Study design
A cross-sectional survey was conducted to assess the QOL of Chinese civil servants and explore the relationship between neuroticism, occupational stress, job satisfaction, and QOL.
Research participants
The participants were civil servants undergoing professional training at the Shandong Academy of Governance from July 2017 to November 2018. Shandong Academy of Governance is managed by the People’s Government of Shandong Province, China, and provides free professional training to the civil servants of Shandong Province. In total, 559 valid responses were received of the 590 questionnaires distributed, representing a response rate of 94.7%.
Measurements
QOL
The Short Form-8 (SF-8) Health Survey was conducted to measure QOL, which was derived from the Short Form-36 Health Survey. SF-8 can be completed in one to 2 mins, and can yield scores comparable to the Short Form-36 [35]. The Chinese version of SF-8 was translated and tested by Wang et al., and demonstrated good internal consistency reliability and criterion validity [35, 36]. Therefore, SF-8 has been applied to Chinese occupational groups [37]. SF-8 includes eight items that separately measure eight sub-scales: general health perceptions (GH), physical functioning (PF), role limitations due to physical health problems (RP), bodily pain (BP), vitality (VT), social functioning (SF), mental health (MH), and role limitations due to emotional problems (RE). Subjects responded to each item on a five-point scale. The sub-scale scores can be transformed into standard scores ranging from 0 to 100, with higher scores indicating better health. A standard score was computed using the following formula: standard scores = (actual raw score − lowest possible raw score possible) × 100 / raw score range [1]. Two summary scores were calculated using the weighted sum of the sub-scale scores: the physical component summary (PCS, includes GH, PF, RP, and BP) and mental component summary (MCS, includes VT, SF, MH, and RE). In this study, SEM was conducted to assess construct validity, and the results confirmed that the factor loadings of the eight indicator variables were no less than 0.62 (see Fig. 2), indicating acceptable construct validity. Cronbach’s α for the SF-8, PCS, and MCS was 0.906, 0.818, and 0.883, respectively, indicating satisfactory internal consistency.
Neuroticism
The neuroticism personality trait was measured using the Chinese version of the neuroticism subscale from the 44 items of the Big Five Inventory [38]. The questionnaire comprises eight items measured on a five-point scale (1 = strongly disagree; 5 = strongly agree), and the final score is the sum of these items. The construct validity of this questionnaire was confirmed through SEM, and the factor loadings of the eight indicator variables were above 0.42 (see Fig. 2). Cronbach’s α for the questionnaire was 0.819 in this study.
Occupational stress
This study combined the OSI-R, JDC, and ERI models to measure professional stress. Therefore, the measurement of occupational stress in the current study was comprised of three parts: occupational roles, JDC, and ERI.
According to the OSI-R model, occupational stressors originating in the work environment influence the perception of work roles [14]. Hence, the first part of occupational stress was assessed using the Occupational Role Questionnaire (ORQ), a sub-scale of OSI-R [14]. The ORQ used in this study was the revised version, which demonstrated good internal consistency reliability and criterion validity for a group of Chinese judges [29]. The revised ORQ includes 22 items and 4 sub-scales: role overload (6 items, an increasing and unreasonable workload), role boundary (5 items, feeling caught between conflicting supervisory demands and factions), responsibility (6 items, responsibility for activities and work performance), and physical environment (6 items, work schedule, working conditions, or feeling personally isolated). As differences in occupation could influence reliability and validity, a confirmatory factor analysis (CFA) was performed to assess the factor structure using AMOS 22.0. One item of the responsibility sub-scale was deleted based on the factor loading of indicator variables (the factor loading was 0.29), and good construct validity (the factor loadings were not less than 0.5) was thus obtained. Participants responded to each item on a five-point scale ranging from “1 (never)” to “5 (always).” A higher score indicated more severe conditions. Cronbach’s α for the questionnaire was 0.868.
The second part of occupational stress was measured according to the JDC model. The model emphasizes that occupational stress occurs when job demand exceeds job control [17]. Two questions were formulated to assess JDC. The first question was: “Is my job demanding of me (for example, intellectual and physical demands)?” The second question was: “Am I in control of my work (for example, I can control the working time, place, progress, objectives, etc.)?” The response range was “1 (very low)” to “5 (very high),” and the second question was reverse scored. Based on the JDC model, occupational stress levels can only be reflected by considering both job requirements and control. Therefore, the JDC score was created using the sum of the two questions with higher scores indicating higher occupational stress.
The third part of occupational stress was measured according to the ERI model where the imbalance between effort and rewards could result in stress [18]. To assess ERI, participants were asked: “Do I pay too much for my work (both physically and mentally)?” and “How much has my work rewarded me (both financially and spiritually)?” The response range was “1 (very low)” to “5 (very high).” The second question was reverse scored. The ERI score was the sum of the two questions, and higher scores indicated a more imbalanced job effort-reward and occupational stress.
SEM was conducted to assess the construct validity of occupational stress using the three scales with ORQ, JDC, and ERI as the first order factors and occupational stress as the second order factor. The results confirmed that the factor loadings for the three indicator variables were no less than 0.43 (see Fig. 2), indicating that the three indicator variables could effectively reflect occupational stress. The final score for occupational stress was the sum of all items of the three parts.
Job satisfaction
We developed three items to measure job satisfaction: (1) Overall, I am very satisfied with my job; (2) I regret doing the job; and (3) I would take the same job if given the chance to choose again. The response range was “1 (strongly disagree)” to “5 (strongly agree),” and item 2 was reverse coded. The final score was the sum, and higher scores indicated greater job satisfaction. A CFA was performed to assess the factor structure, demonstrating good construct validity (the factor loadings were not less than 0.6). Cronbach’s α was 0.713.
Data analysis
The data were analyzed using SPSS (v. 19.0) and AMOS (v. 22.0) software (IBM Corporation, Armonk, New York, USA). Descriptive statistics were performed to describe the socio-demographic factors, neuroticism, occupational stress, job satisfaction, and QOL of Chinese civil servants. T-tests and a one-way ANOVA were conducted to examine the differences in PCS, MCS, and total QOL scores across socio-demographic factors. A Pearson’s correlation coefficient was used to examine the relationships between neuroticism, occupational stress, job satisfaction, and QOL.
SEM is a method for specifying and testing models of linear relationships between observed variables (variables that can be directly measured) and latent variables (variables that cannot be directly measured and represented by multiple observed variables) [39]. SEM can simultaneously test the factor structure of latent variables and the complex relationships among multiple variables, such as direct and indirect relationships. Therefore, SEM was conducted to examine the mediating effect of occupational stress and job satisfaction on neuroticism and QOL in this study. Indirect effects were estimated by bias-corrected bootstrapping (2000 replications). The indirect effect is statistically significant at the 0.05 level if the bias-corrected bootstrap 95% confidence interval (CI) does not include zero [40].
The following indexes were used in the goodness-of-fit tests for the model: the normed Chi-square (χ2/df < 3), root mean square error of approximation (RMSEA < 0.08), goodness-of-fit index (GFI > 0.90), Tucker-Lewis fit index (TLI > 0.90), and comparative fit index (CFI > 0.90) [41]. Poor fitting means that the model is not suitable for the data and needs to be modified. Based on the modification indices suggested by the AMOS, correlating error terms is a method to improve fitting when supported by a strong theoretical justification [41].