Will after-hours technology-mediated work make employees leave? A mediated moderation model

2.1 Motivation for ATW and turnover

2.2 Employees' motivation for ATW and work–life conflict

2.3 The moderating role of time spent on ATW

Different from most prior research contexts where one’s motivation for a behavior highly relates to the subsequent behavior (Sheeran, 2002; Sheeran and Webb, 2016; Deci and Ryan, 2002), in the context of ATW, one’s motivation for ATW does not necessarily relates to one’s actual engagement in ATW. For instance, high motivation for ATW does not mean more time spent on ATW, because the actual happening of ATW cannot be fully determined by employees. It may also depend on the actual work demand, expectations of the organization, availability of related technology and the availability of coworkers using the same technology (Mazmanian et al., 2013; Mellner, 2016). If there is no work that needs to be done after hours or no available contacts (Mazmanian et al., 2013), even though employees have a strong motivation for ATW, ATW will not happen. Besides, low motivation for ATW does not mean less time spent on ATW. This is because, even though employees dislike ATW and intend to separate their work and life, they may still choose to engage in ATW to fulfill their job obligation, establish their professional image, help their organization to maintain business competency and obtain respect and liking of their colleagues (Mazmanian et al., 2013; Ollier-Malaterre et al., 2013; Van Zoonen and Rice, 2017). Hence, in the context of ATW, employees' motivation for ATW is independent of the actual time they spend on ATW, and they could separately or jointly influence employees' turnover intention.

Specifically, we argue that how employees' intrinsic motivation and extrinsic motivation for ATW affect their turnover intention depends on the actual time spent on ATW. For instance, no matter what kind of motivation is held by the employee, if ATW rarely happens, then the impact of intrinsic and extrinsic motivations for ATW on employees' turnover may be weakened or even disappear. If the time of ATW is long, the impact of these two motivations for ATW on employees may be more salient. In this situation, employees are facing extended work hours and become physically and mentally exhausted (Schlachter et al., 2018). Butts et al. (2015) argued that employees may react differently to such physical and mental exhaustion caused by long time of ATW. For employees who want to protect their life from unwanted work intrusion but feel obliged to engage in ATW due to organization’s expectation (Barley et al., 2011), that is being extrinsically motivated for ATW, long time spent on ATW will exacerbate their negative reactions and dampen their positive reactions. Employees who think work is inherently fun and strive to pursue career success will voluntarily devote time and energy to ATW, that is being intrinsically motivation for ATW. As a result, the negative feelings caused by long time spent on ATW will be weakened and the positive reactions will be exacerbated. Therefore, we argue that the impact of employees' intrinsic motivation and extrinsic motivation on work–life conflict and turnover intention will be moderated by the time spent on ATW.

As aforementioned, intrinsically motivated employees will feel that ATW offers them a chance to be better connected to colleagues, develop their potential and take control over their work and life, which reduces their intention to leave. Hence, the more they engage in ATW, the more likely for them to experience these positive feelings. In fact, these positive feelings can neutralize the negative outcomes of ATW, such as physical and mental exhaustion and thus alleviate employees' urge to leave the job. Specifically, because of spending more time on ATW, employees will feel strongly connected to colleagues at work and embedded into the work settings. According to the Job Embeddedness Theory, these employees are enmeshed with an invisible web, and if they leave the job, they need to make great sacrifices (Mitchell et al., 2001). The cost of leaving will decrease their turnover intention (Lo, 2015). Meanwhile, intrinsically motivated employees are willing to invest great efforts into ATW (Deci and Ryan, 1985). They may actively monitor the notices of new work messages, promptly reply to the messages, generously offer help to colleagues and be friendly and patient during their communication (Butts et al., 2015; Shujaat et al., 2019). Hence, for these employees, the happening of ATW offer a stage to show their passion, kindness and competence and therefore can help them establish a good professional image (Schlachter et al., 2018), which may enable them to obtain more work resources and support in the future (Van Zoonen and Rice, 2017; Shujaat et al., 2019), reducing their desirability to leave the organization. Therefore, we propose that,

As discussed earlier, extrinsically motivated employees engage in ATW to avoid punishment or obtain rewards (Deci and Ryan, 2002). Therefore, their commitment and loyalty to the organization are low. When they are required to spend a lot of time on ATW which makes them feel physically and mentally exhausted, they will naturally consider quitting the job without worrying whether their turnover will cause damage or inconvenience to the organization (Cohen, 1993). For those who engage in ATW due to punishment, every episode in which they are forced to respond to the after-hours work demand may make them feel that their organizations do not care about their well-being (Kurtessis et al., 2017). Hence, the more work requests they received after hours, the more negative effect, such as anger and dissatisfaction (Butts et al., 2015), toward their organization will be incurred, which will weaken their attachment to the organization and increase their turnover possibility. Since these employees' ATW is compelled by their organization, they may behave passively during their ATW. For example, they may choose to delay their reply to the work message or reply to the message impatiently or irresponsibly (Butts et al., 2015). As the ATW increases, such passive behavior will ruin their work reputation and relationship with leaders and colleagues (Schlachter et al., 2018; Oksa et al., 2021) and result in a loss of work resources and promotion opportunities, which will increase their intention to leave the organization (Lo, 2015). Therefore, we propose that,

For intrinsically motivated employees, ATW is aligned with their intrinsic needs. Therefore, they can experience positive feelings from ATW, such as happiness, pleasure and satisfaction (Benlian, 2020). The more they engage in ATW, the more positive feelings they will experience, which will promote both their work and life engagement and performance (Van Zoonen and Rice, 2017) and result in low conflict between work and life. Moreover, intrinsically motivated employees are willing to invest time and efforts in ATW (Scott et al., 1997). Their enthusiasm, activeness and responsiveness during ATW can help them accumulate relational capitals and cognitive capitals at work, leading to lower communication cost and easier knowledge sharing with co-workers (Davison et al., 2018; Derks et al., 2016). Hence, when the need of ATW is high, these employees can finish their work more effectively and thus can save more time, attention and energy for their life domain (Michel et al., 2011; Edwards and Rothbard, 2000). Meanwhile, a good relationship with supervisors and colleagues can offer emotional and instrumental support that helps employees to mitigate the pressures from the life domain (Van Zoonen et al., 2020). In fact, intrinsically motivated employees are those who may voluntarily choose to spend long time on work to pursue personal achievement. Korn et al. (1987) posit that such employees are similar to the hyper-performers who are resourceful, adaptable, creative and good at stress management. Thus, they will be more capable of taking care of both their work and life and experience less work–life conflict (Scott et al., 1997). Thus, we propose that,

Time spent on ATW can also moderate the relationship between employees' extrinsic motivation for ATW and their perception of work–life conflict. Firstly, when the need of ATW is low, it means that ATW will not cost employees too much energy and that they can still re-engage in life after finished work. In this situation, the work demand is still in the acceptable range to employees and therefore will not incur extreme reaction in employees (Barley et al., 2011). Therefore, the impact of extrinsic motivation for ATW on work–life conflict will not be very strong. Secondly, extrinsic motivation for ATW reflects employees' negative attitude for ATW and preference for separating work and life. Hence, they have the desire to protect their non-work time from the unwanted work demand and limit the duration of such intrusion (Powell and Greenhaus, 2010). As a result, as the time spent on ATW increases, they will experience negative feelings such as anger, dissatisfaction and unhappiness, which may further hamper their interactions with family members and performance in the life domain, causing work–life conflict (Butts et al., 2015). Butts et al. (2015) posited that preference of work–life separation will exacerbate the negative emotions and lessen the positive emotions in employees during their ATW. In fact, when there are no ATW requests, extrinsically motivated employees may have their own arrangement for family life. However, when there are a lot of ATW, the employees may have to cancel or reschedule their family activities (e.g. seeing a movie or going to a picnic). In such cases, both the employees' life satisfaction and their family members' life satisfaction will be hampered (Benlian, 2020), incurring work–life conflict. Hence, we propose that,

2.4 The mediating role of work–life conflict

When work–life conflict happens, it means that one may have allocated too much time and attention to the work domain which hindered their performance and well-being in the life domain (Greenhaus and Beutell, 1985). Firstly, the extended work hour and increased workload will make employees feel fatigue, burnout, physically and emotionally exhaustion and even cause health problems, such as psychological distress and insomnia (Schlachter et al., 2018; Chen and Karahanna, 2018; Buruck et al., 2020; Park et al., 2020), which will hamper their well-being and makes them what to quit the job. Secondly, in such cases, employees will not have enough recovery time to replenish their physical and mental resources that are depleted by the work demand (Zijlstra et al., 2014; Mazmanian et al., 2013). As a result, these employees may feel that the current job cannot help them achieve their career goal and fully develop their potential. Thirdly, when too much work demand intrudes their life, employees may feel that they are neglected by their organization and therefore develop low organizational commitment and job satisfaction, which will increase their intention to leave (Carlson et al., 2000; Frone et al., 1992; Higgins et al., 1992). Accumulating evidence has confirmed that employees' perception of work–life conflict can arouse their turnover intention (Maertz and Boyar, 2011; e.g. Frone et al., 1992; Carlson et al., 2000; Adams et al., 1996; Wright et al., 2014).

Based on the impact of work–life conflict on turnover intention, we posit that work–life conflict can not only mediates the impact of intrinsic and extrinsic motivations on turnover intention, but also mediates the interaction effect of intrinsic motivation and ATW time and the interaction effect of extrinsic motivation and ATW time on turnover intention. If ATW time is short, both intrinsic and extrinsic motivations will have a relatively weak impact on turnover intention. This is because, regardless the motivation for ATW, if no work needs to be done after hours, it will not interfere with employees' lives to induce work–life conflict. Employees with intrinsic motivation for ATW are those who can experience pleasant feelings from ATW. For example, they may consider ATW an opportunity to help them fit into the work group or to show their talent (Schlachter et al., 2018). Therefore, when ATW time is high, they will consider it as their opportunity. As they fit into the work group and being recognized by their competencies, their achievements and satisfaction in the work-role may alleviate their feeling of loss of private time caused by ATW or even induce work to life enrichment that helps to improve their life role performance (Casper et al., 2018; Jiang and Men, 2017). In such situations, employees' perception of work–life conflict will be weakened and thus have less intention to leave the job. Extrinsically motivated employees dislike ATW and prefer to have their own plans for leisure time if there is low need of ATW (Azar et al., 2018). When more time is spent on ATW, the positive impact of extrinsic motivation on turnover intention will be enhanced. This is because ATW compels them to postpone or cancel their original plan for family, hampering the life satisfaction of the employees and even their family members (Powell and Greenhaus, 2006). Moreover, when employees are compelled to undertake too much ATW, they are likely to experience negative emotions, which have been found to result in low performance in both work and life (Benlian, 2020), intensifying work–life conflict and making employees consider quitting their current job. Accordingly, we argue that the interaction effect of ATW time and employees' motivation (intrinsic motivation and extrinsic motivation) on turnover intention can be mediated by employees' perception of work–life conflict. Therefore, we propose that,

3.1 Data collection

We conducted a survey to collect data. The target population of our survey was all WeChat users who had used WeChat to complete job tasks after hours. With over 1.3 billion monthly active users, WeChat is the second most popular messaging app in the world (Statistica, 2023). We collected data in China because China is the home market of WeChat and WeChat has been well integrated into the work and life of Chinese people. In a survey released by China Economic Herald in April 2022, more than half of the respondents admitted that, during their remote work, they used WeChat to conduct the work, which implied the prevalence of WeChat in ATW.

The participant recruitment was based on two inclusion criteria. First, they needed to have used WeChat for ATW, which ensured that they had relevant knowledge and experience of our research context. To meet this condition, we asked participants to report their time spent on ATW in the past two weeks. If the answer was zero, the participant would be excluded. Second, participants needed to be currently employed so that their reported turnover intention was realistic in the sense that they knew what it entailed if they left their current job. If not, the participant would be excluded.

The questionnaire was administered via an online survey platform. We first uploaded the survey to the survey platform and then released the link of the survey in the WeChat Friend Circle and WeChat groups to invite people to take the survey. Overall, we have received 515 completed questionnaires. After deleting the questionnaires with invalid answers, we got 484 valid returns. The participants were from 32 different regions of China, which alleviated biases related to regional economic and cultural differences. The participants came from a variety of industries and company types, job positions and age groups, which demonstrated the diversity of our sample. Appendix 1 presents their detailed demographics data.

3.2 Measures

All the measurements (see Appendix 2) were adopted from previous studies and rated by a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree). For ATW time (TIME), we adopted the scales from Richardson and Thompson (2012). In our research, there are three after-hours situations: time before work on workdays, time after work on workdays and weekends or holidays. Respondents were asked to choose how much time they spent on ATW per day in these three situations. They could choose from “0–15 min”, “16–30 min”, “31–59 min”, “1–2 h” to “above 2 h” according to their actual experience. For intrinsic motivation (IM) and extrinsic motivation (EM), we adopted the scales from Michael et al. (2013) and Hansen and Levin (2016) to measure how an individual is intrinsically or extrinsically motivated for ATW. The scale of work–life conflict (WLC) was adopted from Netemeyer et al. (1996). The last item of this scale was dropped due to low factor loading, and we used the other four items to measure WLC. Turnover intention (TI) was measured by the scale developed by Mobley et al. (1979) and two of the items were reverse coded. Before the formal data collection, we conducted a pre-test with a group of Information Systems (IS) scholars. Based on their feedback, we slightly revised the wording of some questions to improve readability.

4.1 Measurement validation

Cronbach's alpha and composite reliability were calculated to evaluate the reliability of the measurements. The reliability scores of all constructs were above 0.7 (Table 1), suggesting a high level of reliability (Nunnally, 1978). The average variance extracted (AVE) of all constructs are above 0.60, indicating that the principal constructs have captured more construct-related variance than error variance (Hair et al., 1998). The correlations among all the constructs are below 0.70 and smaller than the square root of AVE, indicating sufficient discriminant validity (Bagozzi et al., 1991). As Table 2 shows, all the item loadings on the assigned construct are above 0.70 (except the first item loading of extrinsic motivation which is 0.654) and greater than the cross-loading on other constructs, which suggests satisfactory convergent validity and discriminant validity of our measures (Chin et al., 2003).

4.2 Common method bias

To eliminate the concern of common method bias, we followed Podsakoff et al. (2003) to apply both procedural and statistical remedies. First, procedurally, we used anonymous measures which did not show the variable names in the questionnaire and used reverse questions for some items. Second, statistically, we conducted the Harman's single factor test and the single latent method factor test (Podsakoff et al., 2003). For the Harman’s single factor test, all the items were entered a Principal Component Analysis (PCA). Among all the extracted factors, the one with the largest eigenvalue only accounted for 30.94% of the variance, indicating that the variance of our data is not coming from one common method source. Therefore, common method bias is not a severe problem in our research. For the single latent method factor test (Podsakoff et al., 2003), we created a latent method factor consisting of the indicators from the largest factor extracted from the PCA and included this method factor into our main analysis as a control variable. The findings are consistent before and after the inclusion of this method factor, suggesting the absence of common method bias.

4.3 Hypothesis testing

We used SmartPLS for hypothesis testing. Table 3 presents the PLS results. First, intrinsic motivation and extrinsic motivation have different impacts on turnover intention and work–life conflict. As the result of Model 1 shows, intrinsic motivation for ATW has a negative direct impact (β = −0.117, p < 0.05), while extrinsic motivation for ATW has a positive direct impact (β = 0.195, p < 0.001) on turnover intention, supporting H1a and H1b. As shown by the result of Model 2, intrinsic motivation for ATW negatively influences work–life conflict (β = −0.429, p < 0.001), while extrinsic motivation for ATW positively influences work–life conflict (β = 0.134, p < 0.01). Therefore, H2a and H2b are both supported.

Second, the moderation effects of ATW time on the impact of intrinsic motivation and extrinsic motivation are different. The interaction term of ATW time and intrinsic motivation on turnover intention is not significant (β = −0.068, ns), failing to support H3a. The interaction term of ATW time and extrinsic motivation on turnover intention is significantly positive (β = 0.150, p < 0.05), suggesting that the positive impact of extrinsic motivation on turnover intention is strengthened as more time is spent on ATW. Hence, H3b is supported. The simple slopes (Figures 2 and 3) also show that the impact of intrinsic motivation on turnover intention is not affected by ATW time, while the impact of extrinsic motivation on turnover intention is stronger when ATW time increases. Similarly, the interaction effect of Time spent on ATW and intrinsic motivation on work–life conflict is not significant (β = −0.015, ns), failing to support H4a. The interaction effect of ATW time and extrinsic motivation on work–life conflict is significant and positive (β = 0.108, p < 0.05), indicating that ATW time can enhance the positive impact of extrinsic motivation on work–life conflict. Therefore, H4b is supported.

To test the mediated moderation effect proposed in H5a and H5b, we followed the three-step procedures proposed by Muller et al. (2005). Muller et al. (2005) contend that there is a mediated moderation effect if the following three conditions are satisfied. First, the interaction effect of the independent variable (IV) and the moderator (MO) on the dependent variable (DV) is significant. Second, IV has significant influence on the mediator (ME) and the interaction effect of IV and MO on ME is significant Third, when ME is added into the model and the interaction effect of ME and MO on DV is controlled, ME has significant effect on DV and the interaction effect of IV and MO on DV is reduced or become insignificant.

In our research, the interaction effect of intrinsic motivation and ATW time is not significant on either turnover intention (β = −0.068, ns) or work–life conflict (β = −0.015, ns). Therefore, there is no mediated moderation effect between intrinsic motivation and turnover intention. H5a is not supported. Nevertheless, we found that the interaction between extrinsic motivation and ATW time on turnover intention is significant (see Model 1 in Table 3), which satisfies Condition 1. Moreover, the impact of extrinsic motivation on work–life conflict is significant, and the interaction between extrinsic motivation and ATW time on work–life conflict is significant (see Model 2 in Table 3), which satisfies Condition 2. We also found that when work–life conflict is added into the model and the interaction between ATW time and work–life conflict on turnover intention is controlled, work–life conflict shows a significant impact on turnover intention (β = 0.455, p < 0.001) and the interaction between extrinsic motivation and ATW time on turnover intention becomes insignificant (β = 0.101, ns) (see Model 3 in Table 3), which satisfies Condition 3. Altogether, the results suggest that work–life conflict fully mediates the interaction between extrinsic motivation and ATW time on turnover intention, supporting H5b.

The impacts of control variables such as age, gender, education, job position and job tenure are reported in Table 3. The results show that age and education have a significant negative impact on turnover intention, indicating that older employees and employees with a higher education level will have lower turnover intention. We also found that age has a significant negative impact on work–life conflict, suggesting that older employees will unlikely perceive work–life conflict. In addition, the interaction between intrinsic motivation and extrinsic motivation for ATW has a significant effect on turnover intention (β = 0.089, p < 0.05) but no significant effect on work–life conflict (β = 0.093, ns).

5.1 Theoretical implication

This study contributes to IS and management research in four aspects. First, it offers a motivational perspective to explain whether ATW will arouse employees' turnover intention. Existing literature has examined the impact of time, frequency and intensity of ATW on employees' turnover intention as well as other outcomes (Ferguson et al., 2016; e.g. Wright et al., 2014; Belkin et al., 2020). A common finding is that the more employees engage in ATW, the more likely they will experience burnout and exhaustion (Ferguson et al., 2016), which will hamper their performance in both work and life (Chen and Karahanna, 2018) and therefore make them want to quit the job (Wright and Nishii, 2007). Our research contends that how employees react to ATW depends on their underlying motivation for ATW. We investigated the impact of employees' intrinsic motivation and extrinsic motivation for ATW on their turnover intention. The results show that intrinsic motivation for ATW decreases, while extrinsic motivation for ATW increases, employees' turnover intention. This finding indicates that, besides the degree of engagement in ATW, employees' motivation for ATW can also affect the turnover intention induced by ATW, which provides a new perspective to understand the impact of ATW on employees and organizations.

Second, this research advances our knowledge about ATW by examining the interaction effect of time spent on ATW and employees' motivation. Existing research only separately examined the main effect of time spent on ATW and some factors that may contribute to the formation of employees' intrinsic motivation for ATW such as perceived job autonomy (Moore, 2000; Carillo et al., 2021), and extrinsic motivation for ATW such as perceived organizational norm and expectation (Richardson and Benbunan-Fich, 2011). Different from the literature, we adopted an integrative approach to examine the interaction effect of employees' motivation and the time spent on ATW. Our results show that more time spent on ATW does not always cause detrimental consequences to organization. It can only intensify the positive impact of extrinsic motivation on work–life conflict and turnover intention but has no moderation effect on the impact of intrinsic motivation on work–life conflict and turnover intention. This finding indicates that the impact of ATW time depends on the nature of employees' motivation for ATW. Under the same amount of time spent on ATW, distinct types of motivations and different intensity levels of the motivation will result in various influences on employees.

Third, our research opens the black box of the relationship between employees' motivation for ATW and turnover intention by illustrating the mediation effect of work–life conflict. As an outcome commonly seen in the literature of ATW (Ragsdale and Hoover, 2016; Köffer et al., 2014), work–life conflict has been found to be a strong predictor of employees' turnover intention (Hom et al., 2017; Lee et al., 2017; Griffeth et al., 2000; Alkahtani, 2015), but researchers have rarely examined the mediating role of work–life conflict in the relationship between ATW and employees' turnover intention. Only Ferguson et al. (2016) and Wright et al. (2014) have used work–life conflict to explain the impact of time spent on ATW on employees' turnover intention, but they did not consider the impact of employees' motivation for ATW. Our research shows that work–life conflict can mediate the impact of employees' intrinsic motivation and extrinsic motivation for ATW on turnover intention and the interaction effect of extrinsic motivation and time spent on ATW on turnover intention. This finding not only deepens our understanding regarding how work–life conflict happens and then results in employees' turnover intention in the context of ATW, but also reveals that work–life conflict is an important mechanism through which ATW engenders impacts on employees.

Lastly, our research offers a more holistic view regarding the impact of ATW by considering employees' motivations for ATW. Existing studies have been mostly focused on the dark side of ATW. They found that high engagement in ATW will result in work–life conflict (Ferguson et al., 2016), turnover intention (Wright et al., 2014), job dissatisfaction (Wright et al., 2014; Diaz et al., 2012), low organizational commitment (Turel et al., 2011), burnout (Wright et al., 2014), exhaustion (Chen and Karahanna, 2018) and anxiety (Becker et al., 2021). In contrast, our research finds that intrinsic motivation for ATW reduce employees' work–life conflict and turnover intention, and this impact will not be affected by the time spent on ATW. Hence, for employees who can perceive autonomy, competence and relatedness during ATW, ATW can produce positive outcomes and benefit the organization. This finding provides a more comprehensive understanding about the impact of ATW on employees and helps to reconcile the inconsistent findings in the ATW literature regarding the consequences of ATW by considering the differential influences of employees' motivations.

5.2 Managerial implication

While previous studies on ATW have mostly examined how the duration, frequency and intensity of ATW influence employees' work and nonwork outcomes (Ferguson et al., 2016; e.g. Wright et al., 2014; Belkin et al., 2020), we contend that employees' motivation to engage in ATW also matters. The findings of our research have important implications for practitioners.

First, our research shows that intrinsic motivation for ATW decreases work–life conflict and then indirectly decreases turnover intention while extrinsic motivation for ATW increases work–life conflict and turnover intention. These findings suggest that ATW does not always result in undesirable outcomes and that the impact of ATW depends on how employees are motivated. For employees who dislike ATW and feel that they are forced to engage in ATW, ATW will induce work–life conflict and turnover intention. Contrarily, employees who enjoy ATW will experience low work–life conflict and turnover intention because they consider ATW as pleasant, satisfying, enjoyable and a viable way for self-achievement. This is consistent with the literature of workaholic, which contends that not everyone suffers from stress due to long work hours and that whether employees feel frustrated or fulfilled about long work hours may depend on why employees spend so much time on work (Porter, 2001). Therefore, we suggest managers to take measures to help employees develop more intrinsic motivation for ATW. For example, manager can organize workshops or lectures about career development to cultivate employees' interests in ATW regarding how to fulfill ambitions faster by taking advantage of ATW, or they can integrate gamification features into ATW to increase employees' pleasant feelings when working after hours. For organizations, the need of ATW or the time of ATW is uncontrollable, because it often depends on the need of their business operations and the need of customers. To accomplish business objectives without losing productivity and talents, it is imperative for organizations to implement effective interventions to cultivate employees' intrinsic motivation for ATW to alleviate the negative feelings caused by ATW when ATW is inevitable.

Second, we find that time spent on ATW can intensify the impact of extrinsic motivation for ATW on work–life conflict and turnover intention. This result indicates that, for employees who feel compelled by their organization to engage in ATW, when they must spend a lot of time on ATW, they are very likely to feel an intense work–life conflict and thus have strong intention to leave the job. To avoid this problem, managers can take two strategies. If the organization must heavily rely on ATW to maintain its business operations, that is when heavy ATW is inevitable, managers can rely on “soft control” measures (Truss et al., 1997; Kelman and Hong, 2016) such as gamification or a honest conversation with employees to express their appreciation to improve the ATW experience or mitigate uncomfortable feelings caused by excessive ATW, rather than “hard control” measures (Kelman and Hong, 2016; Truss et al., 1997) such as sanctions to force their employees to engage in ATW. For the organizations where ATW must be enforced by sanctions, managers need to ensure that the actual time spent on ATW is not too long to reduce employees' perception of work–life conflict and intention to leave the job. For example, managers can set a “real off-work time”, after which employees are not obligated to respond to any work-related messages.

Finally, we find that work–life conflict not only mediates the main effects of intrinsic motivation and extrinsic motivation on employees' turnover intention, but also mediates the interaction effect of extrinsic motivation and time spent on ATW on employees' turnover intention. This implies that employees' perception of work–life conflict is an important precursor of turnover intention in the context of ATW. Hence, if an organization has a strong need for ATW and most employees engage in ATW under fear of being punished or losing competitiveness, managers need to pay close attention to their employees' perception of work–life conflict. Measures such as offering boundary management training (Fenner and Renn, 2010; Jostell and Hemlin, 2018), family-friendly policy (Chou and Cheung, 2013) and supervisory support (Harris et al., 2015) can be adopted, which will not only offer instrumental support to help employees deal with the work–life conflict but also convey the massage to employees that they are cared and valued by the organization (Alkahtani, 2015). As a result, the turnover intention caused by high ATW time in extrinsically motivated employees may be reduced.

5.3 Limitations

This research has two limitations. First, it only examines the separate impacts of intrinsic and extrinsic motivations for ATW on turnover intention. However, researchers have pointed out that intrinsic and extrinsic motivations can co-exist and interact with each other to influence human behaviors (Scott Rigby et al., 1992). Future research should examine whether intrinsic and extrinsic motivations for ATW are complements or substitutes when affecting employees' turnover intention. Second, previous research has shown that individuals with different cultural backgrounds may react differently to the blurred boundary between work and life (Aryee et al., 1999). Our data was collected in China where people tend to embrace the culture of collectivism and have relatively weak awareness to protect their private life from ATW (Triandis, 2018). Future research needs to study samples from other cultures to assess the generalizability of our findings.