The COR theory contends that people actively avoid the potential or actual loss of their own resources and work to accumulate and defend them (Hobfoll, 1989). Pressures related to career development and workplace ostracism may result in the depletion of an employee’s own resources (Mitchell et al., 2019), stimulate negative behaviours such as PP, and prevent constructive actions such as innovation (Chen et al., 2020; Zhang et al., 2021). The COR theory points out that when people are threatened or have lost resources because of pressure, they will take countermeasures (Crant, 2000; Hobfoll, 1989), like adopting organisational citizenship behaviours to help avoid further loss of resources (Halbesleben & Bowler, 2007). Personality can also help individuals seek resources to make up for the excessive loss of resources, so it can alleviate ego depletion (Yu et al., 2022a). By describing the stress-coping process from the standpoint of personal resources, the theory offers a theoretical explanation of perceived stress and AP, as well as the moderating role of personality on ego depletion.

According to the ego depletion theory, a person’s ability to practice self-control may decline as a result of using up all of their available resources (Baumeister et al., 1998). Workplace stresses brought on by unjustified workloads and abusive leadership may cause employees’ resources to be exhausted and further lead to negative feelings and actions, like emotion exhaustion and aggressive behaviours (Barlett et al., 2016; Hagger et al., 2010; Wang et al., 2022). People with depleted egos will make an effort to replenish their own resources and adapt themselves (Tice et al., 2007; Webb & Sheeran, 2003). This theoretically supports the interpretation that ego depletion and AP are associated (Zhang et al., 2022).

According to Chu and Choi (2005), AP is the behaviour of purposefully making delayed decisions to complete the task before the deadline to produce desirable outcomes. Active procrastinators view pressure as a motivator and an incentive, and they feel good while under pressure (Choi & Maron, 2009).

The degree to which a person senses unpredictable, uncontrollable, and overloaded situations or events occurring in life is known as perceived stress (Cohen et al., 1983). When people see stressful situations as dangerous and believe they lack resources to handle them, it might have a negative effect. The difficulty and urgency of job responsibilities place high stress on employees (Chen, 2020), which leads to burnout and health issues (Hao et al., 2015; Tatsuse et al., 2019). According to the COR theory, there are two ways to respond to the loss of resources: to acquire new resources or to suspend existing resource consumption (Hobfoll, 1989). The former response is to actively seek and acquire new resources to compensate for the loss. This can involve identifying alternative sources of resources or finding ways to replenish or replace the depleted resources. The latter response is to suspend or reduce the consumption of existing resources. This means individuals may prioritise and conserve their remaining resources rather than using them up quickly. Examples of such actions include minimising threats, using the adaptive targeting process, rearranging priorities (Folkman, 2008; Folkman & Lazarus, 1985), and adjusting needs (e.g. control, tolerate) that go beyond their own personal resources (Kaiseler et al., 2014).

Active procrastination can be viewed as a reorganisation of cumulative work ordering the more important tasks back on. In the context of civil servants with high stress, AP can be seen as a strategic approach to task management. By purposefully postponing important tasks, civil servants create a temporary reprieve from immediate demands, allowing them to focus attention and limited resources on critical duties. This can serve as a means of conserving resources and regaining focus. By creating a sense of time pressure, civil servants may be more motivated and focused when approaching the postponed tasks. The impending deadline can act as an incentive to work efficiently, utilising their limited resources in a concentrated manner. It is critical to understand that AP differs from PP, which involves unproductive delays and avoidance of responsibilities. Active procrastination involves a purposeful decision to delay tasks and leverage time pressure as a motivational tool for more focused and efficient completion. By engaging in AP, civil servants with high stress may strategically reorganise their workload, enabling them to recover energy, prioritise critical duties, and accomplish tasks within the given constraints. This approach acknowledges the need to manage limited resources effectively and optimise performance under challenging circumstances. Therefore, we suggest the following hypothesis:

Ego depletion refers to a process of temporary decline in one’s capability and willingness to control himself and/or herself, which can damage personal resources (Baumeister et al., 1998; Hagger et al., 2010). The COR theory states that people use their limited resources to fulfil different work requirements (Hobfoll, 1989; Westman et al., 2004). This study suggests that perceived stress can influence the ego depletion of civil servants. Perceived stress in the workplace can lead to more frequent dynamic resource consumption for self-control to cope with work demands and situations (Schmidt & Neubach, 2007). In addition, dealing with stressful situations requires civil servants to expend a significant amount of energy to adjust their emotions and cognitions, which could exhaust their resources (Diestel & Schmidt, 2011). Furthermore, prior studies have shown that individual ego depletion increases when higher stress is perceived (Diestel & Schmidt, 2011). We propose the following hypothesis in accordance with the aforementioned discussion:

The ego depletion theory suggests that people can replenish their exhausted resources (Baumeister et al., 1998). People who suffer from excessive ego depletion, for example, may exhibit cognitive bias because of a lack of resources and control, which could then lead to behaviours that undermine self-regulation, such as lower work engagement and increased procrastination (Fischer et al., 2007). People can also take steps to stop additional resource loss and work towards the goal by systematically controlling various tasks in a planned way (Hobfoll, 1989). Engaging in midday relaxation activities or planned task control can help restore resources. Active procrastination can be seen as one such strategy for resource recovery. When civil servants experience significant ego depletion, they may feel a lack of competence or ability to complete tasks. By deliberately and actively procrastinating, they aim to restore their resources, regain control, and avoid further chaos. We suggest the following hypothesis in light of the given discussion:

Civil servants who experience stress at work will use their limited resources of self-control to handle the emotional disorders caused by stress and pressure. If they are unable to stop resource loss or supplement resources, the speed of resource loss will be further accelerated, resulting in ego depletion (Hobfoll, 1989; Schmidt & Neubach, 2007). Ego depletion can have detrimental effects on job performance and organisational outcomes. Without proper prevention and management of resource loss, civil servants may struggle to stay organised and maintain productivity. In such situations, intentional procrastination can serve as a strategy to prevent and repair resource loss. By putting work off and engaging in brief relaxation or other activities immediately that provide a temporary break from work demands, civil servants may be able to restore their depleted resources and enhance their self-management abilities. This can help them to regain a sense of control, improve focus and concentration, and ultimately improve their job performance. In the light of this, we suggest the following hypothesis:

The Big Five personality model is widely accepted and includes five broad domains (i.e. extroversion, agreeableness, conscientiousness, openness, and neuroticism) (McCrae & Costa, 1997). Extroversion (E) reflects how talkative, forceful, and gregarious a person is (Barrick & Mount, 1991). Those high in extroversion enjoy adventure, social, and other enthusiastic activities (McCrae & Costa, 1997), show more positive emotions (Costa & McCrae, 1980), and focus more on the positive aspects of events. Positive emotions can offset ego depletion (Tice et al., 2007), which helps people feel less emotional exhaustion. Individuals with high extroversion tend to retain a pleasant psychological state under pressure because of their high mood, and stress has less impact on their psychological resource consumption. Therefore, we infer that perceived stress has different impact on ego depletion depending on extroversion and propose the following hypothesis:

Conscientiousness (C) reflects dependability; people with high conscientiousness are more accountable, conscientious, and possess considerable willpower (Barrick & Mount, 1991). According to Ching et al. (2014), conscientiousness can make people act seriously, feel optimistic, and not get fatigued easily. Furthermore, highly conscientious people have strong self-control and self-restraint, and are able to manage stress in an orderly fashion (Qi et al., 2013). Individuals who are more conscientious are less sensitive to perceived stressful events and have higher self-control. They can better exercise self-control and reduce the loss of psychological resources. As a result, we propose the following hypothesis:

Openness (O) refers to a person’s level of imagination, culture, curiosity, originality, broad-mindedness, intelligence, and artistic sensitivity (Barrick & Mount, 1991). People with a high level of openness are more creative, curious, and have greater pressure elasticity (Williams et al., 2009). They are more inclusive and more ready to accept the status quo when under stress (Chen et al., 2015). According to studies, people with high openness personality traits can reassess stressors, adjust themselves to generate positive emotions, and replenish consumed resources (Qi et al., 2013; Williams et al., 2009). It is speculated that people who have greater openness will be more adept at adjusting their own state and mobilising positive emotions when facing stress, resulting in lower levels of ego depletion and psychological energy consumption. Therefore, we propose the following hypothesis:

According to Barrick and Mount (1991), agreeableness (A) means being polite, adaptable, reliable, friendly, cooperative, forgiving, soft-hearted, and patient. Studies show that agreeableness is associated with positive psychological factors such as trust and resilience (Deng et al., 2021). Those with high agreeableness experience higher levels of adaptation, lower levels of perceived stress, and less mental resource depletion, such as energy depletion (Wang et al., 2015). They have the potential for interpersonal intimacy (Graziano et al., 1996), and are adversely correlated with colleagues’ lack of reciprocity (Petrou et al., 2011). They can adapt quickly (Han & Zhou, 2020) and thus use fewer psychological resources to resist stress, thus reducing ego depletion. Therefore, we propose the following hypothesis:

Neuroticism (N) means feeling nervous, anxious, furious, humiliated, sentimental, and insecure (Barrick & Mount, 1991). Individuals with high neuroticism are prone to experiencing higher psychological stress and anxiety (McCrae & Costa, 1997). High neuroticism tends to result in negative feelings such as worry and sadness (Yu et al., 2021), which is known as ‘distress proneness’ (Wilson et al., 2005, p. 12) or ‘psychological stress’ (Qi et al., 2013, p. 102), and is easy to cause ego depletion. Therefore, individuals with high neuroticism may perceive a high level of pressure and find it difficult to self-control the negative emotions caused by stress, making ego depletion more severe. Therefore, we propose the following hypothesis:

Hypothesis 4 proposes that ego depletion mediates the link between perceived stress and AP, while in hypotheses 5a–5e, individuals also differ in the ego depletion brought on by stress depending on their different personalities and different personality levels. Therefore, this study further proposes a moderated mediation model to examine how perceived stress affects AP.

When faced with difficult situations, individuals with high extroversion are optimistic and confident (Ashton et al., 2002), and they will typically focus on the positive aspects and respond positively (Qi et al., 2013). As a result, civil servants have fewer psychological resources to fight against stress, resulting in less AP to deal with resource consumption (Kim et al., 2017). In contrast, individuals low in extroversion have fewer positive emotions in stressful situations. Instead, they lead to an increase in anxiety, impulsivity, psychological pressure, and ego depletion. As a result, people tend to be more willing to try AP to compensate for ego depletion (Qi et al., 2013).

As conscientiousness is related to capacity for self-control (Hou & Gu, 2017), individuals with high conscientiousness are driven by goal-orientation and have strong self-control abilities in the face of stress, so the level of ego depletion is relatively low (Di Fabio, 2006). However, when faced with pressure, individuals with low conscientiousness are difficult to regulate quickly, resulting in higher ego depletion, which is more likely to require AP to cope with the loss. Therefore, a high-conscientiousness personality may weaken the positive influence of individuals’ perceived stress on their own AP through ego depletion.

Individuals with high openness can handle stress better and accept their situation more quickly. They can reevaluate the stress and mobilise positive emotions to regulate it. As a result, fewer control resources are consumed, and resources remain for existing work, which in turn can reduce AP. Conversely, when confronted with stressful events, those with low openness are more prone to stress and generate higher levels of pressure (Qi et al., 2013); as a result, there is more ego depletion caused by resisting stress and then more AP.

Individuals with high agreeableness are more adaptable and will take active measures to deal with stress (Li et al., 2000). Therefore, when they perceive pressure, they can adapt to the situation as soon as possible and effectively resist ego depletion, thus reducing the occurrence of AP. However, individuals with low agreeableness tend to develop negative emotions, such as anger (Lian & Guo, 2017). Such emotions are prone to high ego attrition, which may produce more AP to resist the loss of resources.

When under pressure, individuals with high neuroticism are prone to emotional instability, psychological burnout, anxiety, and other issues (Bakker et al., 2006). It is difficult to change, making it simple for them to induce ego depletion and engage in AP. However, individuals with low neuroticism have relatively stable emotions and are less sensitive to stress, so they may produce fewer negative emotions (Qi et al., 2013), less ego depletion, and further less AP. We suggest the following hypotheses based on the aforementioned analysis:

The data was gathered via an online questionnaire survey and samples were received from Chinese government departments. The participants were civil servants from several provinces, including Xinjiang Province, Sichuan Province, and Fujian Province. The respondents were recruited from a Master of Public Administration (MPA) training course by one author, and these students helped us advertise our study among their colleagues. Participants could fill out the questionnaire at their own convenience, and participation was completely voluntary. After their informed consent, participants completed a short questionnaire regarding their feelings of perceived stress, ego depletion, AP, and their five big personality traits.

Out of a total of 716 distributed questionnaires, 651 valid questionnaires were returned, achieving a 90.9% effective response rate. The demographic characteristics of the sample are as follows: male: 46.6%, female: 53.4%. The average age of civil servants was 34.32 years, with a standard deviation (SD) of 7.23. The educational characteristics of the sample are: junior college degree and below: 15.8%; bachelor’s degree: 77.1%; master’s degree: 6.9%; doctoral degree and above: 0.2%. The mean number of working years was 10.99, with a SD of 7.73. In terms of income level, the average monthly income below ¥5000 accounted for 47.2% of the total, and the monthly income above ¥5000 accounted for 52.8% of the total. And most of them are staff members (39.9%).

Firstly, this study used Harman’s single-factor test to examine the potential common method bias, confirmatory factor analyses (CFA), the average variance extracted (AVE), and composite reliability (CR) to evaluate discriminant validity. The CFA was conducted using Mplus 8.3 software.

Secondly, the means, SD, and Pearson correlation coefficients among the key variables were calculated.

Finally, this study tested our hypotheses. For predictions regarding the direct effect of perceived stress on active procrastination (Hypothesis 1) and the mediating effect of ego depletion between perceived stress and AP (Hypotheses 2–4), hierarchical regression analysis was first used by selecting different independent variables to build different regression models to compare to illustrate the mediation effect. The Hayes Process Macros were used to test the mediation effect again (Hypothesis 4) and the moderating effect (Hypotheses 5a–5e). We used the general path analytic framework (bootstrapping procedure, 1000 iterations) to test the moderated mediating effect (Hypotheses 6a–6e). The Statistical Package for the Social Sciences (SPSS 26.0) software was used for all hypothesis validation.

The scales utilised in this study were modified versions of scales that had been used by other researchers. Following the standard translation-back translation procedure, the items’ Chinese versions were made. A 7-point Likert scale, ranging from 1 (strongly disagree) to 7 (strongly agree), was used to rate the Big Five personality traits. Perceptual stress, AP, and ego depletion were assessed on a 5-point Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree).

Perceived stress was measured with the 4-item scale developed by Motowidlo et al. (1986), which contains items such as ‘I feel a lot of pressure from this job’ and ‘My work is under a lot of pressure’. The total scores of the 4 items were deemed an indicator of the extent of each civil servant’s perceived stress, with higher scores representing more stress. The scale’s Cronbach’s α was 0.78 in this study.

Active procrastination was measured with a scale produced by the Chinese scholars Ni et al. (2011), which included 15 items for four factors that were more appropriate for the Chinese context: result satisfaction, preference to pressure, purposeful choice to procrastinate, as well as capacity to satisfy deadlines (e.g., ‘I will not do badly when the tasks approach the deadline’) that were more suitable for the Chinese context. The total scores of the 15 items were deemed an indicator of the extent of each civil servants’ active procrastination with higher scores representing higher AP. The scale’s Cronbach’s α was 0.81 in this study.

Ego depletion was measured with the Depletion Scale (Twenge et al., 2004), which consists of five items (e.g. ‘I feel exhausted’). The total scores of the five items were deemed an indicator of the extent of each civil servants; ego depletion, with higher scores representing more ego depletion. The scale’s Cronbach’s α was 0.89 in this study.

The Big Five personality traits were measured using a 40-item scale adapted from Saucier (1994), which contains five dimensions: openness, conscientiousness, extroversion, agreeableness, and neuroticism (such as ‘organised’, ‘cooperative’, and ‘relaxed’), and each was scored on a 7-point scale (1 representing ‘strongly disagree’ and 7 representing ‘strongly agree’). Cronbach’s α for openness, conscientiousness, extroversion, agreeableness, and neuroticism were 0.84, 0.81, 0.76, 0.78, and 0.82, respectively.

As control factors, following the procedure used by researchers (Xie et al., 2018; Yu et al., 2022a; Zhou, 2020), we controlled personal factors of gender (1 = male, 2 = female), age, and education level (1 = junior college and below, 2 = bachelor’s degree, 3 = master’s degree, 4 = doctoral degree and above), and work-related factors of working years, income, and civil servant’s position (1 = director of a division, 2 = deputy director, 3 = section chief, 4 = deputy section chief, 5 = principal staff member, 6 = senior staff member, 7 = consultant, 8 = assistant consultant, 9 = staff member).

As variables in this study were all measured via self-reported scales, this raised the possibility of common method bias effects. Accordingly, the potential common method bias that usually appears in self-reported data was examined with Harman’s single-factor test (Zhou & Long, 2004). We extracted a total of 12 components with feature roots larger than 1. Less than 50% of the criteria were explained by the first component, whose explanatory power was only 24.73%. Consequently, there was no common method bias in our research (Podsakoff et al., 2012).

Using statistical software Mplus 8.3, CFA was carried out to evaluate the fit indexes for the eight-factor model against those of other combination models so as to verify discriminant validity between variables (Anderson & Gerbing, 1988). In order to evaluate the model, Bentler (1990) advised using a sample with at least 5 instances of each parameter; our sample size satisfies this recommendation.

The Tucker-Lewis Index (TLI > 0.95 for good level > 0.90 for acceptable level), the ratio of chi square goodness of fit to degrees of freedom (χ²/df < 3 for acceptance), the Root Mean Square Error of Approximation (RMSEA, 0.05 for good level, 0.08 for acceptable level), the comparative fit Index (CFI), and the (RMSEA, 0.05 for good level, 0.08 for acceptable level) were the four indices we used to evaluate model fit (Jackson et al., 2009; Marsh et al., 2004). According to the results (shown in Table 1), the eight-factor model fit best (χ²/df = 2.30; CFI = 0.88; TLI = 0.90; RMSEA = 0.05), while the single-factor model fit poorly (χ²/df = 7.13; CFI = 0. 45; TLI = 0.43; RMSEA = 0.10) (Sörbom, 1989). To further verify the discriminant validity, the AVE and the CR were calculated simultaneously in this study. The AVE for the perceived stress scale was 0.60 and the CR was 0.86; the AVE for the AP scale was 0.62 and the CR was 0.86; the AVE for the ego depletion scale was 0.70 and the CR was 0.92; the AVE for the extroversion scale was 0.48 and the CR was 0.88; the AVE for the conscientiousness scale was 0.52 and the CR was 0.89; the AVE for the openness scale was 0.48 and the CR was 0.88; the AVE for the agreeableness scale was 0.52 and the CR was 0.89; the AVE for the neuroticism scale was 0.53 and the CR was 0.90. According to the standards given by Fornell and Larcker (1981), CR should be > 0.7 and AVE should be > 0.5. However, they added that even if AVE is lower than 0.5 but CR is higher than 0.6, the construct’s convergent validity is still sufficient. According to given criteria, our eight variables were distinct, indicating good discriminant validity.

Table 2 shows the important variables means, SD, AVE, and Pearson correlation coefficients. Perceived stress correlated moderately positively with ego depletion (r = 0.48, p < 0.001), weakly positively with active procrastination (r = 0.26, p < 0.001), and weakly positively with neuroticism (r = 0.28, p < 0.001). Perceived stress had a weakly negative correlation with extroversion (r = -0.24, p < 0.001), a weakly negative correlation with conscientiousness (r = -0.18, p < 0.001), and a weakly negative correlation with openness (r = -0.23, p < 0.001). But it had no correlation with agreeableness (r = -0.07, p = 0.09). Ego depletion correlated strongly positively with active procrastination (r = 0.55, p < 0.001) and with neuroticism (r = 0.51, p < 0.001) and moderately negatively with extroversion (r = -0.36, p < 0.001), conscientiousness (r = -0.41, p < 0.001), agreeableness (r = -0.38, p = 0.09), and openness (r = -0.40, p < 0.001). Active procrastination correlated moderately positively with neuroticism (r = 0.49, p < 0.001) and moderately negatively with extroversion (r = -0.34, p < 0.001), conscientiousness (r = -0.50, p < 0.001), openness (r = -0.32, p < 0.001), and agreeableness (r = -0.48, p < 0.001). According to Cohen’s (1988) suggestion, our results are almost consistent with our theoretical expectations.

To evaluate the direct effects of the variables and the role as a mediator of ego depletion, hierarchical regression was used (Baron & Kenny, 1986; Wen & Ye, 2014). Table 3 presents the results. The main effects among the variables were evaluated after controlling for the six demographic factors of gender, age, working time, education level, income, and position (M1). The results demonstrated that perceived stress had a significantly positive impact on civil servants’ AP (M2: β = 0.25, p < 0.001). H1 is therefore supported.

We then examined the mediating effect. Ego depletion was significantly impacted positively by perceived stress (M6: β = 0.49, p < 0.001), and with active procrastination (M3: r = 0.54, p < 0.001). Based on the main effect model (M2), after adding ego depletion as an independent variable, the regression coefficient for perceived stress was no longer significant (M4: β = -0.02, p > 0.05), whereas ego depletion continued to have a significant positive influence (M4: β = 0.55, p < 0.01). The fact that ego depletion is a full mediator of perceived stress influencing AP is shown by the ∆R² value of 0.23. H2–H4 are therefore supported.

In addition, the effects of mediation were computed using the PROCESS macros running in SPSS 26.0 software (Hayes, 2013), and the six demographic variables were controlled. The bias-corrected confidence intervals (CIs) of each coefficient were calculated after 5000 repeated samplings with returns (bootstrapping procedure), and the results have been presented in Table 4. When perceived stress and ego depletion were taken into consideration, the direct effect was not significant (-0.01, NS). The indirect effect of perceived stress on AP through ego depletion is estimated at 0.18 with 95% CIs [0.15, 0.22]. The indirect effect is significant because the CIs do not include 0, further supporting H4.

The Hayes Process Macro was used to test the moderating effect in SPSS 26.0 software (Hayes, 2013). To prevent potential multicollinearity, we first centred the independent variable (perceived stress) and the moderators (extroversion, conscientiousness, openness, agreeableness, and neuroticism) (Aiken & West, 1991). Meanwhile, we also controlled the influence of other moderators when we tested the moderating effect.

The interaction of perceived stress and extroversion was significantly negatively associated with ego depletion (β = -0.11, p = 0.003) after controlling for the demographic variables, as well as the influence of other moderators (four-factor model fitting index of PS, ED, AP, E: χ²/df = 5.07, CFI = 0.81, RMSEA = 0.08); the interaction of perceived stress and conscientiousness was significantly negatively associated with ego depletion (β = -0.09, p = 0.011. Four-factor model fitting index of PS, ED, AP, C: χ²/df = 4.79, CFI = 0.83, RMSEA = 0.08); the interaction of perceived stress and openness was significantly negatively associated with ego depletion (β = -0.10, p = 0.003. Four-factor model fitting index of PS, ED, AP, O: χ²/df = 5.08, CFI = 0.82, RMSEA = 0.08); the interaction between perceived stress and neuroticism was significant positively (β = 0.08, p = 0.012. Four-factor model fitting index of PS, ED, AP, N: χ²/df = 4.79, CFI = 0.83, RMSEA = 0.08). H5a, H5b, H5c, and H5e were verified. H5d was not verified considering the fact that the interaction between perceived stress and agreeableness was not significant (β = -0.04, p = 0.287), showing that agreeableness had no moderating influence. The moderating effect can be seen in Figure 1.

To examine whether there were significant differences in mediating routes at different levels of extraversion, conscientiousness, openness, and neuroticism, this study adopted a general path analytic framework (which combines moderated regression analysis and path analysis) (Edwards & Lambert, 2007). According to this method, two equations were established based on the hypotheses to test the moderated mediation effect, where E is extroversion, C is conscientiousness, O is openness, N is neuroticism, PS is perceived stress, ED is ego depletion, and AP is active procrastination. These variables were formerly mean-centred. SPSS 26.0 was used to estimate the results of each parameter in Equations 1 and 2 (Yu et al., 2022b), as shown in Table 5:

ED=a05+ax5PS+az5E(C,O,N)+axz5[PS*E(C,O,N)]+em5[Eqn 1]

AP=b04+bx4PS+bm4ED+ey4[Eqn 2]

The coefficients of 1000 bootstrap samples were estimated using the constrained nonlinear regression (CNLR) approach. The constrained nonlinear model used the default loss function to minimise the sum of squared errors and calculate the parameter estimates using the least square method. These estimated coefficients were then imported into Edwards and Lambert’s (2007) Excel template, where the coefficients, differences, and 95% CIs of the mediation model’s first, second, direct, indirect, and total effects with high and low moderator levels were obtained. The results are shown in Table 6.

Table 6 shows significant differences in the indirect effect between the high and low groups of extroversion (r = -0.084, p < 0.01), conscientiousness (r = -0.071, p < 0.01), openness (r = -0.072, p < 0.01), and neuroticism (r = 0.070, p < 0.05). Thus, each of the four moderators significantly moderated the mediating effect. Specifically, ego depletion was a full mediator when the four moderators were high or low (the indirect effect was significant compared with the direct effect), while the mediation effect was stronger with low extroversion (the difference in indirect effects was -0.084, p < 0.01), conscientiousness (the difference in indirect effects was -0.071, p < 0.01), or openness (the difference in indirect effects was -0.072, p < 0.01), and with high neuroticism (the difference in indirect effects was 0.070, p < 0.05). H6a, H6b, H6c, and H6e were supported.