforeseen daily disruptive task setbacks that divert employees’ attention from routinized work tasks and require them to respond adaptively and effortfully. Yet, little is known about how telework employees react to such complex demands and regulate their work behaviors while working from home. Drawing on Hobfoll’s (1989) conservation of resources (COR) theory, we develop a multilevel, two-stage moderated- mediation model arguing that daily COVID-19 task setbacks are stressors that would trigger a resource loss process and will thus be positively related to the employee’s end-of-day emotional exhaustion. The emotion- ally exhausted employee then enters a resource preservation mode that precipitates a positive relationship between end-of-day exhaustion and next-day work withdrawal behaviors. Based on COR, we also predict that the relation between daily COVID-19 task setbacks and exhaustion would be more positive in telework employees who have higher (vs. lower) task interdependence with coworkers, but organizations could alleviate the positive relation between end-of-day exhaustion and next-day work withdrawal behavior by providing employees with higher (vs. lower) telework task support. We collected daily experience-sampling data over 10 workdays from 120 employees (Level 1, n � 1,022) who were teleworking full-time due to the pandemic lockdown. The results generally supported our hypotheses, and their implications for scholars and managers during and beyond the pandemic are discussed.

Keywords: COVID-19 task setbacks, telework, withdrawal, task interdependence, perceived organizational support

Supplemental materials: http://dx.doi.org/10.1037/apl0000843.supp

Many organizations have long offered telework, also known as telecommuting, remote work, or working from home, as a form of flexible work arrangement to enable employees to better manage increasing demands from their work and family (Shockley & Allen, 2010). Under a typical telework arrangement, an employee splits their work time between working at the office and working from home or a preferred off-site location. When physically at the office, the employee can access organizational infrastructure nec-

essary for performing their job and interact face-to-face with coworkers to coordinate work tasks (Gajendran & Harrison, 2007). Past research has associated this type of voluntary, partial telework with reduced stress and withdrawal, as well as better performance, and they attributed these benefits to the greater autonomy and schedule flexibility that telework extends to employees (Allen, Golden, & Shockley, 2015; Gajendran & Harrison, 2007).

However, the coronavirus disease 2019 (COVID-19) that swept rapidly across the globe in early 2020 conferred a new meaning to telework. As this acute respiratory disease transmits through phys- ical contact (World Health Organization, 2020), many organiza- tions have since discouraged or forbidden nonessential employees from physically reporting to work in order to observe social distancing for halting the virus spread (Guyot & Sawhill, 2020). This pushed the incidence of telework to an unprecedented tipping point. In the United States, 65% of the workforce were teleworking full-time in early May 2020 (Gallup, 2020), a multifold increase from the 11% who had access to partial telework pre-COVID-19 (U.S. Bureau of Labor Statistics, 2019). All other regions, includ- ing Europe (Lomas, 2020) and Asia (Liang, 2020; Tay, 2020), saw record telework rates in the period too. We argue that this form of mandatory telework is fundamentally distinct from the aforemen- tioned partial telework offered as a flexible work arrangement.

Editor’s Note. Jonas W. B. Lang served as the action editor for this article.—LTE

This article was published Online First October 15, 2020. X SinHui Chong and Yi Huang, Division of Leadership, Management,

and Organization, Nanyang Technological University; Chu-Hsiang (Daisy) Chang, Department of Psychology, Michigan State University.

This research was supported by a startup grant awarded to SinHui Chong by Nanyang Technological University, Singapore.

Correspondence concerning this article should be addressed to SinHui Chong, Division of Leadership, Management, and Organization, Nanyang Technological University, 50 Nanyang Avenue, S3-01C-101, Singapore 639798. E-mail: [email protected]

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

Journal of Applied Psychology © 2020 American Psychological Association 2020, Vol. 105, No. 12, 1408 –1422 ISSN: 0021-9010 http://dx.doi.org/10.1037/apl0000843

1408

This is because employees now have little or no volition to decide whether and when to telework, and their ability to access to physical infrastructure, tools, and resources in their workplaces is also severely restricted under this work arrangement.

To complicate matters, the COVID-19 crisis has far-reaching impacts on diverse occupations, because effective responses to COVID-19 require joint and collective efforts across nations, governments, industries, and communities. The evolving and vol- atile nature of both the COVID-19 situation and the adaptive countermeasures creates new and unfamiliar problems that disrupt the original job scopes of many telework employees working in diverse fields. For examples, public sector employees have to now deal with pressing issues such as health care provision, abrupt and rapidly rising unemployment, and trade disruptions due to re- strained international mobility (Evans, 2020). Educators have to conduct classes virtually (Lim-Lange, 2020). Hospitality staff need to devise creative ways to maintain revenue while observing social distancing guidelines and travel bans (Djeebet, 2020).

Integrating past research on organizational constraints (Pindek & Spector, 2016) and goal disruption (Zohar, 1999) into the current situation, we define the abovementioned task-related dis- ruptions and inhibitions arising from the COVID-19 situation as C19 task setbacks. They encompass constraints or disruptive events broadly related to the COVID-19 situation, such as but not limited to lockdown regulations, social-distancing rules, trade or travel disruptions, COVID-19’s impact on business operations, preparation for business reopening after the lockdown, or any issues that originated from the COVID-19 situation and did not exist prior to the pandemic. These C19 task setbacks are distinct from conventional organizational constraints like faulty equipment or poor communication that are directly embedded in employees’ immediate work situation (Pindek & Spector, 2016) because they are linked to the dynamic macroenvironment beyond employees’ organization. The unpredictability C19 task setbacks interfere with employees’ focus, divert their attention from their typical day-to- day job scope, and require effortful responses. We therefore con- tend that these unforeseen, unique C19 task setbacks could have undesirable effects on the daily psychological and behavioral outcomes of telework employees.

Drawing on Hobfoll’s (1989) conservation of resources (COR) theory, we develop a multilevel, moderated-mediation model to argue that daily C19 task setbacks are positively related to end- of-day emotional exhaustion and subsequently next-day work withdrawal behaviors in telework employees. We also predict that the link between daily C19 task setbacks and exhaustion would be more positive in telework employees with higher (vs. lower) task interdependence with their coworkers, but organizations can alle- viate the relationship between end-of-day exhaustion and next-day work withdrawal by providing employees with higher (vs. lower) telework task support.

Our research aims to make several contributions. First, we take a theoretically driven approach to generate timely knowledge for understanding how the COVID-19 pandemic affects employees. Being a deadly pandemic, COVID-19 is an “exogenous sanitary shock” to the economy (Strauss-Kahn, 2020). It is accompanied by an onslaught of economic, societal, and mortality concerns that are matchless by both past financial crises (Tooze, 2020) and former day-to-day task disruptions or organizational constraints that em- ployees were used to (Newton, 2020). Our examination of how

daily C19 task setbacks are linked to employees’ daily exhaustion and withdrawal behaviors is therefore critical to assess the extent of damage generated by the pandemic and the readiness of em- ployees and organizations to tackle it.

Second, the COVID-19 pandemic unearths an unstudied domain within the telework literature. Many organizations had little choice but to hastily transition to mandatory, full-time telework to counter the spread of COVID-19. This mandatory form of full-time tele- work is unlike the erstwhile flexible, partial telework that existing research had studied and that employees were used to. In fact, the COVID-19 situation has removed a large degree of flexibility or volition that telework used to offer workers. It is thus essential to investigate how this current form of telework holds up during the pandemic and for whom it is more or less suitable. By examining task interdependence as our hypothesized first-stage moderator, we take a step in identifying a group of employees who may find it more difficult working at decentralized locations as their cowork- ers during a trying period like this.

Finally, examining organizational task support for telework as a second-stage moderator identifies a contextual factor that may mitigate the negative effect of C19 task setbacks. This offers insights into how organizations can assuage the toll that daily C19 task setbacks take on their employees who are now compelled to work from home.

COR Theory and Model Development

The COR theory offers a framework for understanding em- ployee stress from the perspective of why and how employees lose resources, protect existing resources, and gain new resources (Hobfoll, 1989). Resources in this context refer to “anything perceived by the individual to help attain his or her goals” (p. 5) and can include energies and emotions (Halbesleben, Neveu, Paustian-Underdahl, & Westman, 2014). Fluctuation of resources is one of the theory’s key elements, and Halbesleben et al. (2014) recommended researchers to adopt an episodic, experience- sampling approach to examine resource trajectories within the limits of a timeframe when testing COR. We regard the episodic approach highly suitable for this research because COVID-19 is an ongoing pandemic that unfolds bit by bit, and organizations revise their advisories and practices by the day to deal with the frequent turns of events (Ellis, 2020; Lyons, 2020). Telework employees may therefore receive new and sometimes contradictory work instructions on different days that disrupt how they originally perform their work duties (U.S. Centers for Disease Control and Prevention, 2020). These task-related disruptions and complica- tions brought specifically about by the pandemic are daily C19 task setbacks. Considering Halbesleben et al.’s (2014) advice alongside the volatile nature of the pandemic, we focus on exam- ining the within-person relationships between daily C19 task set- backs and next-day work withdrawal behavior via end-of-day exhaustion.

Halbesleben et al. (2014) also highlighted the importance of context in understanding within-person resource fluctuations based on the COR theory. They explained that the context, such as job nature, organizational climate, and organizational support, acts as boundaries that potentially signal to an individual how resource- draining a trigger is or how valuable resources are. Incorporating this premise into our model, we identify task interdependence as a

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1409DAILY COVID-19 TASK SETBACKS AND WORK WITHDRAWAL

first-stage moderator that exacerbates the relation of daily C19 task setbacks with end-of-day exhaustion and organizational telework task support as a second-stage moderator that alleviates the rela- tion of end-of-day exhaustion with next-day work withdrawal in telework employees. Our model is summarized in Figure 1.

Daily C19 Task Setbacks and End-of-Day Emotional Exhaustion

The COR theory argues that a resource loss can be caused by a “shock to one’s cognitive processing that forces the individual to carefully evaluate this new information” (Halbesleben et al., 2014, p. 18). The COVID-19 crisis has generated additional and unpre- dictable work demands for many occupations in different indus- tries. Policy and operational changes enacted jointly by public and private agencies to counter the pandemic permeate from the eco- logical environment into the organizational environment. We re- gard these daily C19 task setbacks as work-related stressors that function in a similar manner as organizational constraints (Pindek & Spector, 2016) and goal-disruptive events (Zohar, 1999) to trigger a resource loss process characterized by end-of-day emo- tional exhaustion.

These C19 task setbacks require employees to appraise unfore- seen problems, unlearn their existing automatic task scripts promptly, develop novel solutions, learn new ways of operations, and adapt to updated rules and advisories, all of which are complex cognitive activities that deplete resources rapidly (Muraven & Baumeister, 2000; Schmeichel, Vohs, & Baumeister, 2003). C19 task setbacks also reinforce the salience of this pandemic’s threats, which may evoke stress and worry about mortality and job security (Tan, 2020). Hence, employees must exert regulatory resources to manage negative thoughts and emotions to continue focusing on their tasks, which further drain their resources (Johns, Inzlicht, & Schmader, 2008). We therefore argue that daily C19 task setbacks can rapidly drain the resources of telework employees and reduce them to emotional exhaustion by the end of the day (Maslach & Jackson, 1981).

Hypothesis 1 (H1): Daily C19 task setbacks are positively related to end-of-day emotional exhaustion.

Task Interdependence as First-Stage Moderator

The COR theory asserts the importance of considering the extraindividual context that can serve as a “fertile or infertile ground” (p. 107) to steep the within-person resource fluctuation process (Hobfoll, Halbesleben, Neveu, & Westman, 2018). We contend that telework employees who have higher (vs. lower) task interdependence with their coworkers will experience greater end- of-day exhaustion from daily C19 task setbacks. Employees with higher task interdependence rely on and interact frequently with their coworkers to coordinate efforts toward achieving common work goals (Morgeson & Humphrey, 2006). They have shared habitual mental models of how to perform work tasks interdepen- dently (Burke, Stagl, Salas, Pierce, & Kendall, 2006). Therefore, when daily C19 task setbacks hit employees with higher task interdependence, the setbacks will likely implicate an entire cluster of interdependent employees. This deviation from shared work routines brought about by daily C19 task setbacks will require not only a teleworker’s direct investment of resources into appraising the situation and developing action plans but also resources into communicating, coordinating, and synchronizing efforts with co- workers in order to resolve the issues at hand collectively (Shep- perd, 1993) while working at different locations. As highlighted by Burke et al. (2006), team-level situational assessment, plan formu- lation, execution, and learning consume a considerable amount of individual and shared resources. Considering employees’ unfamil- iarity with the COVID-19 crisis, it is more likely for interdepen- dent employees to have to expend greater resources into tackling daily C19 task setbacks, thus draining their resources more rapidly and leading to greater exhaustion by the end of the day.

Hypothesis 2 (H2): The relationship between daily C19 task setbacks and end-of-day exhaustion is more positive in tele- work employees who have higher (vs. lower) task interdepen- dence with their coworkers.

End-of-Day Emotional Exhaustion and Next-Day Work Withdrawal Behavior

One of the COR principles states that when people exhaust their resources, they focus on preserving themselves from a further loss

Figure 1. Results of hypothesized moderated-mediation model. Coefficients are unstandardized. The figure only shows hypothesized paths. The full path model includes main effects of task interdependence on emotional exhaustion, task support on next-day withdrawal, control variables of prior-day sleep quality and Friday on emotional exhaustion, and current-day work withdrawal and Friday on next-day withdrawal. These nonhypoth- esized paths are omitted from the figure for parsimony; please refer to Table 5, Model 4 for the full estimated model. � p � .05. �� p � .01.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1410 CHONG, HUANG, AND CHANG

of resources (Hobfoll et al., 2018). This suggests a positive rela- tionship between end-of-day exhaustion and next-day work with- drawal behavior. Withdrawal is a counterproductive work behavior that characterizes an employee’s bid to avoid their work in order to escape from job stressors (Spector et al., 2006). It entails restrict- ing the amount of time and effort spent on work, such as absence, late arrival or early departure, and taking undeserved breaks (Spec- tor et al., 2006). As telework employees are not constantly visible to their supervisor and coworkers, there is comparatively less deterrence for them to engage in work withdrawal behavior. Tele- work employees can usually decide their preferred time to begin or end work with less immediate social sanction from coworkers, and they can engage in personal activities during their work hours.

Under these telework conditions, we expect the experience of exhaustion from C19 task setbacks the previous day to activate the resource preservation mode and prompt work withdrawal the next day in telework employees. Doing so offers them a temporary respite from C19 task setbacks for preventing further loss of resource (Chong, Kim, Lee, Johnson, & Lin, 2020; Kiazad, Seib- ert, & Kraimer, 2014). Relatedly, past research showed that with- drawal is an emotion-based coping strategy for exhausted employ- ees who have little control over their job stressors (Krischer, Penney, & Hunter, 2010). This is analogous to the lack of control that employees have over the COVID-19 situation. Finally, with- drawal from work also buys employees time for their resource pool to replenish (Trougakos & Hideg, 2009). Thus, we hypothesize the following:

Hypothesis 3 (H3): End-of-day emotional exhaustion is pos- itively related to next-day work withdrawal behavior.

Organizational Telework Task Support as Second- Stage Moderator

For the final part of our model, we refer back to the COR for potential ways to mitigate the above hypothesized dysfunctional self- preservation behavior (withdrawal) in response to depleted resources (exhaustion). The COR emphasizes that “ecological conditions [such as social and environmental conditions can] foster and nurture or limit and block resource creation and sustenance” (p. 107) after individuals experience resource loss (Hobfoll et al., 2018). The COVID-19 crisis compelled the majority of employees to switch to full-time telework relatively abruptly without much option to prepare for this work arrangement (Agade, 2020; Liang, 2020; Lomas, 2020). Thus, we propose the importance of organizational telework task support as an environmental factor that expands employees’ capacity and resource pool for mitigating the relationship between end-of-day exhaustion and next-day withdrawal behavior. Adapting the definition of per- ceived organizational support (Eisenberger, Stinglhamber, Vanden- berghe, Sucharski, & Rhoades, 2002), we define organizational tele- work task support as employees’ belief in how much the organization is making an effort to provide them with necessary telework-related resources while they are working from home, such as information technology (IT) support, timely information, relevant work materials, and decision-making authority.

Social cues presented by an organization can shape employees’ responses toward resource loss (Halbesleben et al., 2014). Based on the crossover mechanism in COR, when employees perceive their organizations’ efforts to help them work well from home, they will

recognize that they can draw from this organizational support (i.e., an external resource pool) to replenish their own resources after experi- encing exhaustion from C19 task setbacks the previous day. The telework task support offered by the organization also serves to enhance employees’ perceived control over their environment, thereby regenerating their coping efficacy (Richardson, Yang, Van- denberg, DeJoy, & Wilson, 2008). On the other hand, if an organi- zation is indifferent about offering telework task support, it may further convince exhausted employees that they are isolated and unable to rely on organizational resources to replenish their own resources in their endeavor to overcome the job stressors (Richardson et al., 2008). This crippling feeling will likely strengthen the positive relationship between end-of-day exhaustion and next-day withdrawal behavior.

Hypothesis 4 (H4): The relationship between end-of-day emo- tional exhaustion and next-day work withdrawal behavior is more positive in employees who perceive lower (vs. higher) organizational support for telework.

We integrate the prior hypotheses into a two-stage moderated mediation model to address why daily C19 task setbacks are related to next-day work withdrawal behavior (i.e., via end-of-day exhaustion), for whom the relationship will be more readily observed (i.e., for telework employees who have higher task interdependence with co- workers), and finally how to mitigate this relationship (i.e., by pro- viding higher organizational telework task support). This specific positioning of task interdependence as a first-stage moderator and organizational telework task support as a second-stage moderator is based on two key principles of COR. First, the resource loss process possesses greater momentum than that of preventing resource loss (Hobfoll et al., 2018). This suggests that in the face of an external shock (i.e., daily C19 task setbacks), the moderating effect of a factor that exacerbates resource loss (i.e., higher task interdependence) will be more salient and observable than the moderating effect of a factor that prevents resource loss (i.e., higher organizational telework task support). Thus, we expect task interdependence to function as a first-stage moderator.

Second, the COR theory (Hobfoll et al., 2018) asserts that individ- uals will actively seek to prevent further loss and to regain resources by drawing from internal or external resource pools typically after they have experienced the loss (i.e., emotional exhaustion in our case). This means that individuals experiencing emotional exhaustion will try turning their attention away from stressors that deplete them (i.e., daily C19 task setbacks and task interdependence) and direct their focus on potential means of redress (i.e., organizational telework task support) in order to set their resource reclamation process in motion. Some past research offers preliminary evidence for our arguments. They demonstrated resource pool or capacity (i.e., job control or trait emotional control) as a significant second-stage moderator mitigating the relations between resource loss (e.g., depletion or negative emo- tions) and work behavioral outcomes (Lanaj, Johnson, & Barnes, 2014; Tong, Chong, & Johnson, 2019), instead of a first-stage mod- erator buffering the effects of stressors on resource loss.

Hypothesis 5 (H5): The indirect positive relationship between daily C19 task setbacks and next-day work withdrawal behav- iors via end-of-day emotional exhaustion is the most positive in employees with higher (vs. lower) task interdependence and lower (vs. higher) organizational telework support.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1411DAILY COVID-19 TASK SETBACKS AND WORK WITHDRAWAL

Method

Contextual Background, Procedure, and Participants

The data for this article were part of a broader data collection effort. The procedure was reviewed and approved by the Nanyang Techno- logical University Institutional Review Board (IRB-2019 – 090-037– 01; Title: Work Experience of Individuals). We collected the data in Singapore during its 8-week nationwide lockdown that began on April 7, 2020, during which it was mandatory for all employees to telework where possible. This resulted in 85% of the workforce working from home (Tay, 2020). Only individuals employed in es- sential services such as health care, cleaning, and food preparation were permitted to report physically to their workplaces. The surveys were administered between the fifth and seventh weeks of the lock- down. By this time, employees would have acclimatized to the tele- work arrangement, thus allowing us to avoid any initial honeymoon- hangover effect (Boswell, Boudreau, & Tichy, 2005) associated with the novelty of mandatory telework.

Participants were recruited via snowball sampling. The first author posted a recruitment advertisement on her social media profile and disseminated the advertisement directly to 20 personal and professional contacts working in diverse occupations for snowball sharing. The advertisement indicated that this was a research on work attitudes and behaviors without disclosing the exact research topic. It stated that participants would receive S$65 for completing a baseline survey and 10 daily surveys. Interested individuals accessed a survey link to answer eligibility questions, which included age (at least 18 years old), residence (Singapore), and employment status (employed and teleworking full-time). Out of the 228 individuals who responded to the eligibility survey

within 2 days, 204 met all eligibility criteria and were invited to complete the baseline survey, and 128 completed the baseline survey by the stipulated deadline. In the following week, we emailed and texted a survey link to participants at 8 p.m. for 10 workdays. They had to complete the survey before 3 a.m. the next day. Participants submitted the survey at 9:05 p.m. on average. Out of the participants who completed the baseline survey, 127 com- pleted at least one daily survey, and we received 1,197 daily responses in total (response rate of 93.5%). As the outcome vari- able was work withdrawal behavior on the next workday (i.e., a lagged variable), participants who completed only one daily survey (n � 7) were dropped. The predictor and mediator data collected on days that had missing consecutive next-day responses due to participants skipping a survey (n � 48) and on Day 10 (i.e., there was no Day 11 withdrawal behavior data after the final Day 10 survey; n � 120) were also excluded. This resulted in a final sample of 1,022 daily responses (Level 1 n) from 120 participants (Level 2 n).

Participants were 65.8% females and 97.5% ethnic Chinese, and they had a mean age of 32.3 (SD � 5.0). On average, their organizational tenure was 4.4 years (SD � 4.1) and weekly work hours were 40.6 (SD � 16.3). They came from diverse industries, for example, civil service (20.0%), specialized fields of law or research (15.8%), financial sector (15.0%), and education (7.5%).

Measures

Focal variables. Details of our focal variable measures are presented in Table 1. All adapted or self-developed items for any measures are reported in the Appendix.

Table 1 Measures of Focal Variables

Variable name and source of measure Number of items and response scale Sample item

Baseline survey General task interdependence

(Pearce & Gregersen, 1991) Five items; 1 � not at all; 5 � a

great deal “In my job in general, I frequently must coordinate my efforts

with other coworkers.” Perceived organizational telework

task support Adapted existing supportive

behavior scale (Trougakos, Beal, Cheng, Hideg, & Zweig, 2015) to COVID-19 telework context

Three items; 1 � strongly disagree; 5 � strongly agree

“During this COVID-19 work-from-home period, my company goes out of its way to provide me with task support (e.g., IT support, decision-making authority, etc.).” [Complete list of the items shown in the Appendix.]

Demographics Age, gender, race, industry, organizational tenure, general weekly work hours

Daily survey Daily C19 task setbacks

Adapted disruptive work events scale (Zohar, Tzischinski, & Epstein, 2003) to COVID-19 situation

Three items; 1 � not at all; 5 � a great deal

“Today, I had to divert time or effort from my typical, original work duties to handle issues or reasons related to the COVID-19 situation.” [Complete list of the items shown in the Appendix.]

End-of-day emotional exhaustion (Maslach & Jackson, 1981; Teuchmann, Totterdell, & Parker, 1999)

Two items; 1 � not at all; 5 � a great deal

“I feel burned out at the moment.”

Daily work withdrawal behavior Adapted work withdrawal

behavior scale (Spector et al., 2006) to telework context

Four items; 1 � not at all; 5 � a great deal

“I worked less hours than allowed today.” [Complete list of the items shown in the Appendix.]

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1412 CHONG, HUANG, AND CHANG

Control variables. We controlled for prior day sleep quality (i.e., “How well did you sleep last night?” 1 � very poorly; 5 � very well) when predicting end-of-day exhaustion as sleep quality affects exhaustion (Söderström, Jeding, Ekstedt, Perski, & Åker- stedt, 2012). We also controlled for current-day work withdrawal behavior when predicting next-day work withdrawal behavior to account for autoregression. Finally, we controlled for Friday (1 � Friday; 0 � other days) when predicting exhaustion and with- drawal behavior to account for any “Thank-God-it’s-Friday” ef- fects (Stone, Schneider, & Harter, 2012).

Results

Table 2 presents the correlations of the variables, and Table 3 presents the variances and intraclass correlation ICC(1) values. Within-person variances ranged from 33.13% to 54.55%, justifying the use of experience sampling. We conducted a multilevel confir- matory factor analysis to assess the measurement model. Daily C19 task setbacks, exhaustion, and work withdrawal were estimated on both the within-person and between-person levels, and general task interdependence and organizational telework support were estimated on the between-person level.1 The factor structure had fit indices of �2(131) � 225.489, comparative fix index (CFI) � .98, root mean square error of approximation (RMSEA) � .03, standardized root mean square residual (SRMR)Within � .02, and SRMRBetween � .09. All alternative models, in which items from two factors were loaded onto a single factor, had a poorer fit to the data (218.96 � ��2(4 � �df � 6) � 1,171.94, p � .01), thus supporting our five focal variables as distinct.

To test our hypotheses, we ran a multilevel moderated- mediation model in MPlus v8 (Muthén & Muthén, 2012) using observed variables. The use of observed over latent variables was guided by precedence (Lanaj et al., 2014; Uy, Lin, & Ilies, 2017)

and more importantly informed by scholars who have raised con- cerns about the interpretability of estimating cross-level interac- tions using latent variables (Hayes, Montoya, & Rockwood, 2017; Maslowsky, Jager, & Hemken, 2015). We created a lagged vari- able to match next-workday withdrawal behavior to current-day C19 task setback (predictor) and exhaustion (mediator) for analy- ses. Exogenous between-person variables were grand-mean- centered and exogenous within-person variables were group- mean-centered (Aguinis, Gottfredson, & Culpepper, 2013). To examine slope variability (Bliese, 2016; Snijders & Bosker, 2012), we compared the �2 loglikelihood of a model with and a model without random slopes between daily C19 task setbacks, end-of- day exhaustion, and next-day withdrawal (refer to Table 4). The difference of 67.78 indicated significant between-person slope variability, p � .00, and justified the use random slopes for the a path (i.e., predictor-mediator) and b path (i.e., mediator-outcome). Paths involving control variables were modeled as fixed slopes (Wang, Liao, Zhan, & Shi, 2011).

The results from the full hypothesized moderated-mediation model are summarized in Figure 1 and Table 5 (Model 4). Daily C19 task setbacks were positively related to end-of-day exhaustion (� � .18, SE � 06, p � .00), thus supporting H1. This relationship was significantly moderated by task interdependence (� � .17,

1 Following guidelines from past scale development research for detect- ing common error variance (Avolio, Bass, & Jung, 1999; Miller, Jenkins, Kaplan, & Salonen, 1995), we allowed two items of the withdrawal scale—“I took a longer break from work than allowed today” and “I worked less hours than formally allowed today”—with a high modification index (�200) to covary within the factor, without any change to the factor structure. Before covarying them, the fit indices of the factor structure were �2(133) � 628.19, CFI � .90, RMSEA � .06, SRMRWithin � .08, SRMRBetween � .12.

Table 2 Means, Standard Deviations, and Correlations of Study Variables

Variable M SD 1 2 3 4 5 6 7 8 9 10

Within-person 1. Current-day C19 task setbacks 1.78 .61 (.87) 2. Current end-of-day emotional exhaustion 2.17 .72 .20�� (.88) 3. Next-day withdrawal behavior 1.37 .41 .00 .08� (.61) 4. Previous day sleep quality 3.22 .71 �.05 �.16�� .03 — 5. Current-day withdrawal behavior 1.38 .40 �.01 �.00 .02 �.03 (.61)

Between-person 1. Current-day C19 task setbacks 1.78 .80 (.97) 2. Current end-of-day emotional exhaustion 2.17 1.03 .35�� (.99) 3. Next-day withdrawal behavior 1.37 .40 .31�� .17 (.78) 4. Previous day sleep quality 3.22 .63 �.20� �.36�� �.15 — 5. Current-day withdrawal behavior 1.38 .39 .28�� .18� 1.00�� �.15 (.78) 6. Task interdependence 3.81 .73 �.09 .13 �.18� .16 �.18� (.87) 7. Organizational telework task support 3.64 .93 �.13 �.08 �.01 .09 .03 .08 (.89) 8. Gender .33 .47 .08 .09 �.05 .12 �.03 .14 .02 — 9. Age 32.33 4.96 .07 �.04 �.07 �.06 �.11 .02 �.11 .11 —

10. Organizational tenure 4.42 4.05 .13 �.08 �.06 �.03 �.08 �.06 .06 �.07 .47�� — 11. General work hours per week 40.64 16.33 �.07 .14 �.04 �.08 �.04 .21 �.01 .13 .01 .00

Note. Level 1, n � 1,022; Level 2, n � 120. Correlations with gender (1 � male; 0 � female) are based on n � 117 as 3 people chose not to report their gender. Parentheses on the diagonals indicate the within-person and between-person Cronbach’s alphas of the scales. These multilevel alphas were calculated by [(the squared number of indicators present at a given level of analysis the average interitem covariance at the same level)/the sum of all elements in the full covariance matrix at the same level] (Geldhof, Preacher, & Zyphur, 2014). � p � .05. �� p � .01.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1413DAILY COVID-19 TASK SETBACKS AND WORK WITHDRAWAL

SE � .07, p � .02), such that it was more positive for employees who had higher (vs. lower) task interdependence with their co- workers, thus supporting H2. Figure 2 illustrates the simple slopes (Preacher, Curran, & Bauer, 2006). End-of-day exhaustion was not significantly related to next-workday withdrawal behavior (� � .02, SE � .02, p � .33), thus not supporting H3. Nevertheless, the second-stage moderation was significant and supported H4 (� � �.05, SE � .02, p � .03). The simple slopes (see Figure 3) showed that the relationship between end-of-day exhaustion and next-day withdrawal behavior was positive and significant for employees who perceived lower organizational telework task sup- port, and it was not significant for employees who perceived higher organizational task support for telework. To evaluate effect sizes, we calculated R2 values. There are several R2 approaches for multilevel research. Lang, Bliese, and Runge (2019) recommended using a generalized R2 statistic for estimating simple multilevel models, while Rights and Sterba (2019a, 2019b) encouraged the decomposition of R2 values into their respective levels for complex models. Given that our model is relatively complex with a two- stage, cross-level moderated mediation, we reported the multilevel

R2 values suggested by Rights and Sterba (2019a, 2019b) to understand how our predictors at the two levels contribute to the explained variances of dependent variables (see Table 6).

We calculated conditional indirect effects using a Monte Carlo simulation with 10,000 replications to construct confidence inter- vals around the estimates (Bauer, Preacher, & Gil, 2006). As shown in Table 7, the indirect relationship between daily C19 task setbacks and next-day withdrawal behavior via end-of-day exhaus- tion was most positive and significant for employees whose task interdependence was higher (vs. lower) and whose perceived or- ganizational telework task support was lower (vs. higher), where its 95% CIs [.001, .042] did not include zero. Adopting the subgroup approach by Edwards and Lambert (2007), we tested the significance of the moderated mediation by comparing the indirect effects between two subgroups in which the two moderators had opposing values (i.e., higher interdependence and lower support group vs. lower interdependence and higher support group). The difference was significant with 95% CIs [.002, .043], thus sup- porting H5.

Table 3 Variance Components of Within-Person Variables

Variable Within-person variance (r2) Between-person variance (u2) ICC(1)

Current-day C19 task setbacks .38 .64 .63 Current end-of-day emotional exhaustion .53 1.07 .67 Next-day withdrawal behavior .18 .15 .45

Note. Level 1, n � 1,022; Level 2, n � 120. ICC(1) values represent the proportion of variance that can be explained by the between-person level, which was computed using the equation [between-person variance/ (within-person variance between-person variance)]. ICC(1) � intraclass correlation(1).

Table 4 Comparison Between Using Fixed Versus Random Slopes Between X, M, and Y

Variable

Model with fixed slopes for X ¡ M, M ¡ Y, and X ¡ Y

Model with random slopes for X ¡ M, M ¡ Y, and X ¡ Y

b SE t p b SE t p

Predicting end-of-day emotional exhaustion Daily C19 task setbacks .23 .04 5.99�� .00 .18 .06 3.09�� .00

Predicting next-day work withdrawal behavior Daily C19 task setbacks �.01 .02 �.37 .71 �.03 .02 �1.07 .29 End-of-day emotional exhaustion .05 .02 2.41�� .02 .04 .02 1.74† .08

Within-person residual variance End-of-day emotional exhaustion .51 .02 21.24 .00 .46 .02 20.35 .00 Next-day work withdrawal behavior .17 .01 21.24 .00 .16 .01 19.76 .00

Between-person residual variance End-of-day emotional exhaustion 1.07 .15 7.35 .00 1.07 .15 7.38 .00 Next-day work withdrawal behavior .15 .02 6.79 .00 .08 .02 4.04 .00 Random slope (X ¡ M) — — — — .11 .04 2.85 .00 Random slope (M ¡ Y) — — — — .01 .00 3.48 .00 Random slope (X ¡ Y) — — — — .00 .01 .29 .77 AIC 3,936.44 3,876.66 BIC 3,980.80 3,940.75 Loglikelihood value �1,959.22 �1,925.33

Note. X (daily C19 task setbacks), M (end-of-day emotional exhaustion), and Y (next-day work withdrawal behavior). To examine slope variability (Bliese, 2016; Snijders & Bosker, 2012), we compared the �2 loglikelihood of the two models above (i.e., ��2 � (�2 �1,959.22) � (�2 �1,925.33) � 67.78, p � .00). The above models were estimated to compare fixed and random slopes between X, M, and Y. They do not contain Level 1 control variables or Level 2 variables. Thus, the path coefficients and variance estimates reported here are different from Models 2a and 2b presented in Table 5. AIC � Akaike information criterion; BIC � Bayesian information criterion. † p � .10. �� p � .01.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1414 CHONG, HUANG, AND CHANG

Table 5 Coefficients From Path Models

Variable

Model 0: Null model Model 1: With control variables

added as fixed slopes

b SE t p b SE t p

Predicting end-of-day emotional exhaustion Daily C19 task setbacks — — — — — — — — General task interdependence — — — — — — — — Daily C19 Task Setbacks General Task

Interdependence — — — — — — — — Previous-day sleep quality (control) — — — — �.17 .03 �4.99�� .00 Friday (control) — — — — �.09 .05 �1.63 .10

Predicting next-day work withdrawal — — — — — — — — Daily C19 task setbacks — — — — — — — — End-of-day emotional exhaustion — — — — — — — — Organizational telework task support — — — — — — — — End-of-Day Emotional Exhaustion Organizational

Telework Task Support — — — — — — — — Previous-day work withdrawal (control) — — — — �.04 .03 �1.10 .27 Friday (control) — — — — �.04 .03 �1.15 .25

Within-person residual variance End-of-day emotional exhaustion .53 .03 21.24�� .00 .51 .02 21.24�� .00 Next-day work withdrawal .18 .01 21.34�� .00 .17 .01 21.23�� .00

Between-person residual variance End-of-day emotional exhaustion 1.07 .15 7.33�� .00 1.07 .15 7.34�� .00 Next-day work withdrawal .15 .02 6.79�� .00 .15 .02 6.79�� .00 Random slope (X ¡ M) — — — — — — — — Random slope (M ¡ Y) — — — — — — — — Random slope (X ¡ Y) — — — — — — — —

Model fit AIC 10,514.30 6,317.48 BIC 10,603.03 6,401.28 Loglikelihood value �5,239.15 �3,141.74

Variable

Model 2a: Fixed slopes mediation model without moderators

Model 2b: Random slopes mediation model without moderators

b SE t p b SE t p

Predicting end-of-day emotional exhaustion Daily C19 task setbacks .22 .04 5.75�� .00 .17 .06 2.92�� .00 General task interdependence — — — — — — — — Daily C19 Task Setbacks General Task

Interdependence — — — — — — — — Previous-day sleep quality (control) �.15 .03 �4.68�� .00 �.16 .03 �4.90�� .00 Friday (control) �.08 .05 �1.43 .15 �.07 .05 �1.34 .18

Predicting next-day work withdrawal — — — — — — — — Daily C19 task setbacks �.01 .02 �.41 .68 �.03 .02 �1.09 .28 End-of-day emotional exhaustion .05 .02 2.35� .02 .04 .02 1.71† .09 Organizational telework task support — — — — — — — — End-of-Day Emotional Exhaustion Organizational

Telework Task Support — — — — — — — — Previous-day work withdrawal (control) �.04 .03 �1.11 .27 �.04 .03 �1.20 .23 Friday (control) �.03 .03 �1.02 .31 �.03 .03 �.93 .35

Within-person residual variance End-of-day emotional exhaustion .49 .02 21.24�� .00 .45 .02 20.35�� .00 Next-day work withdrawal .17 .01 21.24�� .00 .16 .01 19.67�� .00

Between-person residual variance End-of-day emotional exhaustion 1.07 .15 7.35�� .00 1.08 .15 7.39�� .00 Next-day work withdrawal .15 .02 6.79�� .00 .08 .02 3.99�� .00 Random slope (X ¡ M) — — — — .11 .04 2.90�� .00

(table continues)

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1415DAILY COVID-19 TASK SETBACKS AND WORK WITHDRAWAL

Finally, we conducted supplemental analyses and estimated two models (see Table 8) in which we switched the stages of the two moderators (Model 5) or included both moderators at both stages (Model 6). Both models supported general task interdependence as a significant first-stage but not second-stage moderator and orga- nizational telework support as a significant second-stage but not first-stage moderator.

Discussion

Theoretical and Practical Implications

Our findings have several theoretical and practical implica- tions, and we discuss them jointly. First, amid the upheaval set

off by the COVID-19 crisis, our research was a timely attempt that examined the work-related psychological implications of the new telework norm brought about by the pandemic. Prior research has associated telework with beneficial outcomes for employees and organizations (Allen et al., 2015; Gajendran & Harrison, 2007). Our research, on the other hand, was based on a unique form of telework that was rather abruptly imposed on employees as a mandatory precautionary measure in light of the pandemic instead of a flexi-work option. The results uncovered potential difficulties and detriments that employees, especially those with higher task interdependence, may face as they tackle unforeseen and unfamiliar daily task setbacks (i.e., C19 task setbacks) while being confined to working from home, physi- cally distantly from their coworkers. Thus, we have identified a

Table 5 (continued)

Variable

Model 2a: Fixed slopes mediation model without moderators

Model 2b: Random slopes mediation model without moderators

b SE t p b SE t p

Random slope (M ¡ Y) — — — — .01 .00 3.48�� .00 Random slope (X ¡ Y) — — — — .00 .01 .37 .71

Model fit AIC 4,512.51 4,450.93 BIC 4,596.31 4,554.45 Loglikelihood value �2,239.25 �2,204.47

Model 3: Random slopes model with Level 2 task interdependence and

telework task support added as main effects without moderating effects yet

Model 4: Full hypothesized random slopes model with

cross-level moderations

Variable b SE t p b SE t p

Predicting end-of-day emotional exhaustion Daily C19 task setbacks .17 .06 2.93�� .00 .18 .06 3.26�� .00 General task interdependence .24 .14 1.71† .09 .24 .14 1.72† .09 Daily C19 Task Setbacks General Task

Interdependence — — — — .17 .07 2.28� .02 Previous-day sleep quality (control) �.16 .03 �4.90�� .00 �.15 .03 �4.80�� .00 Friday (control) �.07 .05 �1.34 .18 �.07 .05 �1.37 .17

Predicting next-day work withdrawal Daily C19 task setbacks �.02 .02 �.99 .32 �.02 .02 �.97 .33 End-of-day emotional exhaustion .03 .02 1.16 .25 .02 .02 .98 .33 Organizational telework task support .05 .04 1.25 .21 .11 .05 2.36� .02 End-of-Day Emotional Exhaustion Organizational

Telework Task Support — — — — �.05 .02 �2.18� .03 Previous-day work withdrawal (control) �.04 .03 �1.19 .23 �.04 .03 �1.06 .29 Friday (control) �.03 .03 �.94 .35 �.03 .03 �.92 .36

Within-person residual variance End-of-day emotional exhaustion .45 .02 20.36�� .00 .45 .02 20.34�� .00 Next-day work withdrawal .16 .01 19.71�� .00 .16 .01 19.81�� .00

Between-person residual variance End-of-day emotional exhaustion 1.06 .14 7.38�� .00 1.06 .14 7.37�� .00 Next-day work withdrawal .08 .02 3.97�� .00 .08 .02 4.05�� .00 Random slope (X ¡ M) .12 .04 2.92�� .00 .10 .04 2.67�� .01 Random slope (M ¡ Y) .01 .00 3.55�� .00 .01 .00 3.33�� .00 Random slope (X ¡ Y) .00 .01 .36 .72 .00 .01 .37 .71

Model fit AIC 4,452.25 4,446.43 BIC 4,570.55 4,574.60 Loglikelihood value �2,202.12 �2,197.21

Note. Level 1, n � 1,022; Level 2, n � 120. Friday (1 � Friday, 0 � other weekdays). X (daily C19 task setbacks), M (end-of-day emotional exhaustion), and Y (next-day work withdrawal behavior). AIC � Akaike information criterion; BIC � Bayesian information criterion. The Mplus output of Model 4 is provided in the online supplemental materials of this article. † p � .10. � p � .05. �� p � .01.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1416 CHONG, HUANG, AND CHANG

specific group of employees (i.e., those with higher interdepen- dence) who are deserving of greater attention and support when transitioning to telework.

Second, we shed light on a new aspect of telework, that is, the use of telework as a crisis response. Our results highlighted the need for further research to understand how to assist organiza- tions and employees to transition smoothly and promptly to telework without compromising on their well-being and perfor- mance. This is especially relevant both during and after the pandemic because some organizations have expressed an intent to implement permanent telework for their employees after the pandemic (Heater, 2020; Rooney, 2020). As the circumstances surrounding large-scale telework may be different (i.e., hasty switch during pandemic vs. planned effort at a later time), we recommend future research to tease out nuances in the construct of telework, such as voluntary versus mandatory telework, planned versus unplanned implementation, partial versus full- time telework, and independent versus interdependent tasks, in order to gain greater insights into how to best execute and facilitate telework.

Third, our results extended past COR findings by showing that exhausted employees could also draw from external re- sources (i.e., organizational telework task support) for replen- ishment. This should encourage further research on cross-level and cross-domain mechanisms of resource exchange (Hobfoll et al., 2018), as these might generate findings to help organiza- tions develop effective interventions for sustaining engaged and resilient employees. Practically, our findings highlighted the need for organizations to provide telework task support for exhausted telework employees who face unanticipated daily C19 task setbacks, especially those whose jobs require higher interdependence.

In addition, we opened a potential avenue for future research on organizational constraints. Conventional organizational con- straints (Pindek & Spector, 2016) largely refer to situational obstacles within the organization that inhibit employees from performing optimally, such as machine malfunctions or poor information flow (Spector & Jex, 1998). C19 task setbacks, in contrast, are broader and arise from more complex and dynamic origins that could be intraorganizational, extraorganizational, or both, as various internal and external stakeholders may join

forces and taking multipronged approaches to fight the pan- demic during this period. Our research therefore highlighted additional sources of constraints worthy of examination by organizational researchers.

Limitations, Future Directions, and Conclusion

We acknowledge several limitations in our research. First, al- though our sample of participants from diverse occupations and industries allowed us to achieve variance in general task interde- pendence and organizational telework task support for testing our hypotheses, we recognize that this could also be a limitation because we did not specifically measure or control for the readi- ness of companies or industries in shifting to telework. Participants from telework-ready companies may feel more equipped to tackle daily C19 task setbacks while they work from home. This can lead to a restricted range of task setbacks as they work from home. Future research can consider studying industrial or organizational characteristics to generate practical insights for specific industries or organizations.

Second, we assessed organizational telework task support based on employee perceptions. This allowed us to establish the impor- tance of perceived organizational telework task support for helping employees overcome daily exhaustion, but it did not shed light on the exact content of effective organizational telework support strategies. Future research can endeavor to develop a checklist of telework task support strategies (Andriessen, 2007) and conduct a field experiment (Eden, 2017) to test their effectiveness.

Third, we could not include all conceivable resource-related variables linked to the pandemic and telework into our model. For instance, telework success may be determined in part by the resources that employees have at home, such as Internet connec- tivity and decent workspace (Allen et al., 2015). Employees from more affluent backgrounds tend to have greater access to these resources (Desilver, 2020), and thus future research could seek to identify vulnerable employees or include socioeconomic back- ground as a control variable.

Finally, we acknowledge that our model is a complex one that included eight variables, a mediating chain, and two cross-level mod- erations. Saylors and Trafimow (2020) warned that complex models

Figure 2. Simple slopes between daily C19 task setbacks and end-of-day emotional exhaustion at different levels of task interdependence.

Figure 3. Simple slopes between end-of-day emotional exhaustion and next-day withdrawal behavior at different levels of organizational telework task support.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1417DAILY COVID-19 TASK SETBACKS AND WORK WITHDRAWAL

have a lower likelihood of being true because it is more difficult for the large number of correlations in its matrix to possess the right causal status. It will therefore be beneficial to unpack our model in future and test each relationship individually using lab or field exper- iments where possible to establish causality.

In conclusion, we demonstrated that daily C19 task setbacks are positively related to next-day work withdrawal behavior via end- of-day emotional exhaustion, especially for telework employees who have higher task interdependence with coworkers and per- ceive lower organizational telework task support. This investiga-

Table 6 Multilevel R2 Measures Based on Within-Person and Between-Person Variances Explained

Variable Within-person residual

variance in M Between-person residual

variance in M

Between-person residual variance in random slope between X

and M

Model 0: Null model .525 1.067 — Model 1: Daily controls (previous-day sleep

quality and Friday) added as Level 1 fixed slopes .509; Rw2(f1) � 3.05% 1.067 —

Model 2a: Daily C19 task setbacks added as Level 1 fixed slope .491; Rw2(f1) � 6.48% 1.069 —

Model 2b: Daily C19 task setbacks added as Level 1 random slope .450; Rw2(v) � 14.29% 1.076 .114

Model 3: General task interdependence added as Level 2 main effect 1.060; Rb2(f2) � 1.49%

Model 4: Random slope (X ¡ M) regressed on Level 2 general task interdependence for cross-level moderation .450 1.060 .099; Rb2(m) � 13.16%

Within-person residual variance in Y

Between-person residual variance in Y

Between-person residual variance in random slope between M

and Y

Model 0: Null model .175 .151 — Model 1: Daily controls (previous-day

withdrawal and Friday) added as Level 1 fixed slopes .174; Rw2(f1) � .57% .151 —

Model 2a: End-of-day exhaustion added as Level 1 fixed slope .173; Rw2(f1) � 1.14% .151 —

Model 2b: End-of-day exhaustion added as Level 1 random slope .164; Rw2(v) � 6.29% .080 .012

Model 3: Organizational telework task support added as Level 2 main effect .079; Rb2(f2) � 1.25%

Model 4: Random slope (M ¡ Y) regressed on Level 2 organizational telework task support for cross-level moderation .163 .082 .011; Rb2(m) � 8.33%

Note. X (daily C19 task setbacks), M (end-of-day emotional exhaustion), and Y (next-day work withdrawal behavior). Model numbers correspond with the models presented in Table 5. Rw2(f1) represents within-person variance explained by Level 1 predictor added as fixed slope; Rw2(v) represents within-person variance explained by Level 1 predictor added as random slope; Rb2(f2) represents between-person variance explained by main effect of Level 2 moderator; Rb

2(m) represents random slope variance explained by estimating cross-level moderation (Rights & Sterba, 2019a, 2019b). The R2 values were calculated using [(baseline residual variance � model residual variance)/baseline residual variance]. Rw2(f1) and Rw2(f1) used Model 0 as the baseline, while Rb2(f2) and Rb2(m)

used Model 2b, in which random slopes were first estimated, as the baseline.

Table 7 Conditional Indirect Effects at Different Levels of Task Interdependence and Organizational Telework Task Support

First-stage moderator (general task

interdependence)

Second-stage moderator (organizational telework

task support)

Conditional indirect effect

95% CIs of the conditional indirect

effect

95% CIs of the difference between conditional

indirect effectsa

Lower (�1 SD) Lower (�1 SD) .004 [�.006, .015] [�.003, .035] Higher ( 1 SD) �.001 [�.006, .004] [.002, .043]

Higher ( 1 SD) Lower (�1 SD) .021 [.001, .042] Comparison baseline Higher ( 1 SD) �.006 [�.026, .013] [�.001, .054]

Note. Indirect effect refers to the relationship of daily C19 task setbacks with next-day withdrawal via end-of-day emotional exhaustion. A Monte Carlo simulation with 10,000 replications was used to compute the 95% CIs around the estimates. CIs that do not contain zero are significant (bolded in table). a Using the subgroup approach (Edwards & Lambert, 2007), indirect effects at higher ( 1 SD) and lower (�1 SD) values of the moderators were compared to the hypothesized strongest indirect effect at higher task interdependence and lower organizational telework task support.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1418 CHONG, HUANG, AND CHANG

tion highlights the importance of supporting telework employees during this trying period.

References

Agade, H. (2020). Coronavirus: Experiences of people working from home. CGTN Africa. Retrieved from https://africa.cgtn.com/2020/03/26/ coronavirus-experiences-of-people-working-from-home/

Aguinis, H., Gottfredson, R. K., & Culpepper, S. A. (2013). Best-practice recommendations for estimating cross-level interaction effects using multilevel modeling. Journal of Management, 39, 1490 –1528. http://dx .doi.org/10.1177/0149206313478188

Allen, T. D., Golden, T. D., & Shockley, K. M. (2015). How effective is telecommuting? Assessing the status of our scientific findings. Psycho- logical Science in the Public Interest, 16, 40 – 68. http://dx.doi.org/10 .1177/1529100615593273

Andriessen, D. (2007). Designing and testing an OD intervention: Reporting intellectual capital to develop organizations. Journal of Applied Behavioral Science, 43, 89 –107. http://dx.doi.org/10.1177/ 0021886306297010

Avolio, B. J., Bass, B. M., & Jung, D. I. (1999). Re-examining the components of transformational and transactional leadership using the Multifactor Leadership. Journal of Occupational and Organizational Psychology, 72, 441– 462. http://dx.doi.org/10.1348/096317999166789

Bauer, D. J., Preacher, K. J., & Gil, K. M. (2006). Conceptualizing and testing random indirect effects and moderated mediation in multilevel models: New procedures and recommendations. Psychological Methods, 11, 142–163. http://dx.doi.org/10.1037/1082-989X.11.2.142

Bliese, P. D. (2016). Multilevel modeling in R (2.6): A brief introduction to R, the multilevel package and the nlme package. Retrieved from https:// cran.r-project.org/doc/contrib/Bliese_Multilevel.pdf

Boswell, W. R., Boudreau, J. W., & Tichy, J. (2005). The relationship between employee job change and job satisfaction: The honeymoon- handover effect. Journal of Applied Psychology, 90, 882– 892. http://dx .doi.org/10.1037/0021-9010.90.5.882

Burke, C. S., Stagl, K. C., Salas, E., Pierce, L., & Kendall, D. (2006). Understanding team adaptation: A conceptual analysis and model. Jour- nal of Applied Psychology, 91, 1189 –1207. http://dx.doi.org/10.1037/ 0021-9010.91.6.1189

Table 8 Models From Supplemental Analyses

Variable

Model 5: Switched organizational telework task support to first-

stage moderator and general task interdependence to second-stage

moderator

Model 6: With task interdependence and

organizational telework task support as concurrent

moderators at both stages

b SE t p b SE t p

Predicting end-of-day emotional exhaustion Daily C19 task setbacks .16 .06 2.84�� .00 .18 .06 3.12�� .00 General task interdependence — — — — .19 .13 1.45 .15 Organizational telework task support �.10 .11 �.99 .32 �.10 .10 �.93 .35 Daily C19 Task Setbacks General Task Interdependence — — — — .16 .07 2.17� .03 Daily C19 Task Setbacks Organizational Telework Task Support �.06 .06 �1.01 .32 �.07 .06 �1.20 .23 Previous-day sleep quality (control) �.16 .03 �4.88�� .00 �.15 .03 �4.79�� .00 Friday (control) �.07 .05 �1.36 .17 �.07 .05 �1.39 .16

Predicting next-day work withdrawal behavior Daily C19 task setbacks �.03 .02 �1.13 .26 �.03 .02 �1.15 .25 End-of-day emotional exhaustion .03 .02 1.28 .20 .03 .02 1.07 .29 General task interdependence �.17 .06 �2.64�� .00 �.17 .06 �2.72�� .01 Organizational telework task support — — — — .12 .05 2.58� .10 End-of-Day Emotional Exhaustion General Task Interdependence .02 .03 .76 .45 .03 .03 1.06 .29 End-of-Day Emotional Exhaustion Organizational Telework Task Support — — — — �.05 .02 �2.37� .02 Previous-day work withdrawal (control) �.04 .03 �1.18 .24 �.03 .03 �1.02 .31 Friday (control) �.03 .03 �.95 .34 �.03 .03 �.92 .36

Within-person residual variance End-of-day emotional exhaustion .45 .02 20.37 .00 .45 .02 20.37 .00 Next-day work withdrawal .16 .01 19.69 .00 .16 .01 19.80 .00

Between-person residual variance End-of-day emotional exhaustion 1.07 .15 7.38 .00 1.05 .14 7.38 .00 Next-day work withdrawal .07 .02 3.78 .00 .07 .02 3.79 .00 Random slope (X ¡ M) .12 .04 2.96 .00 .10 .04 2.74 .01 Random slope (M ¡ Y) .01 .00 3.60 .00 .01 .00 3.42 .00 Random slope (X ¡ Y) .00 .01 .27 .79 .00 .01 .30 .77

Model fit AIC 4,449.68 4,443.51 BIC 4,577.85 4,591.39 Loglikelihood value �2,198.84 �2,191.75

Note. Level 1, n � 1,022; Level 2, n � 120. Friday (1 � Friday, 0 � other weekdays). X (daily C19 task setbacks), M (end-of-day emotional exhaustion), and Y (next-day work withdrawal behavior) In Model 5, the order of our hypothesized moderators was switched. In Model 6, both task interdependence and organizational teleworks support were modeled as concurrent moderators at both stages (i.e., two moderators at both stages). AIC � Akaike information criterion; BIC � Bayesian information criterion. � p � .05. �� p � .01.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1419DAILY COVID-19 TASK SETBACKS AND WORK WITHDRAWAL

Chong, S., Kim, Y. J., Lee, H. W., Johnson, R. E., & Lin, S.-H. J. (2020). Mind your own break! The interactive effect of workday respite activ- ities and mindfulness on employee outcomes via affective linkages. Organizational Behavior and Human Decision Processes, 159, 64 –77. http://dx.doi.org/10.1016/j.obhdp.2019.11.001

Desilver, D. (2020). Before the coronavirus, telework was an optional benefit, mostly for the affluent few. Retrieved from https://www .pewresearch.org/fact-tank/2020/03/20/before-the-coronavirus- telework-was-an-optional-benefit-mostly-for-the-affluent-few/

Djeebet, H. (2020). What is the impact of COVID-19 on the global hospitality industry? Retrieved from https://www.hospitalitynet.org/ opinion/4098062.html

Eden, D. (2017). Field experiments in organizations. Annual Review of Organizational Psychology and Organizational Behavior, 4, 91–122. http://dx.doi.org/10.1146/annurev-orgpsych-041015-062400

Edwards, J. R., & Lambert, L. S. (2007). Methods for integrating moder- ation and mediation: A general analytical framework using moderated path analysis. Psychological Methods, 12, 1–22. http://dx.doi.org/10 .1037/1082-989X.12.1.1

Eisenberger, R., Stinglhamber, F., Vandenberghe, C., Sucharski, I. L., & Rhoades, L. (2002). Perceived supervisor support: Contributions to perceived organiza- tional support and employee retention. Journal of Applied Psychology, 87, 565–573. http://dx.doi.org/10.1037/0021-9010.87.3.565

Ellis, R. (2020). WHO changes stance, says public should wear masks. Retrieved from https://www.webmd.com/lung/news/20200608/who- changes-stance-says-public-should-wear-masks

Evans, J. (2020). European governments scramble to find farm workers for harvest. Financial Times. Retrieved from https://www.ft.com/content/ 601a4dd9-b996-4c49-bc22-8f49bfd06ea0

Gajendran, R. S., & Harrison, D. A. (2007). The good, the bad, and the unknown about telecommuting: Meta-analysis of psychological media- tors and individual consequences. Journal of Applied Psychology, 92, 1524 –1541. http://dx.doi.org/10.1037/0021-9010.92.6.1524

Gallup. (2020). COVID-19: A leader’s guide to developing a work-from-home strategy. Retrieved from https://www.gallup.com/workplace/310988/covid- 19-working-from-home-guide.aspx

Geldhof, G. J., Preacher, K. J., & Zyphur, M. J. (2014). Reliability estimation in a multilevel confirmatory factor analysis framework. Psy- chological Methods, 19, 72–91. http://dx.doi.org/10.1037/a0032138

Guyot, K., & Sawhill, I. V. (2020). Telecommuting will likely continue long after the pandemic. Retrieved from https://www.brookings.edu/blog/up-front/2020/ 04/06/telecommuting-will-likely-continue-long-after-the-pandemic/

Halbesleben, J. R., Neveu, J.-P., Paustian-Underdahl, S. C., & Westman, M. (2014). Getting to the “COR” understanding the role of resources in conservation of resources theory. Journal of Management, 40, 1334 – 1364. http://dx.doi.org/10.1177/0149206314527130

Hayes, A. F., Montoya, A. K., & Rockwood, N. J. (2017). The analysis of mechanisms and their contingencies: PROCESS versus structural equa- tion modeling. Australasian Marketing Journal, 25, 76 – 81. http://dx.doi .org/10.1016/j.ausmj.2017.02.001

Heater, B. (2020). Twitter says staff can continue working from home perma- nently. Tech Crunch. Retrieved from https://techcrunch.com/2020/05/12/ twitter-says-staff-can-continue-working-from-home-permanently/

Hobfoll, S. E. (1989). Conservation of resources: A new attempt at con- ceptualizing stress. American Psychologist, 44, 513–524. http://dx.doi .org/10.1037/0003-066X.44.3.513

Hobfoll, S. E., Halbesleben, J., Neveu, J.-P., & Westman, M. (2018). Conservation of resources in the organizational context: The reality of resources and their consequences. Annual Review of Organizational Psychology and Organizational Behavior, 5, 103–128. http://dx.doi.org/ 10.1146/annurev-orgpsych-032117-104640

Johns, M., Inzlicht, M., & Schmader, T. (2008). Stereotype threat and executive resource depletion: Examining the influence of emotion reg-

ulation. Journal of Experimental Psychology: General, 137, 691–705. http://dx.doi.org/10.1037/a0013834

Kiazad, K., Seibert, S. E., & Kraimer, M. L. (2014). Psychological contract breach and employee innovation: A conservation of resources perspec- tive. Journal of Occupational and Organizational Psychology, 87, 535– 556. http://dx.doi.org/10.1111/joop.12062

Krischer, M. M., Penney, L. M., & Hunter, E. M. (2010). Can counterpro- ductive work behaviors be productive? CWB as emotion-focused cop- ing. Journal of Occupational Health Psychology, 15, 154 –166. http:// dx.doi.org/10.1037/a0018349

Lanaj, K., Johnson, R. E., & Barnes, C. M. (2014). Beginning the workday yet already depleted? Consequences of late-night smartphone use and sleep. Organizational Behavior and Human Decision Processes, 124, 11–23. http://dx.doi.org/10.1016/j.obhdp.2014.01.001

Lang, J. W., Bliese, P. D., & Runge, J. M. (2019). Detecting consensus emergence in organizational multilevel data: Power simulations. Orga- nizational Research Methods. Advance online publication. http://dx.doi .org/10.1177/1094428119873950

Liang, L.-H. (2020). How Covid-19 led to a nationwide work-from-home exper- iment. Retrieved from https://www.bbc.com/worklife/article/20200309- coronavirus-covid-19-advice-chinas-work-at-home-experiment

Lim-Lange, C. (2020). COVID-19’s education revolution—where going dig- ital is just half the battle. Retrieved from https://www.channelnewsasia.com/ news/commentary/covid-19-coronavirus-education-digital-revolution- transformation-12650974

Lomas, N. (2020). European Commission goes teleworking by default over COVID-19. Retrieved from https://techcrunch.com/2020/03/12/ european-commission-goes-teleworking-by-default-over-covid-19/

Lyons, K. (2020). AMC Theaters changes course and will require custom- ers to wear face masks. Retrieved from https://www.theverge.com/2020/ 6/19/21296698/amc-theaters-ceo-masks-coronavirus-covid-19-adam- aron-reopening

Maslach, C., & Jackson, S. E. (1981). The measurement of experienced burnout. Journal of Organizational Behavior, 2, 99 –113. http://dx.doi .org/10.1002/job.4030020205

Maslowsky, J., Jager, J., & Hemken, D. (2015). Estimating and interpreting latent variable interactions: A tutorial for applying the latent moderated structural equations method. International Journal of Behavioral Devel- opment, 39, 87–96. http://dx.doi.org/10.1177/0165025414552301

Miller, T. Q., Jenkins, C. D., Kaplan, G. A., & Salonen, J. T. (1995). Are all hostility scales alike? Factor structure and covariation among mea- sures of hostility. Journal of Applied Social Psychology, 25, 1142–1168. http://dx.doi.org/10.1111/j.1559-1816.1995.tb02611.x

Morgeson, F. P., & Humphrey, S. E. (2006). The Work Design Question- naire (WDQ): Developing and validating a comprehensive measure for assessing job design and the nature of work. Journal of Applied Psy- chology, 91, 1321–1339. http://dx.doi.org/10.1037/0021-9010.91.6.1321

Muraven, M., & Baumeister, R. F. (2000). Self-regulation and depletion of limited resources: Does self-control resemble a muscle? Psychological Bulletin, 126, 247–259. http://dx.doi.org/10.1037/0033-2909.126.2.247

Muthén, L. K., & Muthén, B. O. (2012). Mplus user’s guide (7th ed.). Los Angeles, CA: Author.

Newton, C. (2020). Tech giants are getting creative to manage the COVID-19 crisis. Retrieved from https://www.theverge.com/interface/ 2020/3/17/21181691/google-verily-trump-website-trials-amazon-hiring- covid-19-response

Pearce, J. L., & Gregersen, H. B. (1991). Task interdependence and extrarole behavior: A test of the mediating effects of felt responsibility. Journal of Applied Psychology, 76, 838 – 844. http://dx.doi.org/10.1037/ 0021-9010.76.6.838

Pindek, S., & Spector, P. E. (2016). Organizational constraints: A meta- analysis of a major stressor. Work & Stress, 30, 7–25. http://dx.doi.org/ 10.1080/02678373.2015.1137376

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1420 CHONG, HUANG, AND CHANG

Preacher, K. J., Curran, P. J., & Bauer, D. J. (2006). Computational tools for probing interactions in multiple linear regression, multilevel model- ing, and latent curve analysis. Journal of Educational and Behavioral Statistics, 31, 437– 448. http://dx.doi.org/10.3102/10769986031004437

Richardson, H. A., Yang, J., Vandenberg, R. J., DeJoy, D. M., & Wilson, M. G. (2008). Perceived organizational support’s role in stressor-strain relationships. Journal of Managerial Psychology, 23, 789 – 810. http:// dx.doi.org/10.1108/02683940810896349

Rights, J. D., & Sterba, S. K. (2019a). New recommendations on the use of R-squared differences in multilevel model comparisons. Multivariate Behavioral Research. Advance online publication. http://dx.doi.org/10 .1080/00273171.2019.1660605

Rights, J. D., & Sterba, S. K. (2019b). Quantifying explained variance in multilevel models: An integrative framework for defining R-squared measures. Psychological Methods, 24, 309 –338. http://dx.doi.org/10 .1037/met0000184

Rooney, K. (2020). Square to allow employees to work from home per- manently, following Twitter’s lead. Retrieved from https://www.cnbc .com/2020/05/18/square-lets-employees-work-from-home-permanently .html

Saylors, R., & Trafimow, D. (2020). Why the increasing use of complex causal models is a problem: On the danger sophisticated theoretical narratives pose to truth. Organizational Research Methods. Advance online publication. http://dx.doi.org/10.1177/1094428119893452

Schmeichel, B. J., Vohs, K. D., & Baumeister, R. F. (2003). Intellectual performance and ego depletion: Role of the self in logical reasoning and other information processing. Journal of Personality and Social Psy- chology, 85, 33– 46. http://dx.doi.org/10.1037/0022-3514.85.1.33

Shepperd, J. A. (1993). Productivity loss in performance groups: A moti- vation analysis. Psychological Bulletin, 113, 67– 81. http://dx.doi.org/10 .1037/0033-2909.113.1.67

Shockley, K. M., & Allen, T. D. (2010). Investigating the missing link in flexible work arrangement utilization: An individual difference perspec- tive. Journal of Vocational Behavior, 76, 131–142. http://dx.doi.org/10 .1016/j.jvb.2009.07.002

Snijders, T. A., & Bosker, R. J. (2012). Multilevel analysis: An introduc- tion to basic and advanced multilevel modeling. Thousand Oaks, CA: Sage.

Söderström, M., Jeding, K., Ekstedt, M., Perski, A., & Åkerstedt, T. (2012). Insufficient sleep predicts clinical burnout. Journal of Occupational Health Psychology, 17, 175–183. http://dx.doi.org/10.1037/a0027518

Spector, P. E., Fox, S., Penney, L. M., Bruursema, K., Goh, A., & Kessler, S. (2006). The dimensionality of counterproductivity: Are all counter- productive behaviors created equal? Journal of Vocational Behavior, 68, 446 – 460. http://dx.doi.org/10.1016/j.jvb.2005.10.005

Spector, P. E., & Jex, S. M. (1998). Development of four self-report measures of job stressors and strain: Interpersonal conflict at work scale, organizational constraints scale, quantitative workload inventory, and physical symptoms inventory. Journal of Occupational Health Psychol- ogy, 3, 356 –367. http://dx.doi.org/10.1037/1076-8998.3.4.356

Stone, A. A., Schneider, S., & Harter, J. K. (2012). Day-of-week mood patterns in the United States: On the existence of ‘Blue Monday’, ‘Thank God it’s Friday’ and weekend effects. Journal of Positive Psychology, 7, 306 –314. http://dx.doi.org/10.1080/17439760.2012.691980

Strauss-Kahn, M.-O. (2020). Can we compare the COVID-19 and 2008 crises? Retrieved from https://www.atlanticcouncil.org/blogs/new- atlanticist/can-we-compare-the-covid-19-and-2008-crises/

Tan, T. (2020). Anxiety and worry amid Covid-19 uncertainty. The Straits Times. Retrieved from https://www.straitstimes.com/singapore/anxiety- and-worry-amid-covid-19-uncertainty

Tay, T. F. (2020). Coronavirus: Last digit of IC to determine entry to four markets; essential workforce to be cut to 15 per cent. The Straits Times. Retrieved from https://www.straitstimes.com/singapore/coronavirus- last-digit-of-identity-card-number-to-determine-entry-to-four-markets

Teuchmann, K., Totterdell, P., & Parker, S. K. (1999). Rushed, unhappy, and drained: An experience sampling study of relations between time pressure, perceived control, mood, and emotional exhaustion in a group of accountants. Journal of Occupational Health Psychology, 4, 37–54. http://dx.doi.org/10.1037/1076-8998.4.1.37

Tong, J., Chong, S., & Johnson, R. E. (2019). The indirect relations of workplace incivility with emotional exhaustion and supportive behaviors via self-blame: The moderating roles of observed incivility and trait emotional control. Journal of Organizational Behavior, 40, 931–946. http://dx.doi.org/10.1002/job.2399

Tooze, A. (2020). Is the coronavirus crash worse than the 2008 financial crisis? Retrieved from https://foreignpolicy.com/2020/03/18/ coronavirus-economic-crash-2008-financial-crisis-worse/

Trougakos, J. P., Beal, D. J., Cheng, B. H., Hideg, I., & Zweig, D. (2015). Too drained to help: A resource depletion perspective on daily interper- sonal citizenship behaviors. Journal of Applied Psychology, 100, 227– 236. http://dx.doi.org/10.1037/a0038082

Trougakos, J. P., & Hideg, I. (2009). Momentary work recovery: The role of within-day work breaks. In P. Perrewé, J. Halbesleben, & C. Rose (Eds.), Research in occupational stress and wellbeing (Vol. 7, pp. 37– 84). Bingley, UK: Emerald Group Publishing Limited. http://dx.doi .org/10.1108/S1479-3555(2009)0000007005

U.S. Bureau of Labor Statistics. (2019). Availability of flexible work schedule and student loan repayment benefits in the United States. Retrieved from https://www.bls.gov/ncs/ebs/factsheet/flexible-work- schedule-and-student-loan-repayment.htm

U.S. Centers for Disease Control and Prevention. (2020). COVID-19: Business and workplaces (Plan, prepare, and respond). Retrieved from https://www.cdc.gov/coronavirus/2019-ncov/community/organizations/ businesses-employers.html

Uy, M. A., Lin, K. J., & Ilies, R. (2017). Is it better to give or receive? The role of help in buffering the depleting effects of surface acting. Academy of Management Journal, 60, 1442–1461. http://dx.doi.org/10.5465/amj .2015.0611

Wang, M., Liao, H., Zhan, Y., & Shi, J. (2011). Daily customer mistreat- ment and employee sabotage against customers: Examining emotion and resource perspectives. Academy of Management Journal, 54, 312–334. http://dx.doi.org/10.5465/amj.2011.60263093

Williams, L. J., & Anderson, S. E. (1991). Job satisfaction and organiza- tional commitment as predictors of organizational citizenship and in-role behaviors. Journal of Management, 17, 601– 617. http://dx.doi.org/10 .1177/014920639101700305

World Health Organization. (2020). Coronavirus disease (COVID-19) pandemic. Retrieved from https://www.who.int/emergencies/diseases/ novel-coronavirus-2019

Zohar, D. (1999). When things go wrong: The effect of daily work hassles on effort, exertion and negative mood. Journal of Occupational and Organizational Psychology, 72, 265–283. http://dx.doi.org/10.1348/ 096317999166671

Zohar, D., Tzischinski, O., & Epstein, R. (2003). Effects of energy avail- ability on immediate and delayed emotional reactions to work events. Journal of Applied Psychology, 88, 1082–1093. http://dx.doi.org/10 .1037/0021-9010.88.6.1082

(Appendix follows)

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1421DAILY COVID-19 TASK SETBACKS AND WORK WITHDRAWAL

Appendix

Items of Adapted Scales

Perceived Organizational Telework Task Support Items

Adapted from Trougakos, Beal, Cheng, Hideg, and Zweig (2015) and Williams and Anderson (1991).

1. During this COVID-19 work-from-home period, my company goes out of its way to provide me with task support (e.g., IT support, decision-making authority, etc.).

2. My company takes a personal interest in whether I have all the work tools and resources that I need to work well at home during this COVID-19 period.

3. My company makes it easy for me to retrieve task resources (e.g., information, materials, IT resources) while I work from home during this COVID-19 period.

(1 � strongly disagree; 5 � strongly agree)

Daily C19 Task Setbacks Items

Adapted from Zohar (1999). Instructions: The “COVID-19 situation” in the statements

should be viewed broadly. It can refer to COVID-19 itself or issues linked to or caused by COVID-19, such as the lockdown regula- tions, social distancing, COVID-19’s impact on business opera- tions, the preparation for business reopening after the lockdown, or any issues that have arisen from the COVID-19 situation and did not exist before the pandemic.

1. Today, something related to the COVID-19 situation disrupted me from my planned work goals.

2. Today, I had to divert time or effort from my typical, original work duties to handle issues or reasons related to the COVID-19 situation.

3. Today, I encountered an unforeseen or new difficulty in carrying out my scheduled work plans due to issues or reasons related to the COVID-19 situation.

(1 � not at all; 5 � a great deal)

Daily Work Withdrawal Behavior Items

Adapted from Spector et al. (2006).

1. I joined online/call meetings late without permission today.

2. I skipped online/call meetings without legitimate reason today.

3. I took a longer break from work than allowed today.

4. I worked less hours than allowed today.

(1 � not at all; 5 � a great deal)

Received July 14, 2020 Revision received September 2, 2020

Accepted September 4, 2020 �

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

1422 CHONG, HUANG, AND CHANG