- Email: [email protected]
Relationship between resilience, stress and burnout among civil servants in Beijing, China: Mediating and moderating effect analysis
Personality and Individual Differences 83 (2015) 65–71
Contents lists available at ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Relationship between resilience, stress and burnout among civil servants in Beijing, China: Mediating and moderating effect analysis Shuwei Hao, Wei Hong ⇑, Honghong Xu, Lili Zhou, Zhongyao Xie Medical Psychology Center, Institute of Medical Humanities, Peking University, Beijing, PR China
a r t i c l e
i n f o
Article history: Received 23 September 2014 Received in revised form 17 March 2015 Accepted 27 March 2015 Available online 9 April 2015 Keywords: Resilience Stress Burnout Mediating effect Moderating effect Civil servants Structural equation modeling
a b s t r a c t This study aimed to explore the mediating and moderating effect in the relationship between resilience, stress and burnout among civil servants of Beijing, China. A cross-sectional study was conducted among civil servants in Beijing. Totally 541 civil servants completed a self-report questionnaire including three scales measuring civil servants’ resilience, stress and burnout. The data were analyzed with correlation, multiple regression and structural equation modeling. The results revealed that work stress rather than life and health stress could significantly predict burnout. Resilience played a partial mediating role between work stress and burnout, that is, work stress had both a direct and an indirect, via resilience, impact on burnout. Work stress played a partial mediating role between resilience and burnout, thus, resilience could prevent the development of burnout by relieving work stress, in addition to directly relieving it. Moreover, resilience was a moderator between work stress and burnout, and it could serve as a buffer to mitigate the adverse effects of work stress. These results suggest that resilience could be a positive personality trait for alleviating or eliminating work stress and combating burnout of civil servants of Beijing. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Civil servants who perform the function for the management of the state and government are a relatively special occupational group in China. With the development of the society and the deepening of the reform, the government’s management of civil servants has become more and more standard, more and more stringent. As a result, many civil servants feel ‘‘stressed out’’.
Abbreviations: RTSCA-R, Resilient Trait Scale for Chinese Adults-Revised; OP, optimism; AC, acceptance; CA, controllability; SU, supportiveness; CSSS, Civil Servants Stress Scale; MD, management and development; LR, life relationships; WR, working relationships; HR, health and responsibility; ES, economic stress; WL, working load; WS, work stress; LHS, life and health stress; ES1, two items related to work stress from economic stress dimension; MBI-GS, Maslach Burnout InventoryGeneral Survey; EE, emotional exhaustion; CY, cynicism; PE, professional efficacy; RWS, resilience and work stress interaction term; CA*MD, controllability and management and development product term; OP*WR, optimism and work relationships product term; AC*WL, acceptance and working load product term; SU*ES1, supportiveness and ES1 product term. ⇑ Corresponding author at: Medical Psychology Center, Institute of Medical Humanities, Peking University, 38, Xueyuan Rd., Haidian District, Beijing 100191, PR China. Tel.: +86 10 82801543; fax: +86 10 82802512. E-mail addresses: [email protected] (S. Hao), [email protected] (W. Hong), [email protected] (H. Xu), [email protected] (L. Zhou), [email protected] (Z. Xie). http://dx.doi.org/10.1016/j.paid.2015.03.048 0191-8869/Ó 2015 Elsevier Ltd. All rights reserved.
As is well known, stress is an important negative predictor of human health, and could result in some stress-related illnesses, such as burnout (Boyas & Wind, 2010; Hsu, Chen, Yu, & Lou, 2010). Burnout is an important phenomenon that derives from chronic emotional responses and interpersonal stressors that occur at work, and burnout syndrome is determined by the dimensions of emotional exhaustion, cynicism, and reduced professional efficacy (Maslach & Jackson, 1981; Maslach, Schaufeli, & Leiter, 2001). Although burnout has increased over the last decade, not everyone at the same workplace develops burnout, even facing the similar work stress. It suggested that individual factors may contribute to this phenomenon. In line with the transactional model of stress and coping theory, it indicated that some people are more vulnerable to stressful situations than others, especially those with high depressive symptoms, or have certain personality traits, such as high neuroticism or low optimism (Lazarus & Folkman, 1984). That is to say, contrary personality traits, such as low neuroticism or high optimism, are just the reverse. Consequently, not everyone perceives some life situations as stressful and the causes related to the individual variables may be the personality traits, for example, resilience as a personality trait may follow the same law. Resilience is an interactive phenomenon that is inferred from findings indicating that some individuals have a relatively good outcome despite having experienced serious stresses or adversities –
66
S. Hao et al. / Personality and Individual Differences 83 (2015) 65–71
their outcome being better than that of other individuals who suffered the same experiences (Rutter, 2013). Although there is still considerable debate regarding definitions of resilience, trait orientation suggests that resilience is a personality trait that helps individuals cope with adversity and achieve good adjustment and development (Hu, Zhang, & Wang, 2015). Consistent with the transactional model, resilience as a personality trait has been suggested to play an important role in the relationship between work stress and burnout. Several studies on nurses (García & Calvo, 2012; Mealer et al., 2012) and doctors (Taku, 2014) found resilience was a protective factor that can buffer the relationship between risk factors and burnout. Furthermore, the mediating effect of resilience on life events/learning stress and learning burnout of students (Wang & Zhang, 2011; Zhang, 2013) and the moderating effect of resilience on role stress and job burnout of teachers (Xu, Zhang, Sun, & Tian, 2013) have been proved. For civil servants, burnout threatens not only their own health, but also the government’s work efficiency and social stability. Therefore, it should be given more attention. Nonetheless, the mediating and moderating effect of resilience on stress and burnout among civil servants, especially, the mediating effect of stress on resilience and burnout has not been previously explored. According to the previous studies, we hypothesized that (1) the relationship between stress and burnout is mediated by resilience, and that (2) the relationship between resilience and burnout is mediated by stress, and that (3) resilience moderates the relationship between stress and burnout among civil servants in Beijing, China. Figure 1 depicts the hypothesized conceptual model of this study. 2. Methods 2.1. Participants and procedure A cross-sectional study with a representative sample consisting of 600 civil servants in Beijing, China, were carried out from April 2013 to June 2014. This study was approved by the Biomedical Ethics Committee of Peking University. We contacted the administrators of the departments affiliated to Beijing municipal government, and they assisted us in recruiting the civil servants who had agreed to participate. Participants were invited to complete a self-report questionnaire on resilience, stress, and burnout, and completed the questionnaire in a meeting room environment in their own departments. The questionnaire was anonymous, and the purpose of the research was explained to the respondents. Additionally, their confidentiality was assured, and they were informed that they had the right to drop out whenever they wanted. 541 civil servants completed the questionnaire, the other 59 civil servants were excluded due to missing values in
the relevant items in the questionnaire. The effective response rate was 90.17%. 2.2. Measures 2.2.1. Resilient Trait Scale for Chinese Adults-Revised (RTSCA-R) RTSCA-R (Hao, 2014; Hao & Hong, 2014) consists of 29 items and four dimensions: optimism (OP), acceptance (AC), controllability (CA) and supportiveness (SU). Items are assessed on a 4-point scale that ranges from ‘‘not true at all’’ (scored 1) to ‘‘true nearly all the time’’ (scored 4). Seven items are reverse-scored to reduce acquiescence biases. Higher scores on the scale reflect greater resilience. The reliability and validity of RTSCA-R were satisfied in Chinese civil servants. In the present study, Cronbach’s a coefficient for the total scale was 0.93. Cronbach’s a coefficients of OP, AC, CA and SU ranged from 0.75 to 0.90. 2.2.2. Civil Servants Stress Scale (CSSS) CSSS (Hao, 2014; Hao, Xu, Zhou, Xie, & Hong, 2014) which is designed to assess the perceived stress of Chinese civil servants consists of 38 items and six dimensions: management and development (MD), life relationships (LR), working relationships (WR), health and responsibility (HR), economic stress (ES) and working load (WL). Additionally, the CSSS can be divided into two subscales (Hao, 2014): work stress (WS) subscale and life and health stress (LHS) subscale. Work stress subscale consists of 23 items including MD, WR, WL and two items of ES, and life and health stress subscale consists of 15 items including LR, HR, and three items of ES. Each item is using an 11-point Likert scale ranging from 0 (no pressure) to 10 (maximum pressure). Higher scores on CSSS indicate higher levels of stress of civil servants. In this study, Cronbach’s a coefficient for the total scale was 0.95. Cronbach’s a coefficients of the six dimensions ranged from 0.80 to 0.94, and Cronbach’s a coefficients of work stress subscale and life and health stress subscale were 0.95 and 0.91, respectively. 2.2.3. Maslach Burnout Inventory-General Survey (MBI-GS) The Chinese version of MBI-GS (Li & Shi, 2003) consists of 15 items, using a seven-point Likert-type scale ranging from 0 (never) to 6 (every day). The 15-items scale has three subscales including emotional exhaustion (EE), cynicism (CY), and professional efficacy (PE), which reflect different aspects of the burnout syndromes. Higher scores on EE and CY subscales and lower scores on PE
Table 1 Frequency distribution of civil servants’ demographical characteristics (N = 541). Variables
Groups
n
%
Gender
Male Female Missing value 629 30–39 40–49 P50 Missing value Single Married Junior college or under Undergraduate Graduate Missing value Level 7 Level 8 Level 9 Level 10 Level 11 Level 12 Missing value
274 263 4 80 264 139 51 7 84 457 35 411 91 4 17 23 154 163 162 14 8
50.6 48.6 0.7 14.8 48.8 25.7 9.4 1.3 15.5 84.5 6.5 76.0 16.8 0.7 3.1 4.3 28.5 30.1 29.9 2.6 1.5
Age
Marital status Education
Administrative level
Fig. 1. Mediation and moderation model of resilience, stress and burnout.
67
1 0.35*** 0.35*** 0.33*** 0.12** 1 0.89*** 0.55*** 0.51*** 0.55*** 0.44*** 0.14**
13
1 0.65*** 0.77*** 0.67*** 0.48*** 0.51*** 0.46*** 0.13**
12
1 0.45*** 0.41*** 0.44*** 0.89*** 0.24*** 0.26*** 0.24*** 0.06
11
1 0.34*** 0.48*** 0.62*** 0.76*** 0.51*** 0.37*** 0.36*** 0.30*** 0.17***
10
1 0.50*** 0.39*** 0.38*** 0.43*** 0.47*** 0.71*** 0.27*** 0.21*** 0.23*** 0.16***
9
1 0.39*** 0.56*** 0.41*** 0.73*** 0.77*** 0.94*** 0.56*** 0.46*** 0.45*** 0.48*** 0.13**
8
1 0.87*** 0.63*** 0.73*** 0.69*** 0.81*** 0.83*** 0.94*** 0.86*** 0.51*** 0.50*** 0.48*** 0.16***
7
1 0.40*** 0.74*** 0.71*** 0.36*** 0.28*** 0.30*** 0.34*** 0.20*** 0.30*** 0.32*** 0.35*** 0.30*** 0.46*** 0.34*** 0.39*** 0.29***
1 0.32*** 0.29*** 0.27*** 0.16*** 0.33*** 0.28*** 0.14** 0.20*** 0.24*** 0.24*** 0.25*** 0.35*** 0.23*** 0.29*** 0.25***
1 0.71*** 0.30*** 0.22*** 0.20*** 0.26*** 0.21*** 0.26*** 0.25*** 0.27*** 0.27*** 0.40*** 0.30*** 0.31*** 0.27***
1
0.27*** 0.22*** 0.17*** 0.22*** 0.17*** 0.25*** 0.22*** 0.25*** 0.23*** 0.32*** 0.23*** 0.28*** 0.21***
6
P < 0.01. P < 0.001. **
***
5 3 2
3.01 ± 0.37 3.18 ± 0.48 3.01 ± 0.52 2.96 ± 0.42 2.98 ± 0.43 3.86 ± 1.79 4.29 ± 2.20 1.89 ± 2.17 3.59 ± 2.32 3.88 ± 2.76 4.60 ± 2.55 4.37 ± 1.96 4.22 ± 1.94 3.30 ± 2.04 2.09 ± 1.01 2.32 ± 1.49 1.81 ± 1.39 2.08 ± 1.31 1. Resilience 2. OP 3. AC 4. CA 5. SU 6. Stress 7. MD 8. LR 9. WR 10. HR 11. ES 12. WL 13. WS 14. LHS 15. Burnout 16. EE 17. CY 18. PE
1 M ± SD Variables
Table 2 Means, standard deviations and bivariate correlations (N = 541).
4
3.1. Demographic characteristics of participants
1 0.86*** 0.64*** 0.90*** 0.80*** 0.37*** 0.27*** 0.31*** 0.34*** 0.23*** 0.31*** 0.31*** 0.34*** 0.33*** 0.48*** 0.34*** 0.39*** 0.32***
3. Results
Table 2 presents the means, standard deviations, and Pearson correlation coefficients for the RTSCA-R, CSSS and MBI-GS. For civil servants, resilience, OP, AC, CA and SU were negatively correlated with stress, MD, LR, WR, HR, ES, WL, WS and LS (r = 0.37 to 0.12, P < 0.01). Resilience, OP, AC, CA and SU were negatively correlated with burnout, EE, CY and PE (r = 0.48 to 0.21, P < 0.001). Stress, MD, LR, WR, HR, ES, WL, WS and LS were positively correlated with burnout, EE and CY (r = 0.21 to 0.55, P < 0.001). Stress, MD, LR, WR, ES, WL, WS and LS were positively correlated with PE (r = 0.12 to 0.17, P < 0.01). However, there were no significant correlations between HR and PE (r = 0.06, P > 0.05).
1 0.62*** 0.53*** 0.53*** 0.50*** 0.17***
14
Descriptive statistics, correlation analysis, and Cronbach’s a reliability estimates were conducted using the IBM SPSS Statistics version 20 (SPSS20). The hypothesized models in this study were analyzed with structural equation modeling (SEM) procedures using the Analysis of Moment Structure version 20 (AMOS20). Continuous variables were presented as mean values with standard deviations, whereas, frequencies and percentages were given for the categorical variables. Correlation analyses were computed by way of Pearson’s correlations to determine the direction and size of the relationship between all variables in the structural model. Hierarchical multiple regression analysis was conducted to examine the relationship between control variables (gender, age, marital status, education and administrative level), work stress, life and health stress and burnout. Moreover, SEM was used to analyze the mediating and moderating models. Maximum likelihood estimation was employed as a global test of models because data for all continuous variables were normally distributed. The goodness-of-fit of the models was evaluated by the chi-square (v2) statistic, root mean square error of approximation (RMSEA), goodness-of-fit index (GFI), Tucker-Lewis fit index (TLI), comparative fit index (CFI), parsimony-adjusted normed fit index (PNFI) and parsimony goodnessof-fit index (PGFI). For the RMSEA, values less than 0.10 represent an acceptable fit (Steiger, 1990). For the GFI, TLI and CFI, values greater than 0.90 are considered acceptable (Hu & Bentler, 1999; Wu, 2013). And for the PNFI and PGFI, values greater than 0.50 are considered acceptable (Wu, 2013). It should be noted that all study continuous variables were centered before SEM analysis to increase interpretability of interactions and to control for multicollinearity (Wu, Wen, & Lin, 2009).
3.2. Descriptive statistics and correlations
1 0.78*** 0.81*** 0.62***
16 15
2.3. Statistical analysis
The study population was composed of 541 civil servants with a mean age of (37.08 ± 7.93) years (range 22–58 years). Administrative ranking of civil servants in China were divided into 12 levels, and all the respondents’ administrative ranking were below level 6 in this study. The demographic characteristics of the civil servants were shown in Table 1.
1 0.70*** 0.06
17
subscale indicate higher levels of burnout. Six items of PE dimension are using reverse scoring so that higher total scores on MBIGS could indicate higher levels of burnout. The Chinese version of the MBI-GS has high internal consistency (Li & Shi, 2003). In the present study, Cronbach’s a coefficient for the total scale was 0.89. Cronbach’s a coefficients of EE, CY, and PE were 0.95, 0.89, and 0.89, respectively.
1 0.20***
S. Hao et al. / Personality and Individual Differences 83 (2015) 65–71
68
S. Hao et al. / Personality and Individual Differences 83 (2015) 65–71
3.3. The relationship between work stress, life and health stress and burnout According to the definition of burnout (Maslach & Jackson, 1981; Maslach et al., 2001), it was caused by work stress. In order to investigate whether life and health stress could also have effect on burnout, we analyzed the relationship between control variables (gender, age, marital status, education and administrative level), work stress, life and health stress and burnout using hierarchical multiple regression analysis. The result (see Appendix Table) indicated that control variables had no significant impact on burnout (P > 0.05). Work stress had significant impact on burnout (F = 19.23, B = 0.11, b = 0.47, P < 0.001), nevertheless, life and health stress had no significant impact on burnout (b = 0.05, P > 0.05).
3.4. Mediation effect of resilience on work stress and burnout In line with the results of multiple regression analysis, the variable stress was replaced by work stress in the hypothesis model (Fig. 1). According to the method of mediating effect analysis (Wen, Chang, Hau, & Marsh, 2004; Wen, Hau, & Chang, 2005; Wen & Ye, 2014), we established the hypothesized mediating effect models of resilience and work stress (Figs. 2 and 3). The mediation model of resilience and the standardized coefficients for each variable are shown in Fig. 2. Structural equation model depicting significant regression and correlation paths in the model, all the path coefficients were statistically significant at the level of P < 0.01. The fit indices for the modified model were acceptable: v2 = 250.52, df = 41, RMSEA = 0.097, GFI = 0.92, TLI = 0.91, CFI = 0.93, PGFI = 0.57, PNFI = 0.68. According to the model, burnout was significantly predicted by work stress and resilience, the standardized direct effect value of work stress ? burnout was 0.57 (P < 0.001), and the standardized
Fig. 3. The mediation model of work stress on resilience and burnout (N = 541).
direct effect value of resilience ? burnout was 0.26, (P < 0.001). Resilience was significantly predicted by work stress, the standardized direct effect value of work stress ? resilience was 0.36 (P < 0.001). The standardized indirect effect value of work stress ? resilience ? burnout was ( 0.36 0.26 = 0.0936). The standardized total effect value of work stress on burnout was (0.57 + 0.0936 = 0.6636). Hence, the mediation effect ratio was (0.0936/0.6636 100% = 14.10%). 3.5. Mediation effect of work stress on resilience and burnout The mediation model of work stress and the coefficients for each variable are shown in Fig. 3. All the path coefficients were statistically significant at the level of P < 0.01. The fit indices for the modified model were acceptable: v2 = 250.52, df = 41, RMSEA = 0.097, GFI = 0.92, TLI = 0.91, CFI = 0.93, PGFI = 0.57, PNFI = 0.68. According to the modified model, burnout was significantly predicted by work stress and resilience (P < 0.001), the standardized direct effect value of resilience ? burnout was 0.26 (P < 0.001), and the standardized direct effect value of work stress ? burnout was 0.57 (P < 0.001). Work stress was significantly predicted by resilience, the standardized direct effect value of resilience ? work stress was 0.36 (P < 0.001). The standardized indirect effect value of resilience ? work stress ? burnout was ( 0.36 0.57 = 0.2052). The standardized total effect value of resilience on burnout was ( 0.26 0.2052 = 0.4652). Therefore, the mediation effect ratio was ( 0.2052/ 0.4652 100% = 44.11%). 3.6. Moderation effect of resilience on work stress and burnout
Fig. 2. The mediation model of resilience on work stress and burnout (N = 541). Resilience, RTSCA-R; OP, optimism; AC, acceptance; CA, controllability; SU, supportiveness; work Stress, work stress subscale; MD, management and development; WR, working relationships; ES1, two items related to work stress in economic stress dimension; WL, working load; Burnout, MBI-GS; EE, emotional exhaustion; CY, cynicism; PE, professional efficacy; e1–e13, error term. All measured indicators were centered, and all the path coefficients were standardized.
In accordance with the method of latent Interactions analysis (Wen & Wu, 2010; Wu, Wen, Hau, & Marsh, 2011; Wu et al., 2009), we established the moderation model of resilience (Fig. 4). The interaction term of resilience and work stress with four measured indicators in the model. Four dimensions of RTSCA-R multiplied four dimensions of work stress subscale, respectively, which yielded four product terms (CA*MD, OP*WR, AC*WL and SU*ES1) as measured indicators of resilience and work stress interaction term. The results of SEM indicated that all the path coefficients were statistically significant at the level of P < 0.01. The fit indices for
69
S. Hao et al. / Personality and Individual Differences 83 (2015) 65–71
Fig. 4. Moderation effect of resilience model (N = 541). RWS, resilience and work stress interaction term; CA*MD, controllability and management and development product term; OP*WR, optimism and work relationships product term; AC*WL, acceptance and working load product term; SU*ES1, supportiveness and ES1 product term; e1–e16, error term. All measured indicators were centered, and all the path coefficients were standardized.
4. Discussion The present study investigated the relationship between stress, resilience and burnout among civil servants of Beijing. Consistent with the previous studies, the results of multiple regression analysis revealed that work stress rather than life and health stress could significantly predict burnout, which verified the definition of job burnout. Although the relationship between work stress and burnout is well known and frequently studied, the specific mechanism by which burnout symptoms develop have yet to be fully established. This study expands on the previous findings by demonstrating that the mediating and moderating effect between resilience, work stress and resilience among civil servants of Beijing. On the one hand, the partial mediating effect of resilience on work stress and burnout indicated that work stress could directly and indirectly, via the impairment of resilience, exacerbate burnout symptoms. This phenomenon was confirmed in the previous studies on trauma and resilience (Bonanno, Galea, Bucciarelli, & Vlahov, 2007; Zhu et al., 2012). This result suggested that resilience
Low resilience
High resilience
2.5 2 1.5 Burnout
the modified model were acceptable: v2 = 358.15, df = 84, RMSEA = 0.08, GFI = 0.92, TLI = 0.90, CFI = 0.92, PGFI = 0.65, PNFI = 0.72. According to the moderation model, burnout was significantly predicted by work stress, resilience and work stress and resilience interaction term (RWS), the standardized direct effect value (or total effect value) of resilience ? burnout was 0.27 (P < 0.001), the standardized direct effect value of work stress ? burnout was 0.59 (P < 0.001), and the standardized direct effect value of RWS ? burnout was 0.12 (P < 0.01). In addition, Fig. 5 indicated that burnout symptoms of the low resilience group were more severe than those of the high resilience group. With the increase of work stress, the protective effect was more significant. That is, resilience was a moderator in the work stress–burnout relationship.
1 0.5 0 -0.5
Low work stress
High work stress
-1 -1.5
Work Stress
Fig. 5. Graphic representation of the interaction between work stress and resilience to influence burnout. Low resilience, 2 SDs below the centered mean; high resilience, 2 SDs above the centered mean.
as a personality trait is stable, but it is not unchangeable. The reason may be due to excessive pressure resulting in individuals’ ‘‘self-regulating mechanism’’ overwhelmed, and self-adjustment failed, which could produce mental maladjustment and a series of psychosomatic symptoms (Zhu et al., 2012), such as burnout. This finding is also in line with the prior research showing that high levels of perceived stress are highly predictive of the severity of burnout symptoms (Ahola & Hakanen, 2007; Banovcinova & Baskova, 2014). On the other hand, the partial mediating effect of work stress on resilience and burnout revealed that resilience could directly reduce the level of burnout, and via the reduction of work stress, indirectly reduce the level of burnout. The effect of resilience on burnout is consistent with the results of the previous studies (Cooke, Doust, & Steele, 2013; Mealer et al., 2012) that high resilient individuals are less likely to develop burnout symptoms. Nevertheless, the mediating effect of work stress on resilience and burnout has not been investigated. Therefore, the present
70
S. Hao et al. / Personality and Individual Differences 83 (2015) 65–71
study represents an important advancement of our understanding of the role of resilience in the context of work stress and burnout. One possible explanation for this phenomenon is individuals can actively find and solve problems, actively adapt to the complex natural and social environment. As for the influence of work stress on burnout, individuals will not always passively wait for the presence of stress, but have the initiative to find and solve the problems that might bring pressure, and eliminate stress in advance. They can also take proactive measures to alleviate the adverse effects of work stress on burnout. High resilient individuals will be more confident and more able to positively react to stress (high controllability), can take a more optimistic attitude to face stress (high optimism), are more adept at using the surrounding supporting resources (high supportiveness), and also can calmly accept the consequences while confronting with huge pressure and setbacks (high acceptance). Therefore, high resilient individuals are capable of taking some relatively active measures to reduce work stress and to decrease the possibility of developing burnout symptoms. On the contrary, low resilient individuals, often taking passive way to deal with stress, may present more burnout symptoms such as emotional exhaustion, cynicism and reduced professional efficacy. Moreover, the moderating effect of resilience in the work stress–burnout relationship was also confirmed in the present study. The results indicated that higher resilience were significantly related with lower stress and lower burnout. For civil servants with higher resilience, the positive relationship between work stress and burnout is attenuated. However, for those with lower resilience, their relationship is just the reverse. That is, the relationship between the predictor and the outcome is weaker at higher levels of the moderator, and vice versa. Therefore resilience may serve as a buffer to prevent symptoms associated with burnout in a stressful government work environment. This finding is consistent with the protective factors model (Garmezy, Masten, & Tellegen, 1984). All these findings in the present study contributed to the understanding the relationship between resilience, stress and burnout, and verified the fact that resilience is a kind of positive personality trait for combating burnout among civil servants of Beijing. Additionally, these findings supported the transactional model of stress and coping theory. Various limitations of this study should be noted. Firstly, the cross-sectional design limited its ability to confirm a causal relationship between resilience, stress and burnout. Secondly, the present study relied solely on self-report measures and it was easy to produce common method variance (CMV), although we have already used procedural remedies to control the influence of CMV. Thirdly, because all the participants came from below 6 of 12 levels of civil servants in Beijing, the findings of this study may be more appropriately generalized to the mid-below levels of civil servants of Beijing, and it is not generalizable to the wider Chinese civil servants population. Notwithstanding this study’s limitations, several key implications can be drawn to better understand the relationship between resilience, stress and burnout. The results not only support the mediating and moderating effect of resilience on work stress and burnout, but also suggest work stress as a mediator between resilience and burnout. This study provides a new train of thought for the intervention of burnout among civil servants. By improving resilience of civil servants of Beijing, they can enhance the resistance to work stress, alleviate the burnout symptoms, help them take the initiative to find and solve the problems that may bring pressure, and nip the crisis in the bud, in advance, to eliminate possible hidden dangers, prevent the presence of burnout symptoms, and consequently improve the government’s work efficiency.
Acknowledgments This study was funded by the Key Research Base’s Major Projects of Ministry of Education of China (1JJD190001). We would like to thank all project staff and all the participants for this study, and thank Associate professor Yuling Qiao for her providing language help. Appendix A Hierarchical multiple regression analysis (N = 541). Model Variables 1
(Constant) Age Gender MS Education1 Education2 Rank1 Rank2 Rank3
B
SE 2.425 0.015 0.019 0.158 0.149 0.135 0.289 0.011 0.056
0.310 0.008 0.089 0.133 0.172 0.206 0.207 0.133 0.120
t
b 0.119 0.009 0.057 0.063 0.050 0.076 0.005 0.025
P 7.825 1.979 0.208 1.187 0.863 0.655 1.396 0.083 0.465
0.000 0.048 0.835 0.236 0.389 0.513 0.163 0.934 0.642
R = 0.159, R2 = 0.025, adjR2 = 0.010, DR2 = 0.025, F = 1.680, P > 0.05 2 (Constant) 1.141 0.286 3.995 0.000 Age 0.006 0.007 0.050 0.966 0.335 Gender 0.139 0.078 0.069 1.771 0.077 MS 0.055 0.115 0.020 0.477 0.634 Education1 0.078 0.149 0.033 0.522 0.602 Education2 0.037 0.179 0.014 0.208 0.836 Rank1 0.161 0.179 0.043 0.899 0.369 Rank2 0.058 0.115 0.026 0.502 0.616 Rank3 0.008 0.104 0.004 0.076 0.940 Work stress 0.011 0.001 0.474 9.861 0.000 Life and 0.002 0.002 0.050 1.038 0.300 health stress R = 0.521, R2 = 0.272, adjR2 = 0.258, DR2 = 0.247, F = 19.230, P < 0.001
Gender, marital status, education and administrative level were all converted into dummy variables. Level 6 and 7 = Rank 1, Level 8 = Rank 2, Level 9 = Rank 3, Level 10 and 11 = Rank 4. References Ahola, K., & Hakanen, J. (2007). Job strain, burnout, and depressive symptoms: A prospective study among dentists. Journal of Affective Disorders, 104, 103–110. Banovcinova, L., & Baskova, M. (2014). Sources of work-related stress and their effect on burnout in midwifery. Procedia, Social and Behavioral Sciences, 132, 248–254. Bonanno, G. A., Galea, S., Bucciarelli, A., & Vlahov, D. (2007). What predicts psychological resilience after disaster? The role of demographics, resources, and life and health stress. Journal of Consulting and Clinical Psychology, 75, 671–682. Boyas, J., & Wind, L. H. (2010). Employment-based social capital, job stress, and employee burnout: A public child welfare employee structural model. Children & Youth Services Review, 32, 380–388. Cooke, P. E. J., Doust, A. J., & Steele, C. M. (2013). A survey of resilience, burnout, and tolerance of uncertainty in Australian general practice registrars. BMC Medical Education, 132. García, G. M., & Calvo, J. C. A. (2012). Emotional exhaustion of nursing staff: Influence of emotional annoyance and resilience. International Nursing Review, 59, 101–107. Garmezy, N., Masten, A. S., & Tellegen, A. (1984). The study of stress and competence in children: A building block for developmental psychopathology. Child Development, 55, 97–111. Hao, S. W. (2014). Relationship between resilience, emotionality, stress and mental health among civil servants. (Doctor thesis), Beijing, China: Peking University. Hao, S. W., & Hong, W. (2014). Revision of resilient trait scale for Chinese adults among civil servants. Chinese Journal of Clinical Psychology, 22, 1032–1036.
S. Hao et al. / Personality and Individual Differences 83 (2015) 65–71 Hao, S. W., Xu, H. H., Zhou, L. L., Xie, Z. Y., & Hong, W. (2014). Development and evaluation of the civil servants stress scale. Chinese Journal of Public Health, 30, 1289–1292. Hsu, H. Y., Chen, S. H., Yu, H. Y., & Lou, J. H. (2010). Job stress, achievement motivation and occupational burnout among male nurses. Journal of Advanced Nursing, 66, 1592–1601. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. Hu, T., Zhang, D., & Wang, J. (2015). A meta-analysis of the trait resilience and mental health. Personality and Individual Differences, 76, 18–27. Lazarus, R. S., & Folkman, S. (1984). Stress appraisal and coping. New York: Springer. Li, C. P., & Shi, K. (2003). The influence of distributive justice and procedural justice on job burnout. Acta Psychologica Sinica, 35, 677–684. Maslach, C., & Jackson, S. (1981). The measurement of experienced burnout. Journal of Occupational Behavior, 2, 99–113. Maslach, C., Schaufeli, W. B., & Leiter, M. P. (2001). Job burnout. Annual Review of Psychology, 52, 397–422. Mealer, M., Jones, J., Newman, J., McFann, K. K., Rothbaum, B., & Moss, M. (2012). The presence of resilience is associated with a healthier psychological profile in intensive care unit (ICU) nurses: Results of a national survey. International Journal of Nursing Studies, 49, 292–299. Rutter, M. (2013). Resilience – clinical implications. Journal of Child Psychology and Psychiatry, 54, 474–487. Steiger, J. H. (1990). Structure model evaluation and modification: An interval estimation approach. Multivariate Behavioral Research, 25, 173–180. Taku, K. (2014). Relationships among perceived psychological growth, resilience and burnout in physicians. Personality and Individual Differences, 59, 120–123. Wang, P. J., & Zhang, S. C. (2011). Relationship between life events, resilience and learning burnout in junior school students. Science of Social Psychology, 26, 991–994.
71
Wen, Z. L., Chang, L., Hau, K. T., & Marsh, H. W. (2004). Testing and application of the mediating effects. Acta Psychologica Sinica, 36, 614–620. Wen, Z. L., Hau, K. T., & Chang, L. (2005). A comparison of moderator and mediator and their applications. Acta Psychologica Sinica, 37, 268–274. Wen, Z. L., & Wu, Y. (2010). Evolution and simplification of the approaches to estimating structural equation models with latent interaction. Advances in Psychological Science, 18, 1306–1313. Wen, Z. L., & Ye, B. J. (2014). Analyses of mediating effects: The development of methods and models. Advances in Psychological Science, 22, 731–745. Wu, M. L. (2013). Structural equation modeling: Operation and application of AMOS. Chongqing: Chongqing University Press. Wu, Y., Wen, Z. L., Hau, K. T., & Marsh, H. W. (2011). Appropriate standardized estimates of latent interaction models without the mean structure. Acta Psychologica Sinica, 43, 1219–1228. Wu, Y., Wen, Z. L., & Lin, G. C. (2009). Structural equation modeling of latent interactions without using the mean structure. Acta Psychologica Sinica, 41, 1252–1259. Xu, C. J., Zhang, S., Sun, J., & Tian, X. H. (2013). The relationship between role stress and job burnout of kindergarten teachers: Moderation of resilience. Early Childhood Education (Educational Sciences), 595, 26–30. Zhang, L. Y. (2013). College sports majors’ learning stress and learning burnout: Mediating effect of psychological resilience. Journal of Wuhan Institute of Physical Education, 47(10), 95–100. Zhu, Q., Fan, F., Zheng, Y. H., Sun, S. X., Zhang, L., & Tian, W. W. (2012). Moderating and mediating effects of resilience between negative life events and depression symptoms among adolescents following the 2008 Wenchuan earthquake in China. Chinese Journal of Clinical Psychology, 20, 514–517.
Contents lists available at ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Relationship between resilience, stress and burnout among civil servants in Beijing, China: Mediating and moderating effect analysis Shuwei Hao, Wei Hong ⇑, Honghong Xu, Lili Zhou, Zhongyao Xie Medical Psychology Center, Institute of Medical Humanities, Peking University, Beijing, PR China
a r t i c l e
i n f o
Article history: Received 23 September 2014 Received in revised form 17 March 2015 Accepted 27 March 2015 Available online 9 April 2015 Keywords: Resilience Stress Burnout Mediating effect Moderating effect Civil servants Structural equation modeling
a b s t r a c t This study aimed to explore the mediating and moderating effect in the relationship between resilience, stress and burnout among civil servants of Beijing, China. A cross-sectional study was conducted among civil servants in Beijing. Totally 541 civil servants completed a self-report questionnaire including three scales measuring civil servants’ resilience, stress and burnout. The data were analyzed with correlation, multiple regression and structural equation modeling. The results revealed that work stress rather than life and health stress could significantly predict burnout. Resilience played a partial mediating role between work stress and burnout, that is, work stress had both a direct and an indirect, via resilience, impact on burnout. Work stress played a partial mediating role between resilience and burnout, thus, resilience could prevent the development of burnout by relieving work stress, in addition to directly relieving it. Moreover, resilience was a moderator between work stress and burnout, and it could serve as a buffer to mitigate the adverse effects of work stress. These results suggest that resilience could be a positive personality trait for alleviating or eliminating work stress and combating burnout of civil servants of Beijing. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Civil servants who perform the function for the management of the state and government are a relatively special occupational group in China. With the development of the society and the deepening of the reform, the government’s management of civil servants has become more and more standard, more and more stringent. As a result, many civil servants feel ‘‘stressed out’’.
Abbreviations: RTSCA-R, Resilient Trait Scale for Chinese Adults-Revised; OP, optimism; AC, acceptance; CA, controllability; SU, supportiveness; CSSS, Civil Servants Stress Scale; MD, management and development; LR, life relationships; WR, working relationships; HR, health and responsibility; ES, economic stress; WL, working load; WS, work stress; LHS, life and health stress; ES1, two items related to work stress from economic stress dimension; MBI-GS, Maslach Burnout InventoryGeneral Survey; EE, emotional exhaustion; CY, cynicism; PE, professional efficacy; RWS, resilience and work stress interaction term; CA*MD, controllability and management and development product term; OP*WR, optimism and work relationships product term; AC*WL, acceptance and working load product term; SU*ES1, supportiveness and ES1 product term. ⇑ Corresponding author at: Medical Psychology Center, Institute of Medical Humanities, Peking University, 38, Xueyuan Rd., Haidian District, Beijing 100191, PR China. Tel.: +86 10 82801543; fax: +86 10 82802512. E-mail addresses: [email protected] (S. Hao), [email protected] (W. Hong), [email protected] (H. Xu), [email protected] (L. Zhou), [email protected] (Z. Xie). http://dx.doi.org/10.1016/j.paid.2015.03.048 0191-8869/Ó 2015 Elsevier Ltd. All rights reserved.
As is well known, stress is an important negative predictor of human health, and could result in some stress-related illnesses, such as burnout (Boyas & Wind, 2010; Hsu, Chen, Yu, & Lou, 2010). Burnout is an important phenomenon that derives from chronic emotional responses and interpersonal stressors that occur at work, and burnout syndrome is determined by the dimensions of emotional exhaustion, cynicism, and reduced professional efficacy (Maslach & Jackson, 1981; Maslach, Schaufeli, & Leiter, 2001). Although burnout has increased over the last decade, not everyone at the same workplace develops burnout, even facing the similar work stress. It suggested that individual factors may contribute to this phenomenon. In line with the transactional model of stress and coping theory, it indicated that some people are more vulnerable to stressful situations than others, especially those with high depressive symptoms, or have certain personality traits, such as high neuroticism or low optimism (Lazarus & Folkman, 1984). That is to say, contrary personality traits, such as low neuroticism or high optimism, are just the reverse. Consequently, not everyone perceives some life situations as stressful and the causes related to the individual variables may be the personality traits, for example, resilience as a personality trait may follow the same law. Resilience is an interactive phenomenon that is inferred from findings indicating that some individuals have a relatively good outcome despite having experienced serious stresses or adversities –
66
S. Hao et al. / Personality and Individual Differences 83 (2015) 65–71
their outcome being better than that of other individuals who suffered the same experiences (Rutter, 2013). Although there is still considerable debate regarding definitions of resilience, trait orientation suggests that resilience is a personality trait that helps individuals cope with adversity and achieve good adjustment and development (Hu, Zhang, & Wang, 2015). Consistent with the transactional model, resilience as a personality trait has been suggested to play an important role in the relationship between work stress and burnout. Several studies on nurses (García & Calvo, 2012; Mealer et al., 2012) and doctors (Taku, 2014) found resilience was a protective factor that can buffer the relationship between risk factors and burnout. Furthermore, the mediating effect of resilience on life events/learning stress and learning burnout of students (Wang & Zhang, 2011; Zhang, 2013) and the moderating effect of resilience on role stress and job burnout of teachers (Xu, Zhang, Sun, & Tian, 2013) have been proved. For civil servants, burnout threatens not only their own health, but also the government’s work efficiency and social stability. Therefore, it should be given more attention. Nonetheless, the mediating and moderating effect of resilience on stress and burnout among civil servants, especially, the mediating effect of stress on resilience and burnout has not been previously explored. According to the previous studies, we hypothesized that (1) the relationship between stress and burnout is mediated by resilience, and that (2) the relationship between resilience and burnout is mediated by stress, and that (3) resilience moderates the relationship between stress and burnout among civil servants in Beijing, China. Figure 1 depicts the hypothesized conceptual model of this study. 2. Methods 2.1. Participants and procedure A cross-sectional study with a representative sample consisting of 600 civil servants in Beijing, China, were carried out from April 2013 to June 2014. This study was approved by the Biomedical Ethics Committee of Peking University. We contacted the administrators of the departments affiliated to Beijing municipal government, and they assisted us in recruiting the civil servants who had agreed to participate. Participants were invited to complete a self-report questionnaire on resilience, stress, and burnout, and completed the questionnaire in a meeting room environment in their own departments. The questionnaire was anonymous, and the purpose of the research was explained to the respondents. Additionally, their confidentiality was assured, and they were informed that they had the right to drop out whenever they wanted. 541 civil servants completed the questionnaire, the other 59 civil servants were excluded due to missing values in
the relevant items in the questionnaire. The effective response rate was 90.17%. 2.2. Measures 2.2.1. Resilient Trait Scale for Chinese Adults-Revised (RTSCA-R) RTSCA-R (Hao, 2014; Hao & Hong, 2014) consists of 29 items and four dimensions: optimism (OP), acceptance (AC), controllability (CA) and supportiveness (SU). Items are assessed on a 4-point scale that ranges from ‘‘not true at all’’ (scored 1) to ‘‘true nearly all the time’’ (scored 4). Seven items are reverse-scored to reduce acquiescence biases. Higher scores on the scale reflect greater resilience. The reliability and validity of RTSCA-R were satisfied in Chinese civil servants. In the present study, Cronbach’s a coefficient for the total scale was 0.93. Cronbach’s a coefficients of OP, AC, CA and SU ranged from 0.75 to 0.90. 2.2.2. Civil Servants Stress Scale (CSSS) CSSS (Hao, 2014; Hao, Xu, Zhou, Xie, & Hong, 2014) which is designed to assess the perceived stress of Chinese civil servants consists of 38 items and six dimensions: management and development (MD), life relationships (LR), working relationships (WR), health and responsibility (HR), economic stress (ES) and working load (WL). Additionally, the CSSS can be divided into two subscales (Hao, 2014): work stress (WS) subscale and life and health stress (LHS) subscale. Work stress subscale consists of 23 items including MD, WR, WL and two items of ES, and life and health stress subscale consists of 15 items including LR, HR, and three items of ES. Each item is using an 11-point Likert scale ranging from 0 (no pressure) to 10 (maximum pressure). Higher scores on CSSS indicate higher levels of stress of civil servants. In this study, Cronbach’s a coefficient for the total scale was 0.95. Cronbach’s a coefficients of the six dimensions ranged from 0.80 to 0.94, and Cronbach’s a coefficients of work stress subscale and life and health stress subscale were 0.95 and 0.91, respectively. 2.2.3. Maslach Burnout Inventory-General Survey (MBI-GS) The Chinese version of MBI-GS (Li & Shi, 2003) consists of 15 items, using a seven-point Likert-type scale ranging from 0 (never) to 6 (every day). The 15-items scale has three subscales including emotional exhaustion (EE), cynicism (CY), and professional efficacy (PE), which reflect different aspects of the burnout syndromes. Higher scores on EE and CY subscales and lower scores on PE
Table 1 Frequency distribution of civil servants’ demographical characteristics (N = 541). Variables
Groups
n
%
Gender
Male Female Missing value 629 30–39 40–49 P50 Missing value Single Married Junior college or under Undergraduate Graduate Missing value Level 7 Level 8 Level 9 Level 10 Level 11 Level 12 Missing value
274 263 4 80 264 139 51 7 84 457 35 411 91 4 17 23 154 163 162 14 8
50.6 48.6 0.7 14.8 48.8 25.7 9.4 1.3 15.5 84.5 6.5 76.0 16.8 0.7 3.1 4.3 28.5 30.1 29.9 2.6 1.5
Age
Marital status Education
Administrative level
Fig. 1. Mediation and moderation model of resilience, stress and burnout.
67
1 0.35*** 0.35*** 0.33*** 0.12** 1 0.89*** 0.55*** 0.51*** 0.55*** 0.44*** 0.14**
13
1 0.65*** 0.77*** 0.67*** 0.48*** 0.51*** 0.46*** 0.13**
12
1 0.45*** 0.41*** 0.44*** 0.89*** 0.24*** 0.26*** 0.24*** 0.06
11
1 0.34*** 0.48*** 0.62*** 0.76*** 0.51*** 0.37*** 0.36*** 0.30*** 0.17***
10
1 0.50*** 0.39*** 0.38*** 0.43*** 0.47*** 0.71*** 0.27*** 0.21*** 0.23*** 0.16***
9
1 0.39*** 0.56*** 0.41*** 0.73*** 0.77*** 0.94*** 0.56*** 0.46*** 0.45*** 0.48*** 0.13**
8
1 0.87*** 0.63*** 0.73*** 0.69*** 0.81*** 0.83*** 0.94*** 0.86*** 0.51*** 0.50*** 0.48*** 0.16***
7
1 0.40*** 0.74*** 0.71*** 0.36*** 0.28*** 0.30*** 0.34*** 0.20*** 0.30*** 0.32*** 0.35*** 0.30*** 0.46*** 0.34*** 0.39*** 0.29***
1 0.32*** 0.29*** 0.27*** 0.16*** 0.33*** 0.28*** 0.14** 0.20*** 0.24*** 0.24*** 0.25*** 0.35*** 0.23*** 0.29*** 0.25***
1 0.71*** 0.30*** 0.22*** 0.20*** 0.26*** 0.21*** 0.26*** 0.25*** 0.27*** 0.27*** 0.40*** 0.30*** 0.31*** 0.27***
1
0.27*** 0.22*** 0.17*** 0.22*** 0.17*** 0.25*** 0.22*** 0.25*** 0.23*** 0.32*** 0.23*** 0.28*** 0.21***
6
P < 0.01. P < 0.001. **
***
5 3 2
3.01 ± 0.37 3.18 ± 0.48 3.01 ± 0.52 2.96 ± 0.42 2.98 ± 0.43 3.86 ± 1.79 4.29 ± 2.20 1.89 ± 2.17 3.59 ± 2.32 3.88 ± 2.76 4.60 ± 2.55 4.37 ± 1.96 4.22 ± 1.94 3.30 ± 2.04 2.09 ± 1.01 2.32 ± 1.49 1.81 ± 1.39 2.08 ± 1.31 1. Resilience 2. OP 3. AC 4. CA 5. SU 6. Stress 7. MD 8. LR 9. WR 10. HR 11. ES 12. WL 13. WS 14. LHS 15. Burnout 16. EE 17. CY 18. PE
1 M ± SD Variables
Table 2 Means, standard deviations and bivariate correlations (N = 541).
4
3.1. Demographic characteristics of participants
1 0.86*** 0.64*** 0.90*** 0.80*** 0.37*** 0.27*** 0.31*** 0.34*** 0.23*** 0.31*** 0.31*** 0.34*** 0.33*** 0.48*** 0.34*** 0.39*** 0.32***
3. Results
Table 2 presents the means, standard deviations, and Pearson correlation coefficients for the RTSCA-R, CSSS and MBI-GS. For civil servants, resilience, OP, AC, CA and SU were negatively correlated with stress, MD, LR, WR, HR, ES, WL, WS and LS (r = 0.37 to 0.12, P < 0.01). Resilience, OP, AC, CA and SU were negatively correlated with burnout, EE, CY and PE (r = 0.48 to 0.21, P < 0.001). Stress, MD, LR, WR, HR, ES, WL, WS and LS were positively correlated with burnout, EE and CY (r = 0.21 to 0.55, P < 0.001). Stress, MD, LR, WR, ES, WL, WS and LS were positively correlated with PE (r = 0.12 to 0.17, P < 0.01). However, there were no significant correlations between HR and PE (r = 0.06, P > 0.05).
1 0.62*** 0.53*** 0.53*** 0.50*** 0.17***
14
Descriptive statistics, correlation analysis, and Cronbach’s a reliability estimates were conducted using the IBM SPSS Statistics version 20 (SPSS20). The hypothesized models in this study were analyzed with structural equation modeling (SEM) procedures using the Analysis of Moment Structure version 20 (AMOS20). Continuous variables were presented as mean values with standard deviations, whereas, frequencies and percentages were given for the categorical variables. Correlation analyses were computed by way of Pearson’s correlations to determine the direction and size of the relationship between all variables in the structural model. Hierarchical multiple regression analysis was conducted to examine the relationship between control variables (gender, age, marital status, education and administrative level), work stress, life and health stress and burnout. Moreover, SEM was used to analyze the mediating and moderating models. Maximum likelihood estimation was employed as a global test of models because data for all continuous variables were normally distributed. The goodness-of-fit of the models was evaluated by the chi-square (v2) statistic, root mean square error of approximation (RMSEA), goodness-of-fit index (GFI), Tucker-Lewis fit index (TLI), comparative fit index (CFI), parsimony-adjusted normed fit index (PNFI) and parsimony goodnessof-fit index (PGFI). For the RMSEA, values less than 0.10 represent an acceptable fit (Steiger, 1990). For the GFI, TLI and CFI, values greater than 0.90 are considered acceptable (Hu & Bentler, 1999; Wu, 2013). And for the PNFI and PGFI, values greater than 0.50 are considered acceptable (Wu, 2013). It should be noted that all study continuous variables were centered before SEM analysis to increase interpretability of interactions and to control for multicollinearity (Wu, Wen, & Lin, 2009).
3.2. Descriptive statistics and correlations
1 0.78*** 0.81*** 0.62***
16 15
2.3. Statistical analysis
The study population was composed of 541 civil servants with a mean age of (37.08 ± 7.93) years (range 22–58 years). Administrative ranking of civil servants in China were divided into 12 levels, and all the respondents’ administrative ranking were below level 6 in this study. The demographic characteristics of the civil servants were shown in Table 1.
1 0.70*** 0.06
17
subscale indicate higher levels of burnout. Six items of PE dimension are using reverse scoring so that higher total scores on MBIGS could indicate higher levels of burnout. The Chinese version of the MBI-GS has high internal consistency (Li & Shi, 2003). In the present study, Cronbach’s a coefficient for the total scale was 0.89. Cronbach’s a coefficients of EE, CY, and PE were 0.95, 0.89, and 0.89, respectively.
1 0.20***
S. Hao et al. / Personality and Individual Differences 83 (2015) 65–71
68
S. Hao et al. / Personality and Individual Differences 83 (2015) 65–71
3.3. The relationship between work stress, life and health stress and burnout According to the definition of burnout (Maslach & Jackson, 1981; Maslach et al., 2001), it was caused by work stress. In order to investigate whether life and health stress could also have effect on burnout, we analyzed the relationship between control variables (gender, age, marital status, education and administrative level), work stress, life and health stress and burnout using hierarchical multiple regression analysis. The result (see Appendix Table) indicated that control variables had no significant impact on burnout (P > 0.05). Work stress had significant impact on burnout (F = 19.23, B = 0.11, b = 0.47, P < 0.001), nevertheless, life and health stress had no significant impact on burnout (b = 0.05, P > 0.05).
3.4. Mediation effect of resilience on work stress and burnout In line with the results of multiple regression analysis, the variable stress was replaced by work stress in the hypothesis model (Fig. 1). According to the method of mediating effect analysis (Wen, Chang, Hau, & Marsh, 2004; Wen, Hau, & Chang, 2005; Wen & Ye, 2014), we established the hypothesized mediating effect models of resilience and work stress (Figs. 2 and 3). The mediation model of resilience and the standardized coefficients for each variable are shown in Fig. 2. Structural equation model depicting significant regression and correlation paths in the model, all the path coefficients were statistically significant at the level of P < 0.01. The fit indices for the modified model were acceptable: v2 = 250.52, df = 41, RMSEA = 0.097, GFI = 0.92, TLI = 0.91, CFI = 0.93, PGFI = 0.57, PNFI = 0.68. According to the model, burnout was significantly predicted by work stress and resilience, the standardized direct effect value of work stress ? burnout was 0.57 (P < 0.001), and the standardized
Fig. 3. The mediation model of work stress on resilience and burnout (N = 541).
direct effect value of resilience ? burnout was 0.26, (P < 0.001). Resilience was significantly predicted by work stress, the standardized direct effect value of work stress ? resilience was 0.36 (P < 0.001). The standardized indirect effect value of work stress ? resilience ? burnout was ( 0.36 0.26 = 0.0936). The standardized total effect value of work stress on burnout was (0.57 + 0.0936 = 0.6636). Hence, the mediation effect ratio was (0.0936/0.6636 100% = 14.10%). 3.5. Mediation effect of work stress on resilience and burnout The mediation model of work stress and the coefficients for each variable are shown in Fig. 3. All the path coefficients were statistically significant at the level of P < 0.01. The fit indices for the modified model were acceptable: v2 = 250.52, df = 41, RMSEA = 0.097, GFI = 0.92, TLI = 0.91, CFI = 0.93, PGFI = 0.57, PNFI = 0.68. According to the modified model, burnout was significantly predicted by work stress and resilience (P < 0.001), the standardized direct effect value of resilience ? burnout was 0.26 (P < 0.001), and the standardized direct effect value of work stress ? burnout was 0.57 (P < 0.001). Work stress was significantly predicted by resilience, the standardized direct effect value of resilience ? work stress was 0.36 (P < 0.001). The standardized indirect effect value of resilience ? work stress ? burnout was ( 0.36 0.57 = 0.2052). The standardized total effect value of resilience on burnout was ( 0.26 0.2052 = 0.4652). Therefore, the mediation effect ratio was ( 0.2052/ 0.4652 100% = 44.11%). 3.6. Moderation effect of resilience on work stress and burnout
Fig. 2. The mediation model of resilience on work stress and burnout (N = 541). Resilience, RTSCA-R; OP, optimism; AC, acceptance; CA, controllability; SU, supportiveness; work Stress, work stress subscale; MD, management and development; WR, working relationships; ES1, two items related to work stress in economic stress dimension; WL, working load; Burnout, MBI-GS; EE, emotional exhaustion; CY, cynicism; PE, professional efficacy; e1–e13, error term. All measured indicators were centered, and all the path coefficients were standardized.
In accordance with the method of latent Interactions analysis (Wen & Wu, 2010; Wu, Wen, Hau, & Marsh, 2011; Wu et al., 2009), we established the moderation model of resilience (Fig. 4). The interaction term of resilience and work stress with four measured indicators in the model. Four dimensions of RTSCA-R multiplied four dimensions of work stress subscale, respectively, which yielded four product terms (CA*MD, OP*WR, AC*WL and SU*ES1) as measured indicators of resilience and work stress interaction term. The results of SEM indicated that all the path coefficients were statistically significant at the level of P < 0.01. The fit indices for
69
S. Hao et al. / Personality and Individual Differences 83 (2015) 65–71
Fig. 4. Moderation effect of resilience model (N = 541). RWS, resilience and work stress interaction term; CA*MD, controllability and management and development product term; OP*WR, optimism and work relationships product term; AC*WL, acceptance and working load product term; SU*ES1, supportiveness and ES1 product term; e1–e16, error term. All measured indicators were centered, and all the path coefficients were standardized.
4. Discussion The present study investigated the relationship between stress, resilience and burnout among civil servants of Beijing. Consistent with the previous studies, the results of multiple regression analysis revealed that work stress rather than life and health stress could significantly predict burnout, which verified the definition of job burnout. Although the relationship between work stress and burnout is well known and frequently studied, the specific mechanism by which burnout symptoms develop have yet to be fully established. This study expands on the previous findings by demonstrating that the mediating and moderating effect between resilience, work stress and resilience among civil servants of Beijing. On the one hand, the partial mediating effect of resilience on work stress and burnout indicated that work stress could directly and indirectly, via the impairment of resilience, exacerbate burnout symptoms. This phenomenon was confirmed in the previous studies on trauma and resilience (Bonanno, Galea, Bucciarelli, & Vlahov, 2007; Zhu et al., 2012). This result suggested that resilience
Low resilience
High resilience
2.5 2 1.5 Burnout
the modified model were acceptable: v2 = 358.15, df = 84, RMSEA = 0.08, GFI = 0.92, TLI = 0.90, CFI = 0.92, PGFI = 0.65, PNFI = 0.72. According to the moderation model, burnout was significantly predicted by work stress, resilience and work stress and resilience interaction term (RWS), the standardized direct effect value (or total effect value) of resilience ? burnout was 0.27 (P < 0.001), the standardized direct effect value of work stress ? burnout was 0.59 (P < 0.001), and the standardized direct effect value of RWS ? burnout was 0.12 (P < 0.01). In addition, Fig. 5 indicated that burnout symptoms of the low resilience group were more severe than those of the high resilience group. With the increase of work stress, the protective effect was more significant. That is, resilience was a moderator in the work stress–burnout relationship.
1 0.5 0 -0.5
Low work stress
High work stress
-1 -1.5
Work Stress
Fig. 5. Graphic representation of the interaction between work stress and resilience to influence burnout. Low resilience, 2 SDs below the centered mean; high resilience, 2 SDs above the centered mean.
as a personality trait is stable, but it is not unchangeable. The reason may be due to excessive pressure resulting in individuals’ ‘‘self-regulating mechanism’’ overwhelmed, and self-adjustment failed, which could produce mental maladjustment and a series of psychosomatic symptoms (Zhu et al., 2012), such as burnout. This finding is also in line with the prior research showing that high levels of perceived stress are highly predictive of the severity of burnout symptoms (Ahola & Hakanen, 2007; Banovcinova & Baskova, 2014). On the other hand, the partial mediating effect of work stress on resilience and burnout revealed that resilience could directly reduce the level of burnout, and via the reduction of work stress, indirectly reduce the level of burnout. The effect of resilience on burnout is consistent with the results of the previous studies (Cooke, Doust, & Steele, 2013; Mealer et al., 2012) that high resilient individuals are less likely to develop burnout symptoms. Nevertheless, the mediating effect of work stress on resilience and burnout has not been investigated. Therefore, the present
70
S. Hao et al. / Personality and Individual Differences 83 (2015) 65–71
study represents an important advancement of our understanding of the role of resilience in the context of work stress and burnout. One possible explanation for this phenomenon is individuals can actively find and solve problems, actively adapt to the complex natural and social environment. As for the influence of work stress on burnout, individuals will not always passively wait for the presence of stress, but have the initiative to find and solve the problems that might bring pressure, and eliminate stress in advance. They can also take proactive measures to alleviate the adverse effects of work stress on burnout. High resilient individuals will be more confident and more able to positively react to stress (high controllability), can take a more optimistic attitude to face stress (high optimism), are more adept at using the surrounding supporting resources (high supportiveness), and also can calmly accept the consequences while confronting with huge pressure and setbacks (high acceptance). Therefore, high resilient individuals are capable of taking some relatively active measures to reduce work stress and to decrease the possibility of developing burnout symptoms. On the contrary, low resilient individuals, often taking passive way to deal with stress, may present more burnout symptoms such as emotional exhaustion, cynicism and reduced professional efficacy. Moreover, the moderating effect of resilience in the work stress–burnout relationship was also confirmed in the present study. The results indicated that higher resilience were significantly related with lower stress and lower burnout. For civil servants with higher resilience, the positive relationship between work stress and burnout is attenuated. However, for those with lower resilience, their relationship is just the reverse. That is, the relationship between the predictor and the outcome is weaker at higher levels of the moderator, and vice versa. Therefore resilience may serve as a buffer to prevent symptoms associated with burnout in a stressful government work environment. This finding is consistent with the protective factors model (Garmezy, Masten, & Tellegen, 1984). All these findings in the present study contributed to the understanding the relationship between resilience, stress and burnout, and verified the fact that resilience is a kind of positive personality trait for combating burnout among civil servants of Beijing. Additionally, these findings supported the transactional model of stress and coping theory. Various limitations of this study should be noted. Firstly, the cross-sectional design limited its ability to confirm a causal relationship between resilience, stress and burnout. Secondly, the present study relied solely on self-report measures and it was easy to produce common method variance (CMV), although we have already used procedural remedies to control the influence of CMV. Thirdly, because all the participants came from below 6 of 12 levels of civil servants in Beijing, the findings of this study may be more appropriately generalized to the mid-below levels of civil servants of Beijing, and it is not generalizable to the wider Chinese civil servants population. Notwithstanding this study’s limitations, several key implications can be drawn to better understand the relationship between resilience, stress and burnout. The results not only support the mediating and moderating effect of resilience on work stress and burnout, but also suggest work stress as a mediator between resilience and burnout. This study provides a new train of thought for the intervention of burnout among civil servants. By improving resilience of civil servants of Beijing, they can enhance the resistance to work stress, alleviate the burnout symptoms, help them take the initiative to find and solve the problems that may bring pressure, and nip the crisis in the bud, in advance, to eliminate possible hidden dangers, prevent the presence of burnout symptoms, and consequently improve the government’s work efficiency.
Acknowledgments This study was funded by the Key Research Base’s Major Projects of Ministry of Education of China (1JJD190001). We would like to thank all project staff and all the participants for this study, and thank Associate professor Yuling Qiao for her providing language help. Appendix A Hierarchical multiple regression analysis (N = 541). Model Variables 1
(Constant) Age Gender MS Education1 Education2 Rank1 Rank2 Rank3
B
SE 2.425 0.015 0.019 0.158 0.149 0.135 0.289 0.011 0.056
0.310 0.008 0.089 0.133 0.172 0.206 0.207 0.133 0.120
t
b 0.119 0.009 0.057 0.063 0.050 0.076 0.005 0.025
P 7.825 1.979 0.208 1.187 0.863 0.655 1.396 0.083 0.465
0.000 0.048 0.835 0.236 0.389 0.513 0.163 0.934 0.642
R = 0.159, R2 = 0.025, adjR2 = 0.010, DR2 = 0.025, F = 1.680, P > 0.05 2 (Constant) 1.141 0.286 3.995 0.000 Age 0.006 0.007 0.050 0.966 0.335 Gender 0.139 0.078 0.069 1.771 0.077 MS 0.055 0.115 0.020 0.477 0.634 Education1 0.078 0.149 0.033 0.522 0.602 Education2 0.037 0.179 0.014 0.208 0.836 Rank1 0.161 0.179 0.043 0.899 0.369 Rank2 0.058 0.115 0.026 0.502 0.616 Rank3 0.008 0.104 0.004 0.076 0.940 Work stress 0.011 0.001 0.474 9.861 0.000 Life and 0.002 0.002 0.050 1.038 0.300 health stress R = 0.521, R2 = 0.272, adjR2 = 0.258, DR2 = 0.247, F = 19.230, P < 0.001
Gender, marital status, education and administrative level were all converted into dummy variables. Level 6 and 7 = Rank 1, Level 8 = Rank 2, Level 9 = Rank 3, Level 10 and 11 = Rank 4. References Ahola, K., & Hakanen, J. (2007). Job strain, burnout, and depressive symptoms: A prospective study among dentists. Journal of Affective Disorders, 104, 103–110. Banovcinova, L., & Baskova, M. (2014). Sources of work-related stress and their effect on burnout in midwifery. Procedia, Social and Behavioral Sciences, 132, 248–254. Bonanno, G. A., Galea, S., Bucciarelli, A., & Vlahov, D. (2007). What predicts psychological resilience after disaster? The role of demographics, resources, and life and health stress. Journal of Consulting and Clinical Psychology, 75, 671–682. Boyas, J., & Wind, L. H. (2010). Employment-based social capital, job stress, and employee burnout: A public child welfare employee structural model. Children & Youth Services Review, 32, 380–388. Cooke, P. E. J., Doust, A. J., & Steele, C. M. (2013). A survey of resilience, burnout, and tolerance of uncertainty in Australian general practice registrars. BMC Medical Education, 132. García, G. M., & Calvo, J. C. A. (2012). Emotional exhaustion of nursing staff: Influence of emotional annoyance and resilience. International Nursing Review, 59, 101–107. Garmezy, N., Masten, A. S., & Tellegen, A. (1984). The study of stress and competence in children: A building block for developmental psychopathology. Child Development, 55, 97–111. Hao, S. W. (2014). Relationship between resilience, emotionality, stress and mental health among civil servants. (Doctor thesis), Beijing, China: Peking University. Hao, S. W., & Hong, W. (2014). Revision of resilient trait scale for Chinese adults among civil servants. Chinese Journal of Clinical Psychology, 22, 1032–1036.
S. Hao et al. / Personality and Individual Differences 83 (2015) 65–71 Hao, S. W., Xu, H. H., Zhou, L. L., Xie, Z. Y., & Hong, W. (2014). Development and evaluation of the civil servants stress scale. Chinese Journal of Public Health, 30, 1289–1292. Hsu, H. Y., Chen, S. H., Yu, H. Y., & Lou, J. H. (2010). Job stress, achievement motivation and occupational burnout among male nurses. Journal of Advanced Nursing, 66, 1592–1601. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. Hu, T., Zhang, D., & Wang, J. (2015). A meta-analysis of the trait resilience and mental health. Personality and Individual Differences, 76, 18–27. Lazarus, R. S., & Folkman, S. (1984). Stress appraisal and coping. New York: Springer. Li, C. P., & Shi, K. (2003). The influence of distributive justice and procedural justice on job burnout. Acta Psychologica Sinica, 35, 677–684. Maslach, C., & Jackson, S. (1981). The measurement of experienced burnout. Journal of Occupational Behavior, 2, 99–113. Maslach, C., Schaufeli, W. B., & Leiter, M. P. (2001). Job burnout. Annual Review of Psychology, 52, 397–422. Mealer, M., Jones, J., Newman, J., McFann, K. K., Rothbaum, B., & Moss, M. (2012). The presence of resilience is associated with a healthier psychological profile in intensive care unit (ICU) nurses: Results of a national survey. International Journal of Nursing Studies, 49, 292–299. Rutter, M. (2013). Resilience – clinical implications. Journal of Child Psychology and Psychiatry, 54, 474–487. Steiger, J. H. (1990). Structure model evaluation and modification: An interval estimation approach. Multivariate Behavioral Research, 25, 173–180. Taku, K. (2014). Relationships among perceived psychological growth, resilience and burnout in physicians. Personality and Individual Differences, 59, 120–123. Wang, P. J., & Zhang, S. C. (2011). Relationship between life events, resilience and learning burnout in junior school students. Science of Social Psychology, 26, 991–994.
71
Wen, Z. L., Chang, L., Hau, K. T., & Marsh, H. W. (2004). Testing and application of the mediating effects. Acta Psychologica Sinica, 36, 614–620. Wen, Z. L., Hau, K. T., & Chang, L. (2005). A comparison of moderator and mediator and their applications. Acta Psychologica Sinica, 37, 268–274. Wen, Z. L., & Wu, Y. (2010). Evolution and simplification of the approaches to estimating structural equation models with latent interaction. Advances in Psychological Science, 18, 1306–1313. Wen, Z. L., & Ye, B. J. (2014). Analyses of mediating effects: The development of methods and models. Advances in Psychological Science, 22, 731–745. Wu, M. L. (2013). Structural equation modeling: Operation and application of AMOS. Chongqing: Chongqing University Press. Wu, Y., Wen, Z. L., Hau, K. T., & Marsh, H. W. (2011). Appropriate standardized estimates of latent interaction models without the mean structure. Acta Psychologica Sinica, 43, 1219–1228. Wu, Y., Wen, Z. L., & Lin, G. C. (2009). Structural equation modeling of latent interactions without using the mean structure. Acta Psychologica Sinica, 41, 1252–1259. Xu, C. J., Zhang, S., Sun, J., & Tian, X. H. (2013). The relationship between role stress and job burnout of kindergarten teachers: Moderation of resilience. Early Childhood Education (Educational Sciences), 595, 26–30. Zhang, L. Y. (2013). College sports majors’ learning stress and learning burnout: Mediating effect of psychological resilience. Journal of Wuhan Institute of Physical Education, 47(10), 95–100. Zhu, Q., Fan, F., Zheng, Y. H., Sun, S. X., Zhang, L., & Tian, W. W. (2012). Moderating and mediating effects of resilience between negative life events and depression symptoms among adolescents following the 2008 Wenchuan earthquake in China. Chinese Journal of Clinical Psychology, 20, 514–517.