Modelling Occupational Stress and Employee Health and Wellbeing in a Chinese Higher Education Institution

Abstract

Extensive change is evident in higher education in the People's Republic of China but there have been few studies of the effect of work stress on wellbeing in the higher education sector. The main aim of this study is to test and refine the ASSET (‘An Organizational Stress Screening Tool’) model of occupational stress in a sample of 150 academic and non-academic employees in a Chinese higher education institute. Using partial least squares modelling, the findings showed that job stressors predicted job dissatisfaction but, surprisingly, did not predict perceived commitment. Employees who reported job dissatisfaction tended to perceive that their organisation was less committed towards them and report less commitment towards the organisation. Job stressors resulted in poorer psychological wellbeing. Greater psychological wellbeing was associated with greater physical wellbeing. The findings suggest that university management should introduce strategies aimed at minimising job stressors as these would result in higher level of job satisfaction, higher level of commitment and ultimately resulted in an improvement in physical health.

Internationally, the nature of work is evolving at an accelerated pace. Globalisation and increased demands for greater efficiency, improved service quality, adaptability to continuously changing work environments and uncertainty of employment are all potential sources of occupational stress. Empirical evidence indicates that the experience of occupational stress leads to changes in physiological, psychological and behavioural functions, which may be detrimental to individual health, organisational and national productivity (Lundberg and Cooper, 2011).

The wind of change is also evident within the higher education sector in the People's Republic of China. Enormous social changes have taken place as the state-planned economy has been reformed into a market economy over the past two decades. As with their Western counterparts, Chinese higher education institutions have undergone significant changes, including reduced government funding, increasing enrolments, extensive expansion and employment reforms. Major educational reforms began in 1985 and resulted in a two-tier university system (Huang, 2005; Lai, 2010). Some of the consequences of the educational reform include a new system of personnel practices based on performance and contracts, with an emphasis on competition and rewards for university academics. Lai (2010) noted the continuing influence of the state and university administrations as a negative impact upon academic work in Chinese higher education institutions. The Chinese Ministry of Education reported that reforms of the higher education sector can be divided into five parts: reforms in the provision of education (such as the evaluation of undergraduate education according to the Criteria for National Undergraduate Education, reported in Jing, (2008)), assessment, management, investment, recruitment and job-placement. Management reform is considered the most important and difficult to implement (Ministry of Education of the People's Republic of China, n.d.).

Despite its importance to the wellbeing and productivity of workers, there has been limited systematic research on occupational stress in Chinese higher education institutions (with the exception of Jing (2008) and Lai (2010)). Recent word searches (25 July 2011) on the Web of Knowledge using the key words ‘China’, ‘work stress’ or ‘job stress’ or ‘occupational stress’, ‘job performance’ and ‘higher education’ yielded 20 articles but none examined occupational stress in Chinese higher education institutions.

To improve understanding of work stress mechanisms and their impact on academic and non-academic Chinese staff, research was conducted at a large, research-intensive university in Mainland China. The aim was to evaluate and extend the ASSET (‘An Organizational Stress Screening Tool’) occupational stress model developed by Cartwright and Cooper (2002) that was also used in a national research project in British universities (Tytherleigh et al., 2007). The study sought to evaluate the efficacy of the ASSET model for measuring the stress levels amongst academic and non-academic staff and to explore the relationships between job stressors and individual health, job dissatisfaction and commitment. Path analysis was conducted using partial least squares modelling, a form of structural equation modelling.

Stress in the higher education sector

The academic profession has long been highly respected and higher education institutions have been viewed as secure workplaces in both the East and the West. However, following extensive reforms of the higher educational systems in many countries, work stress in higher education institutions has recently attracted increasing attention from researchers (Gillespie et al., 2001; Winefield et al., 2003; Tytherleigh et al., 2005). A three-year study in 14 British universities revealed that staff stress was significantly higher than in normative data (Tytherleigh et al., 2005). Stress in these universities was significantly correlated with job insecurity, poor work relationships, lack of control and insufficient resources and communication. Another large-scale study of 17 universities in Australia indicated that high levels of stress were associated with insufficient funding, lack of resources, work overload, poor management practices, job insecurity and poor recognition and rewards (Winefield et al., 2003).

After evaluating undergraduate education by the Ministry of Education, Jing (2008) concluded that Chinese academic staff from five universities reported moderate stress from their jobs and organisational practices and that work stress had a significant negative effect on teaching effectiveness but not research productivity. Sun et al. (2011) also noted that Chinese academics have experienced greater occupational stress because increases in enrolments have not been matched by proportional increases in resourcing. Sun and colleagues also noted that the increasing focus by Chinese higher education institutions on research outputs in addition to teaching effectiveness also increased stress. Lai's (2010) research concluded that the on-going push for Chinese academics to obtain nationally reputable research grants leads to stress from role conflict. These studies shed some light on stress levels in the Chinese higher education sector but they did not provide empirical evidence regarding the causal relationships between stress and the physical and psychological wellbeing of Chinese higher education employees.

The ASSET occupational stress model

Growing interest in the negative effects of work has led to extensive research into occupational stress since the 1970s (Hart and Cooper, 2001). Explanations for stress have been divided into four categories: stimulus-based, response-based, interactive and transactional explanations. The stimulus-based approach views stress as a demand, or quality of the external context, that is detrimental to individual physical and psychological health. The response-based approach views stress as a physical and psychological reaction to environmental forces or hazards. The later, more integrative, interactional approach views stress as a cause and effect structural relationship. It interprets stress as both a stimulus (source of stress, or ‘stressor’) and a response (outcome of stress, or ‘strain’) (Cooper et al., 2001). It is characteristic of the interactive theoretical framework in psychology, which postulates that behaviour, attitudes and wellbeing are influenced jointly by person and environment (Cooper et al., 2001).

Transactional definitions are more concerned with the dynamics of cognitive appraisal and coping mechanisms that underpin a stressful experience (Lazarus and Folkman, 1984). Within the transactional framework, stress varies within the person in different conditions over time. It is, therefore, difficult to efficiently measure stress using traditional methodologies. Furthermore, due to the emphasis on individual subjective appraisal, researchers prefer to adopt interactional theories in occupational stress research (Jones and Bright, 2001, p. 186).

Building on existing interactional models, Cartwright and Cooper (2002) developed the ASSET model to study work-related stress across all occupations and conducted a United Kingdom (UK) based survey of 25,000 staff in 26 occupations using the ASSET questionnaire (see Figure 1). The ASSET model proposes predictable relationships between stressors at work, health and job outcomes. Many studies have demonstrated the negative impact of job stressors on employee wellbeing and job commitment. Cooper et al. (2001) reported negative effects of job insecurity on both wellbeing and commitment.

Figure 1. The ASSET Model of Occupational Stress

Source: Adopted from Faragher et al. (2004, p. 193)

Donald et al. (2005) used the ASSET model to examine 16,001 employees across different occupations. They found that the strongest predictor of productivity was psychological wellbeing, followed by commitment. Jacobs et al. (2007) reported similar findings from an analysis of secondary data collected from 13 UK higher education institutions. Performance was influenced by physical health as well as psychological wellbeing and commitment.

By contrast with previous work stress models, the ASSET model identifies commitment, usually conceptualised to be an outcome of stress, as a source of stress (Cartwright and Cooper, 2002). Commitment in the ASSET model is conceptualised as two separate constructs: the employee's commitment to the organisation and the organisation's commitment to the employee. Commitment reflects the psychological bond that ties the employee to the organisation (Meyer and Allen, 1991; Meyer and Maltin, 2010). Meyer and Maltin (2010, p. 324) noted that strong affective commitment to one's organisation might have positive effects on physical and psychological health. However, they noted that most studies that suggest the direct relationship between commitment and wellbeing tend to be correlational in nature.

The ASSET model has been widely used in the Western context but there have been few empirical evaluations of the model in Asian countries. To assess its efficacy in explaining and predicting the causal relationships of work stress of Chinese higher education staff on physical health and psychological wellbeing, the following hypotheses based on the ASSET theoretical framework are proposed (see Figure 2 for the proposed path model). The first five hypotheses examine the relationship of job stressors and perceptions of satisfaction and commitment:

Figure 2. Proposed Model of Occupational Stress in a Chinese Public University

Key:

OC2E = perceived commitment from organisation to employees

EC2O = employees' commitment to organisation

Phy Health = physical health

Psy WB = psychological wellbeing

• Hypothesis 1. Job stressors are positively related to job dissatisfaction.

• Hypothesis 2. Job stressors are negatively related to employees' perceptions of their organisation's commitment to them.

• Hypothesis 3. Job stressors are negatively related to employees' commitment to their organisation.

• Hypothesis 4. Job stressors are negatively related to employees' self-reported psychological wellbeing.

• Hypothesis 5. Job stressors are negatively related to employees' self-reported physical health.

The next four hypotheses explore the correlates of dissatisfaction:

• Hypothesis 6. Job dissatisfaction is negatively related to employees' perceptions of their organisation's commitment to them.

• Hypothesis 7. Job dissatisfaction is negatively related to employees' commitment to their organisation.

• Hypothesis 8. Job dissatisfaction is negatively related to employees' self-reported psychological wellbeing.

• Hypothesis 9. Job dissatisfaction is negatively related to employees' self-reported physical health.

The next five hypotheses explore commitment:

• Hypothesis 10. Employees' perceptions of their organisation's commitment to them are positively related to employees' commitment to the organisation.

• Hypothesis 11. Employees' perceptions of their organisation's commitment to them are positively related to their psychological wellbeing.

• Hypothesis 12. Employees' perceptions of their organisation's commitment to them are positively related to their physical health.

• Hypothesis 13. Employees' commitment to the organisation is positively related to their psychological wellbeing.

• Hypothesis 14. Employees' commitment to the organisation is positively related to their physical health.

The final hypotheses addresses the relationship between psychological wellbeing and physical health:

• Hypothesis 15. Employees' psychological wellbeing is positively related to their physical health.

Method

The study was conducted in a large, comprehensive mainland public university that is directly controlled by the Ministry of Education. It was established in the early 1900s and has close to 40,000 students studying in 71 disciplines across 25 schools on three campuses. The university has been undergoing reform, consistent with the changes of Project 985, and has expanded rapidly.

To ensure a high response rate, the questionnaires were administered jointly by the researcher and the Counselling Service Unit of the university. The questionnaire clarified the purpose of the survey and assured anonymity and confidentiality. The subjects were asked to return the answered questionnaires to a mailbox designated for the stress audit in each department or school.

Sample characteristics

Stratified random sampling was used to ensure the sample selected was representative of all staff categories. Faculties and administrative departments were randomly selected according to their proportion in the university, comprising two out of 14 schools from the Faculty of Arts, one out of seven schools from Faculty of Social Science, four out of 28 schools from the Faculty of Science and one out of seven administrative departments. Participants from each school included academic and non-academic staff. Participants from the administrative unit were non-academic staff.

In total, 320 questionnaires were distributed. After discarding those questionnaires with missing data, the final valid sample was 150 (response rate of 46.9 per cent), including 102 academic staff and 48 non-academic staff, of whom 40 per cent (n = 60) were female and 60 per cent (n = 90) were male. Females were over-represented when compared with the overall gender ratio in the institution (which was three males for every female). The sample included 97 lecturers and professors (65 per cent of the total sample), five laboratory research fellows (3.3 per cent), 19 clerical and administrative stuff (12.7 per cent) and 29 IT support staff (20 per cent). The distribution of participants by job category is similar to their distribution within the university.

A majority of the sample had been working in the university for at least ten years (61.3 per cent) and most earned between 10,000 and 30,000 RMB per year. A large majority of the respondents (86 per cent) of the staff are married. In China, the legal working hours are eight hours per day and forty hours per week. However, the culture of long hours is becoming more and more prevalent in Chinese workplaces, including higher education institutions. Individuals who worked over 40 hours per week made up a high percentage of the sample (66.5 per cent) and most of them held senior positions (81 per cent). Approximately half of the mid- and junior-level staff worked over 40 hours per week.

Measurement and model estimation

The ASSET questionnaire (Cartwright and Cooper, 2002) used in this study consisted of four parts. The first three related to the ASSET model and together evaluated the participants' perceptions of sources of pressure and work stress. The final part collected demographical data. Only demographic items were revised to adapt to the Chinese educational environment. Approval was obtained from the original authors of the ASSET questionnaire to translate the instrument into Mandarin. Prior to data collection, the survey was pilot-tested with 15 members of the university staff who did not participate in the final survey. The survey was translated from English into Mandarin by a bilingual Chinese academic. It was then back translated into English, consistent with Brislin (1979). The ASSET questionnaire has proven to have good validity and reliability in studies conducted in the UK, Greece and South Africa, as summarised in Table 1 (see Johnson and Cooper, 2003; Faragher et al., 2004; Donald et al., 2005; Vakola and Nikolaou, 2005; Jackson and Rothmann, 2006).

Table 1. Comparison of internal consistency coefficients for ASSET model of occupational stress

Context	Current Study	Tytherleigh et al. (2007)	Tytherleigh et al. (2005)	Jackson and Rothmann (2006)	Johnson and Cooper (2003)	Faragher et al. (2004)	Donald et al. (2005)	Vakola and Nikolaou (2005)
	N = 150	N = 3,550	N = 3,808	N = 1,170	N = 130	N = 2,552 (study 1), 6644 (study 2)	N = 16,001	N = 292
	HEI in China	UK HEI	UK HEI	HEI in South Africa	UK local authority	Cross sectional; UK public and private sector organisations	Cross sectional, including occupations in higher education institutions	Cross sectional from Greece
Key: n.a., not available; HEI, higher education institution.
Work relationships	0.87	0.84	0.84	0.79	0.81	0.84	0.76	0.80
Overload	0.76	0.82	0.82	0.68	0.77	0.82	0.77	0.76
Control	0.70	0.84	0.61	0.72	0.70	0.81	0.67	0.68
Job security	0.61	0.63	0.84	0.57	0.71	0.60	0.54	0.60
Resources and communication	0.77	0.73	0.73	0.59	0.60	0.69	0.65	0.67
Work-life balance	0.60	0.72	0.64	0.69	0.76	0.75	0.70	0.57
Pay and benefits	—	—	—	—	—	—	—	—
Your job (renamed job dissatisfaction for this study)	0.66	0.63	0.74	0.61	0.65	0.66	0.57	0.49
OC2E	0.79	0.83	0.82	0.83	0.82	0.83	0.68	0.82
EC2O	0.75	0.78	0.75	0.65	0.74	0.77	0.79	0.75
Psychological wellbeing	0.91	0.90	0.82	0.89	0.91	0.93	0.90	n.a.
Physical health	0.83	0.78	0.75	0.79	0.74	0.78	0.75	n.a.

The proposed structural model (Figure 1) was tested by generating a partial least squares (PLS) solution using SmartPLS v.2 (Ringle et al., 2005). A series of ANOVAs was also conducted in PAWS 18 to test the effects of the path model for different control variables such as category of employee (academic versus non-academic), gender, marital status and levels of appointment (according to junior, middle and senior level).

In partial least squares, the path coefficients are standardised regression coefficients and the loadings are similar to component loadings. The significance of relationships between variables is determined using the bootstrap procedure (Ringle et al., 2005). Bootstrapping is carried out to provide extra confidence that the results are not sample-specific by using repeated random samples drawn from the data. In this instance, 500 re-samples were used.

When testing any specific model, PLS requires a smaller sample than linear structural relations (LISREL) and other covariance-based structural equation modelling techniques. Another advantage of using PLS over LISREL is that PLS does not require multivariate normal data. PLS is also considered to be more appropriate for exploratory model building. Given the number of variables in the proposed model, the present sample size is within the range considered to be suitable for PLS analysis (Chin and Newsted, 1999, p. 314). According to Green (1991, p. 503), the minimum sample size required for a path model with six independent variables is 97 cases.

PLS allows the operationalisation of constructs as formative scales, instead of the reflective scales used in LISREL and analysis of moment structures (AMOS). The only reflective construct in the model with multiple sub-scales is the ‘stressor’ construct, which has a composite reliability coefficient of 0.89. The validity and reliability of the construct can be checked by examining the average variance extracted (AVE). AVE for the construct was 0.55, which is greater than the threshold of 0.50. Furthermore, the square root of the AVE is larger than its correlation with any other construct, indicating discriminant validity.

The procedures outlined in Petter et al. (2007) were used to evaluate the validity and reliability of formative scales. Face and content validity of the formative constructs are derived from theory (in this instance, occupational stress). The literature indicates that the constructs are abstract and complex. Items with low loadings were retained to maintain the theoretical meaning of the constructs. Multicollinearity was evaluated using the variance inflation factors of the items composing each of the formative scales. There was no evidence of multicollinearity, as all variance inflation factors were less than 3.33 (Cenfetelli and Bassellier, 2009).

Several steps were undertaken to limit the impact of common method variance. First, only validated scales and objective items were used (Chang et al., 2010). The ASSET model has been previously validated in various national (UK, Greece and South Africa) and occupational (higher education, emergency service and health care) contexts. Second, Harman's ex post one-factor test was used to test whether the current study did not suffer from common method variance (Podsakoff and Organ, 1986). The analysis showed that 18 factors with eigenvalues greater than 1.0 explained 73 per cent of the variance. This finding suggests common method variance was not an issue. Finally, discriminant analyses of the items and their associated latent variables in the outer model and bootstrapping procedures were undertaken on the relationships between the latent variables in their inner models using SmartPLS.

Sample items from the scales used are reported in Appendix 1. The full ASSET questionnaire cannot be provided for copyright reasons (see Cartwright and Cooper, 2002). The construct validity and reliability of the instrument in the higher education context have been reported previously (in particular, see Faragher et al., 2004 for detailed reporting of the validity of the instrument). Table 1 reports the internal consistency coefficients from the current study and other studies using the ASSET instrument. With the exception of ‘your job’ (job dissatisfaction)’, ‘work-life balance’ and ‘job security’, which has Cronbach's alphas less than 0.70, the internal consistencies of the scales were consistent with evidence from the UK. An advantage of using ASSET model in the UK is that the normative database of the ASSET stress audit can aid organisations benchmark employee wellbeing and job outcome with respect of stress in their workplace (Donald et al., 2005; Johnson, 2009).

Job stressors (formative scale)

This is a second order latent variable, formed with seven reflective scales. The sub-scales are ‘work relation’ (eight items), ‘work-life balance’ (four items), ‘job overload’ (four items), ‘job insecurity’ (four items), ‘control’ (four items), ‘resources and communication’ (four items) and ‘pay and benefit’ (one item). Each is operationalised as a composite index (that is, all of the scales are summed to form an overall job stressors scale). The items were rated on a six-point Likert-type scale ranging from ‘1’ = strongly disagree to ‘6’ = strongly agree. Sample items are reported in Appendix 1. As reported in Table 1, most of the scales have good internal consistency.

Job dissatisfaction (reflective scale)

Cartwright and Cooper's (2002) ‘Aspects of your job’ scale was adapted to measure job dissatisfaction. This scale was viewed as an indicator of job dissatisfaction because of its very high negative correlation with established job-satisfaction measures (Johnson, 2009, 142). The construct was renamed ‘job dissatisfaction’. The scale was measured with a six-point Likert scale, ranging from ‘1’ = strongly disagree to ‘6’ = strongly agree. It is measured by an eight-item composite index. Higher scores represent greater job dissatisfaction.

Commitment (formative scale)

Following the ASSET model, commitment is measured with two second-order formative constructs. Perceived commitment of the organisation to the employee is measured with a five-item scale. Commitment of the employee to the organisation is measured with four items. The items were rated from ‘1’ = strongly disagree’ to ‘6’ = strongly agree.

Employee health and wellbeing (formative scales)

Two constructs were used to operationalise employee health and wellbeing. The first, ‘physical health’ (six items), gives an insight into physical problems related to stress, not detailed clinical data. The second construct, ‘psychological wellbeing’ (11 items), measures the symptoms of stress-induced psychological ill health. Responses were anchored on a 4-point Likert scale, ranging from ‘1’ = never to ‘4’ = often. These were reverse coded so that a higher score indicates better health.

Control variables

Many variables influence job stress in higher education institutions. They include functional role (academic versus administrative), gender, marital status and working hours. Winefield et al. (2003) found that academic staff reported less job satisfaction than administrative staff. Pick et al. (2012) noted that administrative staff experienced a variety of stressors that impact on the level of job satisfaction and psychological wellbeing in a sample of support staff in three public universities in Australia. Tytherleigh et al. (2005) used the ASSET model of occupational stress in a sample of UK higher education institutions where they reported that academic and research staff were the most stressed especially in relation to work–life balance and overload. Similarly Jacobs et al. (2007) found empirical support for variations in the relationship between job performance and stress between academic and non-academic staff and they conclude that job factors influence the findings. It has been argued that, for some occupational groups, high levels of commitment and job satisfaction might moderate the effects of stress (Langan-Fox and Cooper, 2011).

The effect of gender differences on stress effects have been examined (Tytherleigh et al., 2005; Langan-Fox and Cooper, 2011). However, there were inconsistent findings. Vakola and Nikolaou (2005) concluded that males were found to have higher-level commitment of the employee to the organisation in Greece. On the other hand, Tytherleigh et al. (2007) found that there were statistically differences between males and females in relation to the negative consequences of pay and benefits and health outcomes. Johnson and Cooper (2003) did not find any significant difference in psychological wellbeing according to gender. Similarly, Jackson and Rothmann (2006) did not find any difference in occupational stress, commitment and health outcomes according to gender difference.

Marital status has been used as a control variable in occupational stress research. Scholars (Swanson et al., 1998; August and Waltman, 2004; Tytherleigh et al., 2007) have argued that one of the challenges faced by female academics in the higher education sector is family duties, as their work commitment placed constraints on their family time. Other scholars in the work-family conflict area (Dewe et al., 2010) have also shown that dual career couples experienced work stressors when compared to those who are not married.

Various studies have shown that excessive working hours impact positively on job stress. Some studies have shown that that academic and research staff tend to work excessively long hours (Tytherleigh et al., 2005). However, Tytherleigh et al. did not find a relationship between long working hours and physical or psychological health outcomes.

Before testing the causality of job stressors on health and wellbeing, several tests were conducted to examine the effects of demographic differences, such as category of employee (academic versus non-academic),¹ gender, marital status and number of excess hours worked. Results of the ANOVA showed that there was a significant difference between academic staff and non-academic staff in the level of stressors, with academic staff receiving poorer scores, particularly for overload and work-life balance. With respect to the number of excess hours worked per week, there were significant differences between those who worked less than 11 hours of extra work per week, 11–20 and above 30 hours for overload and work-life balance stressors. In relation to employees' commitment to the organisation, the difference was between those who worked less than 11 extra hours and those who worked more than 21 hours. There were no statistically significant differences in the stress items according to gender and marital status.

A MANCOVA was subsequently performed for function by number of excess hours worked, as independent t-tests analyses showed that these two variables were found to have significant impact on the model. It revealed that overload and work-life balance were statistically significant, after controlling for staff function and number of excess hours worked. Consequently, an additional variable, the interaction of function and excess hours worked, was introduced into the path model. A path was introduced from this construct to the remaining constructs in the path model. The finding suggests that academic staff experienced higher levels of workload and less beneficial work-life balance than non-academic staff. Therefore, to increase their physical and psychological wellbeing, university management must implement appropriate human resource strategies to allow academic staff to work less, such that they have an improved level of work-life balance. Further implications of this finding will be discussed in the section on managerial implications.

Findings

Descriptive statistics and inter-correlations for the constructs used in the ASSET Model are reported in Table 2. The respondents' level of job dissatisfaction was below the mid-point level of the six-point Likert scale, suggesting that they were on average, satisfied with their job. There was no difference in the level of job satisfaction between male and female respondents. The respondents also reported that their perception of their organisation's commitment towards them and their level of commitment towards their organisation, were both above the mid-point of a six-point scale. On the other hand, their level of psychological wellbeing and physical health were both below the mid-point level of a four-point scale.

Table 2. Descriptive statistics and inter-correlations of second order constructs (N = 150)

	M	SD	1	2	3	4	5	6	7
OC2E, perceived commitment from organisation to employees; EC2O, employees' commitment to organisation. *P < 0.05, **P < 0.01, ***P < 0.001.
1. Function by excess hrs worked	1.32	1.07	1.00
2. Stressors (formative scale)	20.13	5.36	0.34***	1.00
3. Job dissatisfaction (reflective scale)	2.73	.72	0.10	0.73***	1.00
4. OC2E (formative scale)	21.58	4.08	0.03	−0.40***	−0.46***	1.00
5. EC2O (formative scale)	18.19	3.28	0.19*	−0.18*	−0.35***	0.70***	1.00
6. Psychological wellbeing (formative scale)	20.69	6.91	−0.21*	−0.47***	−0.47***	0.24**	0.27**	1.00
7. Physical health (formative scale)	11.59	4.10	−0.13	−0.40***	−0.34***	0.20*	0.23**	0.75***	1.00

The path model (Figure 3), explained 65.2 per cent of the variance in physical health and 32 per cent of psychological health. The results (Table 3) showed that the ASSET Model of Occupational Stress can be used to explain the level of wellbeing of Chinese higher education institution employees (as indicated by the global goodness of fit index recommended by Tenenhaus et al., 2005) of 0.514.

Figure 3. Results of Path Analysis

Note: Figures denote path coefficients

Table 3. Results of Path Analysis (N = 150)

Paths	Coefficients	t-statistic	Sig. level
Note: Function: ‘0’ = Non-academic, ‘1’ = Academic. * P < 0.05, ** P < 0.01; *** P < 0.001; n.s. not significant.
Function x Excess Hrs → Job Stressors	0.2902	3.3456	***
Function x Excess Hrs → Job Dissatisfaction	−0.0953	2.2715	*
Function x Excess Hrs → Perceived Commitment from Organisation to Employees	0.1773	2.2978	*
Function x Excess Hrs → Employees' Commitment towards the Organisation	0.1534	2.0639	*
Function x Excess Hrs → Psychological Wellbeing	−0.0326	0.4334	n.s.
Function x Excess Hrs → Physical Health	−0.0872	1.5635	n.s.
Job Stressors → Job Dissatisfaction	0.8819	30.8024	***
Job Stressors → Perceived Commitment from Organisation to Employees	−0.3446	1.9956	*
Job Stressors → Employees' Commitment towards the Organisation	0.1605	1.4538	n.s.
Job Stressors → Psychological Wellbeing	−0.3597	2.6862	**
Job Stressors → Physical Health	−0.0793	0.7621	n.s.
Job Dissatisfaction → Perceived Commitment from Organisation to Employees	−0.1752	1.2083	n.s.
Job Dissatisfaction → Employees' Commitment towards the Organisation	−0.1839	1.4309	n.s.
Job Dissatisfaction → Psychological Wellbeing	−0.0188	0.1317	n.s.
Job Dissatisfaction → Physical Health	−0.1458	1.4413	n.s.
Perceived Commitment from Organisation to Employees → Employees' Commitment towards the Organisation	0.5267	6.0397	***
Perceived Commitment from Organisation to Employees → Psychological Wellbeing	−0.0549	0.5675	n.s.
Perceived Commitment from Organisation to Employees → Physical Health	−0.0458	0.5877	n.s.
Employees' Commitment towards the Organisation → Psychological Wellbeing	0.1467	1.5365	n.s.
Employees' Commitment towards the Organisation → Physical Health	0.1217	1.983	*
Psychological Wellbeing → Physical Health	0.6481	10.5198	***

Implications for higher education management

The primary aim of the current study was to apply the ASSET model of occupational stress developed by Cartwright and Cooper (2002) in a sample of academic and non-academic employees in a research intensive Chinese university. The operationalisation of the constructs within the ASSET model was extended by introducing two separate paths within the ‘commitment’ and ‘health and wellbeing’ constructs. The findings showed that the ASSET model is appropriate for understanding the effect of occupational stress on physical and psychological wellbeing within a sample of Chinese university employees.

Respondents in the current study worked in excess of 40 hours per week. The culture of long hours is becoming more and more prevalent in Chinese workplaces, including higher education institutions. This result is consistent with a previous study of the working hours of university teachers in China (Chen and Zhuang, 2005) and a study in UK higher education institutions (Tytherleigh et al., 2005). Due to the high ratio of students to academics in Chinese universities, teaching and administration loads have increased. The emphasis on quantity of publications in performance management also drives academics to work long hours in research (Lai, 2010; Sun et al., 2011).

A key finding was that academic staff reported higher level of job stressors and job dissatisfaction in the sample population. In particular, stressors relating to workload and work-life balance contribute to the significant difference in job stressors between academic and non-academic staff. Job stressors were found to negatively lead to commitment. Job dissatisfaction also contributed negatively to commitment, especially as academic staff reported more negative assessment of the level of commitment. This has a negative consequence as it impacts on the level of negative symptoms of physical health reported by the respondents. Job stressors were found to directly impact negatively on the level of psychological wellbeing, which resulted in a deterioration of physical health.

The current study made several contributions in relation to the conceptualisation of the ASSET model and empirical understanding of the extent of occupational stress in the Chinese higher education context. First, the current study adopts a causal modelling approach using partial least squares technique to operationalise and develop the path relationships between job stressors, commitment, health and wellbeing. While Faragher et al. (2004) adopted a structural-equation-modelling approach to examine the measurement and structural properties of the ASSET model, most studies have utilised a multiple-regression approach to examine the predictive relationships of the antecedents and consequences of occupational stress. The current study adds to the literature by developing a causal-modelling perspective to explain the relationships between variables.

The second contribution of the study to the literature is in the manner in which two new path relationships were created in the path model. Hypothesis 5 posits that employees' perception of their university's commitment leads to their own commitment to the university. Hypothesis 15 posits that employee's psychological wellbeing can positively lead to physical health. Both of these hypotheses were supported in the path analysis.

This result is generally consistent with the causal relationships identified in western studies on the relationships between stressors and wellbeing in higher education institutions (Dua, 1994; Winefield et al., 2003). In addition, the finding provides additional empirical evidence that there is a causational relationship from employee's commitment to the organisations to physical wellbeing; thus supporting the commonly held view that strong affective commitment to one's organisation might have positive effects on wellbeing (Meyer and Maltin, 2010).

As noted by Johnson and Cooper (2003, p. 194), there is good convergent validity for the psychological wellbeing scale, the GHQ12; a common scale used to measure psychological wellbeing. The study finding (Hypothesis 4) showed that occupational stressors have a negative consequence on the mental health and wellbeing of academic and non-academic staff in Chinese higher education institutions. There has been support in the literature for the findings (He et al., 2000; Siu, 2002; Johnson and Cooper, 2003). Also, the study found that individuals who reported a higher level of psychological wellbeing also tend to report a higher level of physical wellbeing. This finding corroborates the evidence in the literature, which support the relationship between physical and psychological wellbeing (Cooper et al., 2001).

Job stressors are a proven source of job dissatisfaction in the university resulting in low commitment. Job dissatisfaction as a major effect of strain was well documented in occupational stress research (Sullivan and Bhagat, 1992). A negative significant correlation between stress and job satisfaction was found in a study of British higher education institution staff (Abouserie, 1996). The stress effect role of job dissatisfaction in the ASSET model was validated by this study of Chinese higher education staff. Noticeably, the scale used to measure job dissatisfaction was reported to have low internal consistency (below 0.6) (Table 1). This is consistent in most of the studies conducted with the ASSET Model. Therefore, it is important that the construct is operationalised as a formative scale, rather than a reflective scale (which meant that items would have to be deleted into order to achieve good model fit in structural equation modelling), similar with all of the studies using the ASSET model.

Further examination of the means for job dissatisfaction, commitment and health and wellbeing variables revealed they were above the mid-point level. These results indicate that the participants had a high level of engagement, despite a high level of job stressors and job dissatisfaction. The influence of job dissatisfaction on commitment was not found. This finding is contrary to previous evidence (Cass et al., 2003) that job stressors have negative effect on employees' commitment to the organisation. It implies that job satisfaction does not affect commitment and health and wellbeing in the Chinese higher education sector. This finding may be explained by the existence of the ‘iron rice bowl’ syndrome of employment relationship in China that guarantees lifetime employment to public-sector employees. It could also be explained by the cultural values of the Chinese society (Hofstede, 1984) in which employees have a strong sense of commitment and loyalty to their organisation, despite increasingly stressful working conditions (for example, effective teaching and greater research outputs). In addition, during the 1990's, there was a trend whereby university staff in China who experienced low job satisfaction tended to leave the sector for a career in industry or as entrepreneurs. Due to the high level of autonomy of academic staff and low turnover in Chinese higher education institutions, there might be other factors that are important influences on commitment in the sample of higher education institution staff. Further research should be conducted to improve understanding of the relationship between job stressors, job satisfaction, commitment and health and wellbeing.

The findings of this study provide support for the causal relationship between job stressors and employee wellbeing and the relationship between job dissatisfaction and commitment. The findings can provide organisations a simple human relations framework of antecedents and outcomes of work stress, particularly when many organisational changes take place in the higher education sector. The findings from this stress audit in a Chinese university could be used to help construct a stress prevention strategy and action plan for resolving any problems identified in higher education institutions. The identification of the mediating role of job dissatisfaction between job stressors and organisational commitment provided some implication to occupational stress theory and human resources management. Job dissatisfaction and commitment are well studied in human resource management research yielding inconsistent results in their causal relationship (Currivan, 1999). In occupational stress research models, job dissatisfaction and organisational commitment are often conceptualised as the stress outcomes (Cartwright and Cooper, 2002). There is little examination of the relationship between the two constructs in occupational stress models. The identification of the mediating role of job satisfaction between job stressors and commitment might imply that in stress intervention, human resources managers in Chinese higher education institutions should consider job satisfaction as an important organisational effectiveness indicator like the human resources managers in the private sector in Western countries. A commitment-based human resource management system (Whitener, 2001) could be implemented to ensure that commitment of higher education institution employees could reduce their intention to leave as Chinese higher education becomes more research-performance focused, as those in the USA, UK and Australia.

Job satisfaction of staff in Chinese higher education institutions could be improved at two levels. At the job level, work redesign for teaching and administrative or support positions may reduce workload and increase job control. At the organisational level, job security, sound, equitable and fair pay and benefit, frequent communication from university management and provision of resources for academic and non-academic work could be introduced by the human resource management function. Further, accommodating work-life balance policies into human resource strategy could be another possible means of reducing occupational stress, improving job satisfaction and performance (Houston et al., 2006).

Conclusion

The findings of this study provide support for the applicability of the ASSET model of occupational stress in a Chinese higher education institution. It indicates that job stress has a negative impact on employee wellbeing and attitudes to work in the Chinese higher education sector. In addition, the current study provided the empirical evidence to indicate that there was a causal relationship from stressors to job dissatisfaction, which leads to a reduction in commitment and physical health. There was also empirical evidence in support for the causal effect of employees' perception of their organisation's commitment towards them on their level of commitment, which subsequently led to an improvement in physical health.

Despite the potential effect of common method bias, tests were conducted to ensure the effects are minimised. In addition to Harman's one factor test, the research include checks for multicollinearity in formative scales, discriminant analysis, AVE for reflective scale and goodness-of-fit index. Together with the support from other studies found in the literature, the ASSET model and our findings have a degree of generalisability to higher education institutions in other countries.

This study is the first to test the ASSET model of occupational stress (Cartwright and Cooper, 2002) in an emerging economy. Future research could be conducted on a larger sample in other higher education institutions in China. Multiple data collection techniques should be undertaken (including peer evaluations) and collection of the dependent and independent measures at different time periods. This would ensure that common method variance is minimised. While the current study did not find any support for the statistical relationship between commitment and psychological wellbeing, future study could explore this relationship as a possible means of better understanding the meaning of employee engagement, such as the recent study by Robertson and Cooper (2010). A high commitment human resource management theoretical perspective (Whitener, 2001) could be adopted to further explore the causal relationships between stress, human resource practices, commitment and employee performance.

Note

• 1

  We thank one of the anonymous reviewers for suggesting we analyse the impact of occupational stress according to staff category.

Appendix

Appendix 1. Sample items from the ASSET questionnaire (Cartwright and Cooper, 2002)

• Job insecurity: level of job security (four items): ‘My job is insecure’

• Work relationships: degree of support from colleagues and peers (eight items): ‘I am not sure what is expected of me by my boss’

• Pay and benefits: adequacy of pay (single item): ‘My pay and benefits are not as good as other people doing the same or similar work’

• Job overload: experienced workload and time pressures (four items): ‘I do not have enough time to do my job as well as I would like’

• Job control: amount of control over work (four items): ‘I am not involved in decisions affecting my job’

• Resources and communication: adequacy of the available equipment and resources (four items): ‘I am not adequately trained to do many aspects of my job

• Work-life balance: extent to which the demands of work interfere with personal life (four items): ‘My work interferes with my home and personal life’

• Job dissatisfaction (reflective scale, eight items): ‘My work is dull and repetitive’

• Commitment (reflective scales): perceived commitment of the organisation to the employee (five item scale): ‘I feel valued and trusted by the organisation’

• Commitment (reflective scales): commitment of the employee to the organisation (four items): ‘I feel that it is worthwhile to work hard for this organisation’

Employee health and wellbeing (formative scales):

• • 
  Physical health (six items): ‘How often you have the symptom of insomnia?’
• • 
  Psychological health (11 items): ‘How often you have constant irritability?’