Long working hours among mobile-flexible employees

As mentioned above, for purposes of lowering costs, A-FOs are mostly combined with mobile-flexible working arrangements, which describes the new way of working (NWW). The NWW is characterized by (1) time and location flexibility, (2) job autonomy, and (3) the increased use of information and communication technologies (ICT) (^{Austin-Egole et al.}, 2020; ^{Rafnsdóttir & Heijstra}, 2013). A systematic review from ^Tavares (2017) concludes that working hours have extended into evenings, nights, and weekends, especially for employees in flexible working arrangements. ^{Rafnsdóttir and Heijstra} (2013) have also shown that while flexibility simplifies some aspects of working life, it also initiates a higher workload, triggering an extension of working hours and a higher demand for availability. ^Eurofound (2020) supported these findings confirming that flexible working arrangements are associated with increased workloads and extended working hours. Scholars (^{Demerouti et al.}, 2014; ^{Derks & Bakker}, 2010; ^Eurofound, 2020; ^Sonnentag, 2018) explain longer working hours among mobile-flexible employees as follows: Constant connectivity leads to more work interruptions with an increased volume of work and pressure to respond to demands as quickly as possible. The interruptions force workers to extend their working hours to finish the work they intended to complete within the normal workday.

Long working hours and exhaustion among mobile-flexible employees

In a cross-sectional study, ^{Buruck et al.} (2020) show that increased workload is related to increased exhaustion in mobile-flexible employees. This finding is closely linked to ^Sonnentag`s (2018) recovery paradox, which states that workers in stressful occupations are more likely to work during non-working time and be consistently available on their mobile devices (^{Bliese et al.}, 2017; ^{Demerouti et al.}, 2014; ^Eurofound, 2020; ^Sonnentag, 2018). Such a working mode leads to the paradoxical combination of high need for recovery due to high job stressors with a low propensity to rest. According to the recovery paradox, the constant connection to technology and thereby to work leads to more frequent states of negative activation and a depletion of energy resources. Mobile availability and mobile device use during non-working time create a behavioral pathway that combines high occupational stressors with low psychological detachment from work. A sign of such low psychological detachment is frequent rumination about work issues. According to ^Sonnentag (2018), this state of high stress and low detachment hinders the recovery process and natural recharge of energy resources (^{Åkerstedt et al.}, 2002; ^{Litwiller et al.}, 2017; ^{Pereira et al.}, 2016), which in turn may lead to the increased experience of exhaustion.

Exhaustion is defined by ^{Demerouti et al.} (2003) as the result of intense physical, affective, and cognitive stress and is the long-term result of prolonged exposure to specific work demands (p.14). Several studies could show that quantitative job demands, such as high workload or time pressure, are the strongest predictors of exhaustion (^{Bennett et al.}, 2018; ^{Seidler et al.}, 2014; ^Sonnentag, 2018). Exhaustion can be explained by the job demands-resources (JD-R) model (^{Demerouti et al.}, 2001), that postulates that every working environment has its own risks and protective factors in terms of health and motivation. These factors can be categorized into job demands (e.g., time pressure) and job resources (e.g., support by work colleagues; ^{Bakker & Demerouti}, 2017; ^{Demerouti et al.}, 2001). The JD-R model proposes two psychological processes. The first is the health impairment process, which begins with persistent work demands that can decrease employees' energy, leading to constant overwork and exhaustion (^{Bakker & Demerouti}, 2017; ^{Kattenbach et al.}, 2010). The second is the motivational process, which starts with the availability of job resources and is likely to lead to work engagement and positive outcomes, such as higher organizational commitment and enhanced job performance (^{Bakker & Demerouti}, 2017; ^{Demerouti et al.}, 2003; ^{Kattenbach et al.}, 2010). As these two processes show, job demands are not primarily negative, but they can turn into job stressors when the satisfaction of the demands involves great efforts (physical and mental) from which employees are unable to recover sufficiently and, therefore, experience physiological and/or psychological consequences (^{Bakker & Demerouti}, 2017; ^{Demerouti et al.}, 2007; ^{Meijman & Mulder}, 1998).

Long working hours, exhaustion, and rumination among mobile-flexible employees

The aforementioned studies point out that mobile-flexible employees are exposed to several work stressors such as long working hours, and that these may lead to rumination. Rumination is defined as repeatedly thinking about negative personal feelings and/or the effects, causes, and meanings of a negative mood (^{Kinnunen et al.}, 2017). Rumination involves cognitive and emotional processing, which is the process of actively thinking about a stressor, the thoughts and feelings it elicits, and its impact on one's life and future (^Watkins, 2008). According to the perseverative cognition (PC) hypothesis, repeated thinking about a problem prolongs immediate psychological and physiological responses to life events and daily stressors (^Watkins, 2008). As a result, the body activates its biological stress-response systems chronically, which can increase the likelihood of disease (^{Kinnunen et al.}, 2017; ^Watkins, 2008). The PC hypothesis explains why rumination is linked to exhaustion: The act of rumination does not allow for recovery during non-working time (^{Berset et al.}, 2011; ^{Eggli et al.}, 2021; ^{Jimenez et al.}, 2021; ^{Kinnunen et al.}, 2017; ^{Nolen-Hoeksema et al.}, 2008; ^{Weigelt et al.}, 2019). In the stressor-detachment model, ^{Sonnentag and Fritz} (2015) propose that harm to the organism is not mainly the acute stress response but rather the persistent activation, even when the stressor is no longer present (p. 75). Multiple studies suggest that constantly thinking about previously experienced negative events is detrimental to health (^{Clark et al.}, 2020; ^{Fritz & Sonnentag}, 2006; ^Watkins, 2008). Cross-sectional studies (^{Galliker et al.}, 2020; ^{Vandevala et al.}, 2017; ^{Zoupanou & Rydstedt}, 2019) as well as diary studies (^{Cropley et al.}, 2006; ^{Eggli et al.}, 2021) have shown that rumination reactivates the emotional and physiological stress response. After-work rumination is associated with physical and mental health risks such as increased risks of cardiovascular diseases (^{Cropley et al.}, 2017), increased cortisol responses (^{Cropley et al.}, 2015; ^Rydstedt, 2009), sleep issues (^{Berset et al.}, 2011), and work-related fatigue (^{Querstret & Cropley}, 2012). ^{Kinnunen et al.} (2019) showed in a three-wave longitudinal study that work-related rumination predicted exhaustion and ^{Firoozabadi et al.} (2018) confirmed these findings in a five-day diary study. Regarding the relationship between work demands and rumination, several studies found that work stressors combining high demands and low control are significantly associated with work-related rumination (^{Cropley and Millward Purvis},2003; ^{Cropley et al.}, 2006; ^{Grebner et al.}, 2005). Additionally, ^{Höge and Hornung} (2013) showed that mobile-flexible work is related to increased rumination. Hence, the present study assumes work-related rumination to be, like exhaustion, an individual stress reaction among mobile-flexible employees toward work stressors, such as long working hours.

The Role of Rumination as a possible Mediator

Another interesting question is whether the extension of previous day working hours directly affects mental health or whether rumination can act as a mediator. According to ^{Verkuil et al.} (2010), perseverative cognition like rumination or worry is an individual reaction to stressful situations. ^{Brosschot et al.} (2005) define the term preservative cognition as the repeated or chronic activation of cognitive representations with stress-related content (p. 1045). Work-related preservative cognition is defined as not being able to psychologically detach from work (^{Berset et al.}, 2011; ^{Cropley et al.}, 2006). Interestingly, ^{Gilboa and Revelle} (1994) reported that even small negative everyday life events can activate preservative cognition lasting up to 11 hours after the event. What does this finding mean with respect to mobile-flexible workers? Based on a systematic review by ^Tavares (2017), working hours of employees in flexible work models have extended to evenings, nights and weekends. Thus, it can be assumed that mobile-flexible workers have less time for the recovery process. There may only be few, if any, hours between work and sleep and sleep itself, an important aspect of recovery, may be limited. Even in already very limited off-times, where the stressors are no longer present in the true sense, they may interfere with recovery in the form of work-related rumination. ^{Donahue et al.} (2012) show that rumination has mediated the effect of obsessive passion for work on exhaustion. The meta-analytical studies from ^{Blanco-Encomienda et al.} (2020) and ^{Jimenez et al.} (2021) demonstrate that rumination is positively related to exhaustion and can act as a mediator, especially when recovery is restricted during non-working time.

Purpose of this Study

According to our current knowledge, there is a gap in the literature concerning how the extension of working hours and rumination are related to exhaustion among mobile-flexible workers based in A-FOs. The purpose of this study is to explore this research gap. Thus, we hypothesized–on a daily basis–that an extension in working hours is positively related to exhaustion in the evening (H1; long working hours → exhaustion) as well as to rumination after the end of work (H2; long working hours → rumination). Furthermore, we expected that rumination after work is positively related to exhaustion in the evening (H3; rumination → exhaustion). Finally, we assumed that rumination after work mediates the association between the long working hours and exhaustion in the evening (H4; long working hours → rumination → exhaustion).

Sample and Procedure

The present longitudinal study selected only participants who had worked in an A-FO for one year or longer. Therefore, suitable A-FO organizations were directly contacted and asked to participate. Only those organizations in the German-speaking part of Switzerland that comprised the most important design elements of activity-related work, such as different zones (e.g., retreat, social, or quiet work) were contacted. All study participants provided informed consent and the study design was approved by the ethics committee of the University of Bern, Switzerland (Nr. 2010-08-00003). Only participants employed in Switzerland at a minimum of 40%, calculated on the basis of Swiss full-time employment, were included in the study. Originally, a total of 91 subjects were contacted and 57 of them participated by filling out the questionnaires, which corresponds to a response rate of 53%. For conclusive data, only participants who had completed more than three consecutive days of panel data were considered (^Graham, 2009; ^{Ohly et al.}, 2010), resulting in the exclusion of 5 participants. Three further participants were excluded due to an unrealistic questionnaire completion time (less than 4 minutes). The final analysis included 49 participants. The average number of questionnaires filled out by each participant was M = 4.53, SD = 2.32.

The final sample included 17 (35%) male and 32 (65%) female participants. The average age was 36.8 years old, ranging between 20 to 59 years (SD = 10.93). Participants were employed in different sectors including administration, public services, education, health and social services. Data collection took place between January and February 2020. Once participants completed the general online questionnaire, they entered the online diary study and filled in daily questionnaires from Monday to Sunday. The sample size at Level 1 varies for each variable (e.g., extended working hours, rumination, and exhaustion, daily hassles at home, social support at home) due to missing values. A well-known approach to deal with missing data is complete case or listwise deletion, where each case with missing data in any of the analyzed variables is removed from the analysis (^Graham, 2009; ^Allison, 2003). Due to listwise deletion the sample size of Level 1 (day level) ranged from 227 to 238, and the sample size of Level 2 (person level) ranged from 33 to 38.

Instruments

Questionnaire: Before collecting the diary measures, participants completed a general questionnaire assessing demographic and occupational background (Level 2 variables). Completion of the general questionnaire was obligatory before proceeding with the diary study.

Diary Study: A morning questionnaire was used for the data collection of Level 1 variables to measure changes in rumination and exhaustion and to record the long working hours. The morning questionnaire had to be completed on all workdays before work. The morning questions were related to the experience of the previous day's nonwork time.

Daily long working hours. The ratings on the extension of working hours were created based on the four-item subscale from the self-endangering work behavior questionnaire of Krause et al. (2015). The questions were rated on a 5-point Likert scale, ranging from 1 (rarely/none) to 5 (frequently/constantly). The item was introduced by “To what extent do the following statements apply to you? Yesterday evening after work, I had ...”. An example item is “Foregone sufficient sleep in favor of work.” The internal consistency for the scale was (α = 0.87).

Daily rumination. Rumination was measured with the three-item scale by ^{Mohr et al}. (2006). The questions were rated on a 5-point Likert scale, ranging from 1 (rarely/none) to 5 (frequently/constantly). The item was introduced by “To what extent do the following statements apply to you? Yesterday evening, I had . . .”. An example item is “Difficulty relaxing after work.” The Cronbach's alpha coefficient was (α = 0.93).

Daily exhaustion. In the morning questionnaire, exhaustion was measured with four items following a German version of the Oldenburg Burnout Inventory from Demerouti and Bakker (1999). The items of the exhaustion subscale refer to general feelings of emptiness, over-taxation from work, a strong need for rest, and a state of physical exhaustion (^{Demerouti et al.}, 2001; ^{Demerouti et al.}, 2003). The questions were rated on a 4-point Likert scale, ranging from 1 (totally disagree) to 4 (totally agree). An example item is “Yesterday evening after work . . . I felt worn out and weary.” The internal consistency for the exhaustion scale was (α = 0.90).

Control variables. Exhaustion has been suggested to differ depending on age and gender (^{Elfering et al}., 2018); wherefore, these two variables were controlled to avoid influencing the statistical model.

Daily hassles at home have been linked to psychological distress (^{Elfering et al.}, 2021). To make sure that the relationship between long working hours and exhaustion was not influenced by daily hassles at home, we controlled for this variable. Daily hassles at home were measured daily (Level 1) every morning through one item: “Have you had any conflicts yesterday with the following people?” The five answer categories were “spouse,” “children,” “friend,” “family member”, and “other person.”

Social support at home has shown evidence for decreasing workplace stress (^Wang, 2018). Thus, we controlled social support at home. This control variable was assessed daily (Level 1) every morning with one item: “Did the following people help you yesterday with problems or concerns?” Five answer categories (“spouse,” “children,” “friend,” “family member”, and “other person”) were given.

Statistical Analysis

The statistical analysis was done using the software program R 4.0.2 (R Core Team, 2020). The linear and logistic regressions were analyzed to estimate the association between rumination and long working hours as a predictor variable and exhaustion as an outcome variable. The tests were two-tailed. For hypotheses testing, multilevel analysis was used. Multilevel data with repeated measurements is nested within individuals and leads to a two-level model with series of repeated measures at the day level (within persons; n = 227-238) and individual persons at the person level (between persons; n = 33-38). In this study, the day level is represented by Level 1 variables, whereas person-level variables represent Level 2. The sample size on Level 1 varied for different variables (i.e., long working hours, rumination, exhaustion, daily hassles at home, and social support at home), but the variation did not affect the multilevel analysis (^Allison, 2003; ^{Bell et al.}, 2018; ^Graham, 2009). The predictor on Level 1, long working hours, was group-mean centered, as was the mediator rumination. This allowed the effect of the variable to be interpreted in relation to its own mean across all days. The age variable on Level 2 was grand-mean centered. Gender and the outcome variable exhaustion remained uncentered. It is worth mentioning that participants' perceptions for each workday may have varied depending on what they experienced. Therefore, we controlled the Level 1 variables for the previous day's score (i.e., long working hours, rumination, exhaustion). For this purpose, the stability of the previous day's scores was calculated using means of autocorrelations for each day except for the first day of the diary study. For multilevel mediation testing, the Monte Carlo method was used. This method is recommended since it allows for multilevel indirect effect analyses and parametric bootstrapping and is suitable for producing more accurate results for small sample sizes (^{Selig & Preacher}, 2008).

Advantages and Limitations of the Study

Psychological responses are dependent on the situational context and are bound to daily fluctuations (^{Ohly et al.}, 2010). Thus, the main advantage of the present study is its longitudinal design. By using a diary design, we analyzed the interindividual daily variations (^{Ohly et al.}, 2010).

Although this study provided interesting findings, some limitations should be noted.

First, studies that use self-report measures are susceptible to common method bias (^{Podsakoff et al.}, 2003).

Second, if all variables in a study are based on one measurement method, significant results are likely to be tainted by common method variance (^{Donaldson & Grant-Vallone}, 2002).

Third, a further disadvantage is that all included variables concern the evening of the last day, estimated the next morning. It would be advisable to measure the variables in a daily morning as well as a daily evening questionnaire (^{Donaldson & Grant-Vallone}, 2002).

Furthermore, according to ^{Maas and Hox} (2004), for robust estimations of fixed effects in multilevel modeling, a sample of at least 50 at Level 2 is recommended. The present sample size at Level 2 (n = 33-38) provides only limited statistical power for the required analyses. In this respect, we would like to mention that the original study was planned for a longer period with more subjects. The questionnaire period fell in the first lockdown declared by the state of Switzerland because of COVID-19. We waited for a year, but to date many of the study participants still work mainly from home and are very rarely, if at all, in the A-FOs.

Future Research

Little research has been conducted on how job-related resources such as social support from supervisors or individual cognitive coping strategies influence mental health in mobile-flexible employees. For example, ^{Bakker and van Wingerden} (2021) showed that employees who use their imagination, fantasy, and humor to proactively reshape work tasks reported fewer depressive symptoms and fatigue as well as increased vitality. Consequently, ^{Bakker and van Wingerden} (2021) emphasized that self-determined behavior can be used to buffer the impact of negative thoughts and improve psychological and social resources. Interestingly, the JD-R model has recently been extended to include the role of personal resources assisting the individual in the balance of work demands and job-related resources (^{Demerouti & Nachreiner}, 2018). We strongly suggest future research to examine mobile-flexible employees' individual coping strategies, such as playful work design (^{Bakker & van Wingerden}, 2021) or mindfulness exercises (^{Bakker & Demerouti}, 2017; ^Sonnentag, 2018).

Furthermore, we suggest examining the role of job-related resources such as appreciation by colleagues or supervisor support among mobile-flexible employees. Future research should investigate whether conclusions from present and previous findings on the relationship between job characteristics, job demands, and job resources need additions or modifications, as it must be assumed that new behavioral strategies or job demands have resulted of the COVID-19 pandemic.

Practical Implications

The present results provide preliminary evidence that daily long working hours increase rumination and exhaustion among mobile-flexible employees in A-FOs. The results further suggest that daily rumination predicts exhaustion. What can we conclude from these relationships? The shift from traditional to mobile-flexible working arrangements in the NWW is not only reflected in modern flexible working arrangements, but also in novel office design concepts (^{Appel-Meulenbroek et al.}, 2020; ^{Engelen et al.}, 2019; ^{Rolfö et al.}, 2018). While A-FOs are being built in many places in Switzerland, it is crucial to consider the change in stressors associated with introducing new work environments and new forms of work into practice. ^{Derks and Bakker} (2010) describe the increased technical flexibility as a disproportionate burden that entails high pressure to respond to every request coupled with the feeling of constant accessibility. Additionally, several scientists (^{Costa et al.}, 2004; ^{Demerouti et al.}, 2014; ^Kvande, 2007) argued that one reason why long working hours are likely to be a major stressor among mobile-flexible employees is the shift from external control and guidance to more self-control and self-management (^{Costa et al.}, 2004; ^{Demerouti et al.}, 2014; ^Kvande, 2007). ^{Costa et al.} (2004) state that flexible working models increase demands on self-management and thus on coping strategies (^Graf, 2018; ^{Kattenbach et al.}, 2010; ^{Krause et al.}, 2015). However, ^{Derks and Bakker} (2010) emphasized something very essential: The ongoing digitalization is neither a work demand nor a job resource in itself. Rather, it can become either work demand or resource through the way new working arrangements used by the individual. Therefore, to prevent negative effects of mobile-flexible work, employees should be given the chance to practice alternative coping mechanisms that support recovery processes, especially psychological detachment during non-working time (^Sonnentag, 2018). When designing mobile-flexible work models, employers should be aware that mobile-flexible working arrangements involve a high level of self-regulation and the necessity of developing strategies to detach from work (^Kvande, 2007). Because these skills cannot be taken for granted in every individual, they require the organizations' support and development (^{Bakker & Demerouti}, 2017; ^Kniffin, 2021; ^Sonnentag, 2018). Moreover, psychological processes such as motivation, dependent on job resource availability, are likely to lead to increased work engagement and positive outcomes in terms of both work output and the employee's emotional state. The provision of sufficient job resources should always be considered when designing future flexible working models (^{Bakker & Demerouti}, 2017).

Authors’ contributions

All authors contributed to the study’s conception and design. Material preparation, data collection, and analysis were performed by [MSW]. The first draft of the manuscript was written by [MSW], and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Ethical approval

Ethical approval (Ethics No. 2010-08-00003) was obtained from the Ethics Commission of the University of Bern, Switzerland, prior to data collection.

Declaration of images

Health Promotion Switzerland granted the authors permission to reproduce and publish their images as Figure 2 in this publication as an example of an activity-oriented flexible office (A-FO).