The effect of job demands and job resources on burnout and its impact on work–related attitudes : a study of workload, role conflict, workplace bullying, autonomy and role ambiguity related to burnout and the influence

(1)

MSc thesis

The eﬀect of job demands and job resources on

burnout and its impact on work–related attitudes

A study of workload, role conﬂict, workplace bullying, autonomy and role ambiguity related to burnout and the inﬂuence of burnout on job satisfaction, intention to leave

and work-related well-being.

drs. H.A. Trompetter

student number

10687807

April 8, 2015

(2)

(3)

Abstract

What is the reason that people get sick of their jobs, or otherwise become dissatisfied and leave? This research is about the effect of job demands (workload, role conflict and workplace bully-ing) and job resources (autonomy and role ambiguity) on burnout and its impact on work-related attitudes (job satisfaction, intention to leave and work-related well-being). The research model consists of five dependent variables; burnout, workplace bullying, job satisfaction, intention to leave and work-related well-being. In which workplace bullying also serves as independent vari-able for burnout. Burnout relates as independent varivari-able to job satisfaction, intention to leave and work-related well-being.

The expectations are that job demands contribute to an increased risk of burnout and that job resources reduces burnout. For role ambiguity, one of the job resources, also an effect on workplace bullying is expected. Moderation of job resources, acting as a buffer, on the relation between job demands and burnout are expected. Burnout is expected to influence job satisfaction, intention to leave and work-related well-being.

The relationships are examined by means of Pearson’s correlation analysis and hierarchical re-gression analysis. There is a significant effect if p ≤ .05 (Boer et al., 1994). Regression analysis provides information on the effect of the predictor (β) and shows what the proportion of explained variance in the dependent variable is by R2.

The study is carried out by using a paper–and–pencil questionnaire among 109 respondents work-ing at the same organization of around 200 employees. Compared to the Dutch workwork-ing population the sample counts more female workers, is overrepresented in the age group 26–35 years and the educational level of the respondents is above average. Important findings of the statistical analysis:

• The results show that workplace bullying exhibits the strongest predictor for burnout. • Role ambiguity serves as a plausible explanation of workplace bullying.

• No stronger effect is found on burnout, because of moderation of one of the job resources (autonomy, role ambiguity) with one of the job demands (workload, workplace bullying). • Burnout has negative effects on job satisfaction and work-related well-being, and victims

(6)

(7)

Introduction

The idea of job design as a catalyst for reduced staff turnover and absenteeism, served as the inspi-ration for this research, which is about the effect of job demands and job resources on burnout and its impact on work-related attitudes. The need is obvious; in the Netherlands in 2010 more than 13 percent of the employees are at home, because of burnout signals1.

Many employees suffer from strain and stress as a result of bad job design. In the long run, this may lead to burnout, or some employees decide to quit before its onset. But what is the reason that people get sick of, or otherwise become dissatisfied with their jobs and leave? What are the aspects of a job that contribute to, or can prevent this from happening? In other words, what are the main causes of unhappy personnel? This research revolves around the question; which aspects of a job affect employees negatively.

So how to go about this? The first step is to define all the factors and to order these factors into groups. The first group of factors are the job demands, these are the factors that the employee has no or little control over, namely: workload, role conflict and workplace bullying. The second group of factors are the job resources, these are the factors that the employee has control over, namely: autonomy and role ambiguity. Especially the topic of workplace bullying is interesting, since it has recently become a focus in the research about workplace bullying to investigate its relation with job characteristics and the effects on health outcomes, such as burnout.

The second step is finding out what the relationships are between the job demands and the job resources. From there on, their combined influence on burnout can be determined. Finally, the effect of burnout on reduced staff turnover and absenteeism can be examined. In order to study the effect of burnout on work-related attitudes, the variables job satisfaction, intention to leave and work-related well-being are measured.

In order to find out how to design jobs for happy workers, one must first know the mechanisms that negatively affect them. With this knowledge, the adverse outcomes of reduced staff turnover and absenteeism may be prevented. This study is underpinned by the following research question:

1_Retrieved _from _{http://www.cbs.nl/nl-NL/menu/themas/arbeid-sociale-zekerheid/} publicaties/artikelen/archief/2011/2011-3493-wm.htm on April 8, 2015.

(8)

Introduction

Research question

What is the effect of workload, role conflict, workplace bullying, autonomy and role ambiguity on burnout and what is its impact on job satisfaction, intention to leave and work-related well-being?

This research question will be answered by way of the scientific method, the procedure of which can be illustrated by a series of consecutive steps or phases shown in Figure 1. The structure of this thesis follows this diagram as indicated by the milestones <0…4>, which also represent the respective chapters. 1. observation 2. question 0 3. research 4. hypotheses 1 5. survey 2 6. analysis 3 7. conclusion 4

Figure 1: Phasing diagram of the scientific method, with milestones indicating the chapters (van Hinte & van Tooren, 2008).

This chapter (0) covers the first two steps or phases of Figure 1, namely: ‘observation’ and ‘question’.

In order to operationalize the research question, it will be researched in phase three; ‘research’, this is where the theoretical framework will be presented, explaining burnout, job demand, job resources and work-related attitudes. The models which were used will be discussed, namely: the Job Demands Control Model of Karasek (1979) and the Job Demands-Resources Model of Demerouti et al. (2001). These serve as a basis for a new conceptual model described by the hypotheses that are formulated. The process of formulating hypotheses is indicated by step four in Figure 1. Phase three and four of the scientific method are covered in Chapter 1, as indicated by the milestone.

The next step is the ‘survey’, which will be treated in Chapter 2. This is where the research method used and the measurement procedure are described.

Chapter 3 contains the ‘analysis’ phase (six), and presents the numerical results gathered from the operationalization of the used variables of the conceptual model by the survey.

Finally, the results are summarized in the ‘conclusion’, Chapter 4, with a discussion of the results and suggestions for further research.

(9)

1 Theoretical framework

The central theme of this research is burnout and its relationship to either job demands and job resources or work-related attitudes. In this context, job demands means the combined effect of the variables workload, role conflict and workplace bullying. Job resources are the variables autonomy and role ambiguity. Work-related attitudes are represented by the variables job satisfaction, work-related well-being and an employee’s intention to leave. Burnout is not subdivided and is one variable. In this chapter, these variables and their interrelationships will be visualized in a coherent conceptual model.

1.1 Job design for reduced staﬀ turnover and absenteeism

Between 1880 and 1910, Frederick Winslow Taylor created the trend of Scientific Management which is regarded as the foundation for redesigning jobs. In his method, Taylor separated work tasks from the mental part of the work for the purpose of efficiency and increased control pre-dictability (Huczynski & Buchanan, 2007). The downside of the simplification of jobs was that they became repetitive and dull.

During the economic growth in the US around 1900, work was plentiful and workers had a chance to speak their mind about the simplifications made to their jobs. Strikes occurred, absen-teeism increased and more workers quit to seek their fortune elsewhere. The changes in the labor market spiked renewed interest in redesigning jobs for organizations wanting to mitigate these circumstances.

The subject of redesigning jobs was studied by Hackman & Oldham (1976), who created the Job Characteristic Model (JCM). The JCM describes the effects of job characteristics on the psycho-logical states of employees and the effects on individual and organizational outcomes. The tested job characteristics are skill variety, task identity, task significance, autonomy and feedback. A simplification of the Job Characteristic Model is shown in Figure 2.

(10)

1 Theoretical framework

Job Characteristics Psychological States Outcomes

Figure 2: Job Characteristics Model.

A few years later, Karasek (1979) came up with a more refined model for describing job char-acteristics. Karasek’s Job Demands Control Model (JDC-model) explains different job demands and control mechanisms relating to strain and stress symptoms. The JDC-model considers work-load and conflict as job demands instead of job characteristics such as in the JCM. Moreover, skill direction (not shown) and autonomy are regarded as control options instead of job characteristics. Figure 3 shows a simplification of Karasek’s JDC-model. Applied to our case it looks like Figure 4, with workload and role conflict as job demands, and autonomy as a moderator.

Job Demands Strain

Control

Figure 3: Job Demands Control Model.

Workload, Role

Con�ict Burnout

Autonomy

Figure 4: Applied JDC-model.

Demerouti et al. (2001) generalized the JDC-model and expanded it in order to make it applicable for more types of job demands and job resources. Demerouti et al. (2001) dubbed their model the Job Demands-Resources model (JD-R model). The JD-R model can be seen as an expanded version of the JDC-model, adding the direct relationship between job resources and job demands and displaying the direct effect of reduced health/energy on organizational outcomes. A simplified version of the JD-R model relevant to this research is shown in Figure 5. It can be seen in Figure 6 that workplace bullying is considered a job demand. The effects of role ambiguity and autonomy are seen as job resources.

(11)

1.1 Job design for reduced staff turnover and absenteeism

Job Resources Job Demands

Reduced Health/Energy

Organizational Outcomes

Figure 5: Job Demands-Resources Model.

Autonomy, Role Ambiguity Workplace Bullying Burnout Job Satisfaction, Work Related Well-being, Intention to Leave

Figure 6: Applied JD-R model.

When combined, Karasek’s JDC-model and Demerouti et al.’s JD-R model together give rise to the conceptual model shown in Figure 7 which contains all the studied variables and their re-lationships. The conceptual model in Figure 7 also indicates the expected positive and negative relationships between the independent and dependent variables. Each individual relationship will be repeated separately for each hypothesis to be formulated. Because burnout is the linchpin con-necting the variables it will be addressed first.

(12)

1.2 The concept of burnout at work

In 2010, one out of every eight employees in the Netherlands suffered symptoms of burnout and was unable to do their job1. Burnout, a term coined in the 1970s by the American psychotherapist Herbert Freudenberger and Professor of Psychology Christina Maslach, assumed that it was related to chronic occupational stress. At first, burnout was primarily found in professions where the involvement of human interaction was high, like doctors, nurses and teachers (Bakker & Schaufeli, 2000; Demerouti et al., 2000). Being exposed to ‘people work’ makes one more prone to strain and stress, because these jobs have higher emotional and mental demands than jobs with sparse interaction with people.

Maslach (1982) characterized burnout as “A syndrome of emotional exhaustion, depersonaliza-tion, and reduced personal accomplishment that can occur among people who do ‘people work’ of some kind”. It is therefore a complex of three symptoms, where emotional exhaustion refers to the feeling of being completely empty; all reserves are used up, the battery is exhausted and can no longer be recharged. The mental and physical exhaustion experienced by employees ensures that they will behave cold and detached towards their clients. The latter is called depersonalization and unlike in psychiatry, depersonalization is not about alienation from oneself, but alienation towards others. Typical is a negative, cold, cynical attitude towards the people with whom one has to deal with professionally. Reduced personal accomplishment refers to the feeling of under-performing at work, which is accompanied by feelings of failure and doubt of one’s own abilities (Bakker et al., 2004).

In occupations with less or no ‘people work’, research has shown that burnout also occurs. Only now, the depersonalization is not towards people, but towards the work and the duties one has to perform. In other words, burnout is a syndrome characterized by feelings of exhaustion, increased detachment and a negative perception of one’s competence (Bakker & Schaufeli, 2000).

The cause of strain has been associated with the extent to which an employee has control over his or her tasks during the working day as discussed by (Karasek, 1979). If an employee experiences high work pressure and simultaneously does not have the freedom to determine how to deal with the stress, one could speak of a high workload and low autonomy. Stress occurs when people cannot deal with the demands. The stress in this situation increases the risk of burnout as compared to a situation where there is high autonomy. This is because high autonomy gives an employee more freedom to decide when, what and how to perform than low autonomy. In terms of job design

(13)

1.3 The JDC-model for workload, role conflict and autonomy

this implies that for a job with high demands and stress, either it needs to be compensated with opportunities for higher autonomy, or the workload needs to be adjusted to a more manageable level in order to prevent burnout. In his Job Demand Control Model, Karasek (1979) addresses this by stating that work demands and opportunities for control are interrelated.

1.3 The JDC-model for workload, role conﬂict and autonomy

Robert Karasek created the Job Demand Control Model (JDC-model) in 1979 while working as a sociologist at the University of Massachusetts in Lowell. In this model, demands are job re-quirements such as work quantity and work pace, both are a product of workload. Control is conceptualized by two different components: the ability of workers to learn and use their skills, called skill direction and the ability to have some say or autonomy in the work process. Job auton-omy refers to the employee’s ability or freedom to make decisions about his or her work activities (De Jonge & Kompier, 1997; De Witte et al., 2007). The major implication of Karasek’s work is that by redesigning work processes, mental strain experienced by workers can be reduced by al-lowing them increased decision latitude. Karasek (1979) says that this is possible without affecting the job demands that may plausibly be associated with organizational output levels.

“Having to much work” is a complaint often heard from employees. Evidence that workload induces strain and job autonomy decreases strain is abundant (Häusser et al., 2010). The positive influence of autonomy plays an important part for redesigning jobs as a control mechanism for reducing the effect of workload on burnout. Karasek (1979) discusses that high job demands are not necessarily bad when combined with enough autonomy. Possibilities to create control mechanisms of high autonomy are: taking breaks when you need them, opportunities to do your work at home for example if your child is sick or simply by working more slowly. Adequate control mechanisms make handling high work pressure possible, so that it is perceived as challenging and motivating instead (Karasek, 1979).

Karasek’s Job Demand Control Model builds upon three hypotheses: • workload relates positively to strain (main effect workload),

• job autonomy relates negatively to strain (main effect job autonomy), and • workload and job autonomy interact in a synergistic way (Baillien et al., 2011). These ideas are formulated in the following three hypotheses:

(14)

Hypothesis 1 Workload is positively related to burnout.

Workload

+

Burnout

Hypothesis 2 Autonomy is negatively related to burnout.

Autonomy

-

Burnout

Hypothesis 3 The positive relationship between workload and burnout is moderated by

auton-omy, such that this relationship is weaker for higher values of autonomy.

Workload

+

Burnout

Autonomy

-Besides workload, there are more job-related characteristics that play a role in causing a burnout. Karasek (1979) constructed the scale of job demands for his JDC-model with the goal of measuring the psychological stressors involved in accomplishing the workload. These include conflicting demands as related to strain and stress symptoms such as stressors related to unexpected tasks, and stressors of job-related personal conflict.

Job-related personal conflict is also called role conflict and occurs when two or more sets of role pressures exist in an individual’s workspace. Compliance with one of these pressures impedes the accomplishment of another (Kahn et al., 1964). The negative effect of role conflict is that it introduces uncertainty, because the employee is not certain whether all role requirements are suc-cessfully balanced (Jawahar et al., 2007). Experiencing incompatible or irreconcilable expectations associated with multiple roles, or with a single role, is presumed to be psychologically uncomfort-able for individuals and generates negative emotional reactions (Schaubroeck et al., 1989). These emotional reactions as a consequence of uncertainty accompanying a role conflict are an indicator for feeling stressed. Role conflict is a factor in causing a burnout via stress and needs to be taken

(15)

1.4 The JD-R model for workplace bullying, autonomy and role ambiguity

into account when designing a job. The mechanism of role conflict is illustrated by Lee & Ash-forth (1996) who reported a strong correlations between role conflict and emotional exhaustion, role conflict and depersonalization, and role conflict and reduced personal accomplishment, which are also the three elements of burnout according to Maslach (1982). This is summarized in the following hypothesis:

Hypothesis 4 Role conflict is positively related to burnout.

Role Con�ict

+

Burnout

Although the JDC-model is limited in its primary focus on three broadly defined work character-istics, its simplicity is a major advantage in its applicability. In addition to the job demands, control options and social support, there are also other work characteristics that result in stress, e.g. men-tal, emotional or physical demands. In order to meet these shortcomings of the JDC-model, the research by Demerouti et al. (2001) established a relationship between job demands, job resources and burnout in the Job Demands-Resources Model. The JD-R model is applicable to different work characteristics in order to explain job strain. As opposed to the JDC-model, the JD-R model is not about specific negative or positive work characteristics, but sees each work characteristic as a potential source of stress or energy (engagement) that can be included in the model. Since job design is more about job characteristics than merely job demand and control, the JD-R model is used to explain relationships between workplace bullying, autonomy and role ambiguity and their effect on burnout.

1.4 The JD-R model for workplace bullying, autonomy and role ambiguity

The Job Demands-Resources model (JD-R model) was developed by Demerouti et al. (2001) aim-ing to provide insight into the causes of burnout. In this model, job demands refer to the physical, psychological, social, or organizational aspects of the job that require sustained physical and/or psychological (cognitive and emotional) effort or skills and are therefore associated with certain physiological and/or psychological costs (Bakker et al., 2006). Examples of job demands are high work pressure, an unfavorable physical environment, or emotionally demanding interactions e.g. with clients. Job demands are not necessarily negative, but they may turn into stressors when ef-forts to meet the requirements of the work are either too high or insufficient recovery is taking

(16)

place (Schaufeli & Taris, 2013; Bakker et al., 2006).

Job resources, as used in the JD-R model, are the aspects of a job that are functional in achieving work goals, reduce job demands and reduce the associated physiological and psychological costs of those job demands (Bakker et al., 2006). Job resources are found at different levels and should stimulate an employee’s personal growth, learning and development. At the level of the organi-zation at large, job resources are visible in e.g. what kind of pay, career opportunities and job security the organization offers. Job resources also exist in interpersonal and social relations with supervisor and co-workers or in a team climate. Other places that job resources appear are in the organization of the work, for example, role clarity and participation in decision making, and also at the level of the task which manifests in skill variety, task identity, task significance, autonomy and performance feedback (Bakker et al., 2006).

The JD-R model splits job resources from job demands and discusses the different effects of the job characteristics on stress symptoms and engagement (Bakker et al., 2006). In this research, job demands and job resources are applied as perceived by the employees on stress symptoms like burnout. The organization of the work (read: job design) expressed by role ambiguity and autonomy are used as moderating effects on job demands, as experienced by the employees in the form of burnout. See also Figure 7 the conceptual model for an overview of the relationships. An advantage of the JD-R model is its wide applicability for all types of job demands or job resources. Workplace bullying is another example of a job demand, which can be applied into the JD-R model. This will be discussed in the next section.

Workplace bullying and burnout

The original JD-R model proposes that there are two processes that play a role in the development of burnout. Following Robert & Hockey’s views (1997) on compensatory control, the JD-R model assumes that in a situation of excessive work demands, additional efforts must be made to keep the work performance up to standard. In other words, operating under high work pressure requires extra energy and the price to be paid comes in the form of physical and psychological costs, such as fatigue and irritation. For example, to recover from these extra efforts employees can take a break, switch tasks or work slower. However, employees not able to recover properly remain in a state of sustained activation, which can ultimately lead to physical and mental exhaustion. This is the first process or symptom of burnout as described by Maslach et al. (2001).

The second process, according to the JD-R model, is about a lack of energy which leads to an employee who is not being able to respond adequately to job demands, and is unable to meet set

(17)

work objectives. Over time, not meeting work objectives brings about sustained feelings of under-performing at work, which are often accompanied by feelings of failure and doubt of one’s own abilities (Bakker et al., 2004). The employee experiences reduced personal accomplishment, which is the second symptom of burnout. The combination of a lack of energy and failure to achieve work goals leads to mental detachment in the form of cynicism, which is the third component of burnout (Maslach et al., 2001). In short burnout is a result of high work pressure and high work pressure is seen as a job demand.

Research on burnout shows different demanding processes from work. Besides work pressure, bullying at work is related with burnout. In the last 10 to 15 years more research is being done on the subject of bullying at work (Agervold, 2009). In the meta-analysis of Nielsen & Einarsen (2012) findings show that exposure to bullying is associated with both job-related and health- and well-being-related outcomes, such as mental and physical health problems, symptoms of post-traumatic stress, burnout, increased intentions to leave, and reduced job satisfaction and organizational com-mitment. Organizations and employees alike need to be made aware that exposure to bullying, as a form of inappropriate organizational behavior, has these negative consequences.

The concept of workplace bullying refers to situations where an employee is persistently exposed to negative and aggressive behaviors at work primarily of a psychological nature (Leymann, 1996). Workplace bullying is the persistent exposure to interpersonal aggression and mistreatment from co-workers, managers and subordinates, and is a prevalent problem with devastating effects on both employees and organizations (Einarsen et al., 2009).

Workplace bullying comprises continuously evolving and often escalating relationships with fea-tures of repetition (frequency), duration (over a period of time) and patterning (of a variety of behaviors involved) (Einarsen et al., 2003). Although the negative and unwanted nature of the be-havior involved is essential to the concept of bullying, its core characteristic is not the nature of the behaviors, but rather the persistence of the experience (Einarsen et al., 2003) Thus, the emphasis is as much on the frequency and duration of what is done, as on what is done, and how it is done (Einarsen et al., 2009).

In bullying behavior, a distinction exists between direct actions and indirect acts of aggression. For an explanation of direct actions, one can think of accusations, verbal abuse and public humil-iation. Indirect acts of aggression involve rumors, gossiping and social isolation (O’Moore et al., 1998). A further distinction has been made by Einarsen (1999) between work-related behaviors and person-related behaviors of bullying. So slander, social isolation and insinuation about some-one’s mental health are examples of person-related bullying, and giving a person too many, too few or too simple tasks, or persistently criticizing a person or their work, can be seen as work-related

(18)

bullying (Einarsen et al., 2009).

Workplace bullying can be thought of as an emotional job demand. Threats and aggression associated with workplace bullying yield feelings of powerlessness and results in the victim ex-periencing strain. Workplace bullying as a job demand, is suspected to be related to burnout like other job demands such as workload and role conflict. If there is no opportunity to recover from workplace bullying, especially if it takes place on a regular basis, then the long duration of the bullying will drain the employee’s energy and is likely to result in a burnout. The effect of bullying on burnout is illustrated by the following hypothesis:

Hypothesis 5 Workplace bullying is positively related to burnout.

Workplace Bullying

+

Burnout

Inﬂuences of workplace bullying, autonomy, role ambiguity on burnout

The Job Demands-Resources model in Figure 5 indicates a moderating effect of job resources on job demands and reduced health/energy. Workload as a job demand and autonomy as control was already shown to buffer the impact of strain in the JDC-model by Karasek of Figure 3. The JD-R model expands this view and states that job resources buffer the impact of job demands on job strain, here labeled reduced health/energy, that includes burnout (Bakker et al., 2003). Bakker et al. (2006) also mentions that, if a stressor is predictable or controllable by the person who experiences it, it can be seen as a job resource. Job resources like role clarity and decision autonomy, act as moderators on burnout.

Xanthopoulou et al. (2009) tested part of the JD-R model among employees from two home-care organizations and showed that autonomy is the most important buffer of job demands, in this case patient harassment (read: bullying), in predicting exhaustion and cynicism. Autonomy is expected to show comparable effects in relation to bullying at work and burnout. The latter is encompassed in the following hypothesis:

(19)

Hypothesis 6 The positive relationship between workplace bullying and burnout is moderated

by autonomy, such that this relationship is weaker for higher values of autonomy.

Workplace Bullying

+

Burnout

Autonomy

-Some job resources (autonomy, role clarity, social support) are expected to have a positive effect on burnout (Bakker et al., 2005). Jawahar et al. (2007) discusses that it is important to continue searching for mechanisms that reduce the negative effects of job stress, because job stress has major significance for both employees and organizations. For employees it is frustrating not to be able to participate in the labor market and for the organization there are financial consequences. Role ambiguity refers to the clarity of role requirements at work (Notelaers et al., 2010). The Michigan model of job stress (Kahn et al., 1964) emphasizes role ambiguity as a stressor which leads to strain. This results in the following hypothesis:

Hypothesis 7 Role ambiguity is positively related to burnout.

Role Ambiguity

+

Burnout

Bakker et al. (2005) states that the extent to which the onset of a stressor is predictable (read: role clarity) may act as a moderator/buffer against job strain. In other words, predictability of the stressor is also a job resource which effectively functions as role clarity. Role clarity can be regarded as a job resource which buffers the effects of job demands, e.g workplace bullying, on burnout.

The opposite of role clarity is role ambiguity, i.e., the lack of clarity of role requirements at work. It is expected that role ambiguity, already being a stressor for strain, also shows a mod-erating effect with respect to the relationship of workplace bullying on burnout. Just like role clarity, what acts like moderator/buffer against strain, it is expected for role ambiguity to act like a moderator/increaser of strain. Only role ambiguity is regarded as a job resource, which instead of buffering it, exacerbates the positive relationship of job demands (i.e. workplace bullying) and burnout. It is expected that because of the unpredictability of role ambiguity, role ambiguity may

(20)

act as a moderator and will add stress on top of the already experienced stress level of workplace bullying in relation to burnout:

Hypothesis 8 The positive relationship between workplace bullying and burnout is moderated

by role ambiguity, such that this relationship is stronger for higher values of role ambiguity.

Workplace Bullying

+

Burnout

Role Ambiguity

+

Job resources, such as autonomy and role ambiguity, are elements of a job design. A well de-signed job shows less signs of role ambiguity and imparts enough autonomy to buffer the negative job demands such as workload and role conflict. However, a bully-proof job design has yet to be invented. In other words, a job design can compensate for, but will never be resilient to workplace bullying. In case of an unclear job design, or no job design whatsoever, the resulting role ambi-guity can give rise to workplace bullying. Hoel et al. (2002) and Einarsen et al. (1994) showed that bullying thrives when employees perceive their job requirements as unclear. Therefore role ambiguity as a consequence of an incomplete job design is expected to have an effect on workplace bullying, as formulated by the hypothesis:

Hypothesis 9 Role ambiguity is positively related to workplace bullying.

Role Ambiguity

+

Workplace Bullying

1.5 Burnout and job satisfaction, intention to leave and work-related

well-being

The Job Characteristics Model (JCM) by Hackman & Oldham (1976) proposes that positive psy-chological states enhance an employee’s intrinsic motivation, job satisfaction and performance, while reducing absenteeism and staff turnover. The states that need to be engendered are: the experience of meaningfulness of the work, feelings of responsibility for the outcomes, and under-standing the results of their effort. A meta-analysis by Lee & Ashforth (1996) showed that burnout,

(21)

1.5 Burnout and job satisfaction, intention to leave and work-related well-being

as a state, negatively affects job satisfaction. So burnout is expected to reduce job satisfaction:

Hypothesis 10 Burnout is positively related to job satisfaction.

Burnout

-

Job Satisfaction

It is found that burnout creates undesired behaviors such as personnel turnover and absenteeism (Lee & Ashforth, 1996). Burnout is expected to exacerbate absenteeism and turnover. Staff turnover, can be seen as the result of an employee’s intention to leave the organization:

Hypothesis 11 Burnout is positively related to intention to leave.

Burnout

+

Intention to Leave

Diener et al. (1999) defined well-being as referring to global and rather stable evaluations of one’s life and context-specific aspects, as well as to momentary mood. In this study, well-being is seen as the well-being of an employee at the end of a workday. It refers to an employee’s well-being when leaving the workplace and returning home, Sonnentag (2001) described this as ‘situational being’. Job satisfaction has frequently been used as a measure of workers’ health and well-being (Wanous et al., 1997). Work-related well-well-being has often been narrowly operationalized as job satisfaction (Rothmann, 2008), and because the latter is negatively affected by burnout (Lee & Ashforth, 1996), it is expected for burnout to have a negative effect on work-related well-being:

Hypothesis 12 Burnout is negatively related to work-related well-being.

Burnout

-

Work Related Well-_Being

To make a statement about which Hypothesis 1–12, will be supported/rejected, first all the vari-ables in question will be measured. These are: burnout, the job demands; workload, role conflict and workplace bullying, the job resources; autonomy and role ambiguity, and the work-related at-titudes; job satisfaction, intention to leave and work-related well-being. The next chapter is about the research design and describes how the variables are measured.

(22)

(23)

2 Research design

This chapter is about the method of data collection and it presents the description of the sample. The used research instruments for each independent and dependent variable are shown by opera-tionalizing them. Furthermore, the data analysis of the research is given.

2.1 Method of data collection

The present study is about the influence of job demands (workload, role conflict and workplace bullying) and job resources (autonomy and role ambiguity) on burnout and the effect of burnout on work-related attitudes (job satisfaction, intention to leave and work-related well-being).

Before the actual data collection took place, a preliminary data research was done at the orga-nization where the research is performed. In order to gain access to the data a meeting with the CEO was scheduled for permission. The preliminary study gathered the rates of absenteeism and turnover over a period of two years, namely from September 2012 to September 2014. This was done to ensure that the addressed problems in the research question also exist in this organization. The data is collected in the Netherlands at an organization with around 200 employees.

The data collection is conducted among the employees of one organization by using a survey. The questionnaires used to collect the data are distributed among the working population before, during and after the Christmas holidays in 2014. The data is collected at one moment in time like a cross-sectional survey. A paper-and-pencil questionnaire is used and personally handed over by the researcher to each of the respondents. The distributing of the questionnaire was ad-random among the working population of the organization. To every respondent the researcher introduced the questionnaire, almost equally by explaining it was voluntary and anonymous. The questionnaire was a self-report filled out by the respondents. The respondents took the questionnaires with them; most were handed in at the same day they were handed out.

All the employees names, who were given an questionnaire, were written down on a note block, so no one could be asked twice. The questionnaire is added in the Appendix A (in Dutch).

(24)

2 Research design

Before the questionnaire was launched, a test group of two academic friends provided feedback about the questionnaire. This resulted in shortening the questionnaire from 99 items to 57 items. They gave comments on overlapping items, unclarity in items, the long duration of answering the questionnaire (between 30 and 40 minutes) and the boredom that arose by completing it.

The questionnaire was shortened by deleting the 12 items scale of job related affective well being by Warr (1990), because this scale is often used in diary studies instead of samples and therefore did not fit this type of measurement. The added value of this scale was also questionable, because well-being was already measured with the scale of Sonnentag (2001). Therefore the scale was removed.

In the first version of the questionnaire workload was measured by 11 items of work pace and workload from the questionnaire experience and judgement of work (VBBA) by Van Veldhoven & Meijman (1994). The final questionnaire only measured workload with five items, because after reconsideration work pace was not applicable for the type of organization the questionnaire was administered.

Another adjustment to shorten the questionnaire was made by using another measurement for autonomy. Instead of the 11-item version of Van Veldhoven & Meijman (1994) the three-item scale of decision authority as a measurement for autonomy of the Job Content Questionnaire of Karasek (1998) was used.

The seven-item scale of performance feedback was removed from the questionnaire for a less complex research model. Research of Humphrey et al. (2007) discussed feedback as a social char-acteristics of job charchar-acteristics explaining a variances of 24 percent in turnover intentions. In an earlier stage of this research design feedback seemed important to measure. Only the scope of the research changed during the research process, as a consequence performance feedback was taken out of the research.

Finally, the control variable for part-time or full-time contract was removed, because the value perception of the results are seen as important for all employees regardless of their contract type.

Description of the sample

The research sample is an organization in the Netherlands, which counted in December 2014, 194 employees whose 49 percent men and 51 percent women. The organization employs, given or take, 20 managers and 40 specialists and has an operational staff around 120 employees. The remainder of the employees (about 60) are administrative employees or middle managers.

As a rough rule of thumb, within the limit of respondents’ time availability, N = 100 is deemed

(25)

2.2 Description of the research instruments

adequate. To test the individual predictors within the model 104 respondents plus one for each predictor is the aim. Workload, role conflict, workplace bullying, autonomy and role ambiguity are the five predictors in this study, so 109 respondents is the minimum required for this study as final sample.

2.2 Description of the research instruments

The methods used to collect the data are described and explained in this section. The different scales are adopted from previous researches and are all used with their original items. If possible Dutch scales were selected, because the respondents are working in the Netherlands and most of them are Dutch. If this was not the case, the scale is translated using a back-translation-back method in consultation with an excellent English writer and speaker, whose native language is Dutch. The description of the variables from the conceptual model are operationalized below.

Workload

Van Veldhoven & Meijman (1994) developed the questionnaire ‘experience and judgment of work’ (VBBA), by an analysis of 50 existing instruments in the field of psycho-social workload and job stress. The original scale of (Van Veldhoven & Meijman, 1994) measured work pace and workload by using 11 items. Drenth & Sijtsma (1990) suggests a reliable scale must have a least a α of .80. The reliability of the 11-item scale Cronbach’s α = .89 is excellent. Only the five items that measured workload are used in this questionnaire. One of those items is coded reversed. Respondents are asked to answer on a four-point Likert scale from ‘never’ to ‘always’. A higher score refers to the experience of higher workload. Example items are ‘Do you have too little work?’ (reversed coded) and ‘Are you behind in your work?’. The original items are in Dutch and identically formulated in this study.

Role Conﬂict

Karasek (1979) originally discussed besides workload in the Job Demands Control Model also conflict which placed the individual in a motivated or energized state of “stress”. In the ques-tionnaire of Van Veldhoven & Meijman (1994) items are based on Karasek (1985) ‘Job Content Questionnaire’. The six-item scale of Van Veldhoven & Meijman (1994) is used in this question-naire, because it is a tested Dutch scale with a reliability of α = .80. Respondents are asked to

(26)

2 Research design

answer on a four-point Likert scale from ‘never’ to ‘always’. A higher score refers to the experi-ence of higher role conflict. Example items are ‘Are you getting conflicting orders?’ and ‘Do you ever have a disagreement with your supervisor about the content of your duties?’. The original items are in Dutch and applied in this study.

Workplace Bullying

Workplace bullying is measured with the 9-item Short Negative Acts Questionnaire for targets by Notelaers & Einarsen (2008). The underlying factors are four items of personal bullying and five items of work-related bullying. Respondents need to indicate how often they are exposed to workplace bullying on a five-point frequency scale ranging from ‘never’ to ‘daily’. A higher score refers to the more exposure to personal and work-related bullying, measured in time. Example items are ‘Are you having repeated reminders of your errors or mistakes?’, which measures personal bullying and ‘Are you being ordered to do work below your level of competence?’, as a form of work-related bullying. The original items are in English and therefore translated with the back-translation-back method, reliability of α = 85 (Notelaers & Einarsen, 2008).

Autonomy

Autonomy is measured by the three items of decision authority of the Job Content Questionnaire (JCQ) of Karasek (1998). Those items are scaled on a five-points from ‘completely disagree’ to ‘completely agree’. All items are formulated in the same positive direction. A higher score means that the respondent experienced higher autonomy, because of more decision authority. Example items are ‘My job allows me to make a lot of decisions on my own.’ and ‘I have a lot to say about what happens on my job’. The original scale is in English and therefore translated with the back-translation-back method (α = .81).

Role Ambiguity

Role ambiguity is measured with six items of the questionnaire ‘experience and judgment of work’ (VBBA) of Van Veldhoven & Meijman (1994), which they framed as ‘unclarity about the task’ (α = .81). Respondents are asked to answer on a four-point Likert scale from ‘never’ to ‘always’. A higher score refers to the experience of lower role ambiguity, or in other words more role clarity. In order for the items to be formulated in the same direction as their answer scale all the items need

(27)

2.2 Description of the research instruments

to be recoded. Example items are ‘Are you aware of what others expect of you at work?’ and ‘Do you know your responsibilities?’. The original items of VBBA are in Dutch and used in this study.

Burnout

The concept of burnout is measured by the Utrechtse Burnout Scale (UBOS) of Schaufeli & Dieren-donck (2000), the Dutch version of the Maslach Burnout Inventory. Is sees burnout as three parts, namely exhaustion, cynicism and professional efficacy. The reliability and validity is good, an in-ternal consistency was given of .70 for exhaustion and cynicism and .60 for professional efficacy. The 15 items are measured on a five-point scale from ‘completely disagree’ to ‘completely agree’ in stead of the original seven-point scale. Six of the 15 items in the UBOS are coded reversed. A higher score refers to an employee with more symptoms of burnout. Example items are ‘I feel mentally exhausted by my work’ and ‘I doubt the usefulness of my work’. The original items of UBOS are in Dutch and are taken over in the questionnaire.

Job Satisfaction

The meta-analysis by Wanous et al. (1997) compared single-item measures to multi-item measures for job satisfaction and found no differences in the results. In this study job satisfaction is measured with single-item namely: ‘Are you satisfied with you job?’. This item is answered on a five-point Smiley Faces Scale; 😭, ☹, 😐, ☺, 😃.. The Faces Scale of job satisfaction measures overall job satisfaction with just one item which participants respond to by choosing a face. Originally the scale had male faces developed by Kunin (1955) and 20 years later the matching female faces were created by Dunham & Herman (1975). A higher score means an employee is more satisfied with his or her job.

Work-related Well-being

Sonnentag (2001) measured situational well-being at the end of the workday when employees are leaving their workplace and returning home. This three-item scale of well-being is measured with a five-point Likert-type items from ‘never’ to ‘always’. One of the items is coded reversed. A higher score means that the well-being is higher as a positive consequence of the work, a low score means the well-being of the employee was negatively influenced by his or her work. Example items are ‘I felt tense when coming home from work.’ (reverse coded) and ‘I was in a good mood

(28)

2 Research design

when coming home from work.’. The original items are in English and therefore translated with the back-translation-back method, with a reliability of α = .76.

Intention to Leave

Intention to leave is measured using four items of Van Veldhoven & Meijman (1994) questionnaire ‘experience and judgment of work’ (VBBA). The scale is measured on a ‘no’ or ‘yes’ scale, with a Cronbach’s α = .90. More items answers with yes means a higher intention to leave the organi-zation. Example items are ‘Are you thinking about changing jobs?’ and ‘I am planning to change my job next year’. The original items of VBBA are in Dutch and takenover in the questionnaire.

Control variables

The control variables: gender, age, level of education, organizational tenure, department and man-agerial job are added to the survey, all with a category scale. The variable department is added to the questionnaire as favor for the organization were the survey was conducted. Therefore it is not included as control variable in this study. Each of the control variables could have an unintended effect on the results of the research. Therefore the effect of the control variables on the relationship of the different independent and dependent variables is reported during the measurements.

2.3 Data analysis

The data of the questionnaires are entered in Statistical Program for Social Sciences (SPSS). Ques-tionnaires with missing items are not removed, the missing items are coded as missing values. Only the items that are answered, are taken into account in the analysis.

All the variables are checked with frequencies reports for input errors. The numbers of the questionnaires responds with the numbers of the input lines. Therefore it is possible to check the input errors with the numbered questionnaires, to restore them or replace them for missing.

The reversed-coded items are recoded so all the answers have the same direction, before analysing the data.

For each scale a reliability test is performed, which measures the internal consistency of the scales by Cronbach’s α. Drenth & Sijtsma (1990) suggests a reliable scale must have a least a α of .80. In this research, as a rule of thumb, scales are seen as reliable between α .70–.80. If a scale scores below α = .60, this is seen as not acceptable and if possible the scale need to be corrected. A

(29)

2.3 Data analysis

low scale score can be corrected by deleting an item of the scale, if an item scores on the item-total correlation < .20 than it is qualified to delete (Boer et al., 1994). An item-total correlation explains which responses on the item do not vary in line with those of other items. By deleting such an item the reliability of the scale can increase.

1.0 1.6 2.2 2.8 3.4 4.0 Workload 0 10 20 30 40 50 60 Frequency SD = 0.49 M = 2.07 N = 109 1.0 1.6 2.2 2.8 3.4 4.0 Role Conflict 0 10 20 30 40 50 60 Frequency SD = 0.39 M = 1.71 N = 109 1.0 1.8 2.6 3.4 4.2 5.0 Workplace Bullying 0 10 20 30 40 50 60 Frequency SD = 0.5 M = 1.72 N = 109 1.0 1.8 2.6 3.4 4.2 5.0 Autonomy 0 10 20 30 40 50 60 Frequency SD = 0.85 M = 3.36 N = 109 1.0 1.6 2.2 2.8 3.4 4.0 Role Ambiguity 0 10 20 30 40 50 60 Frequency SD = 0.48 M = 1.81 N = 109 1.0 1.8 2.6 3.4 4.2 5.0 Burnout 0 10 20 30 40 50 60 Frequency SD = 0.48 M = 2.04 N = 109 1.0 1.2 1.4 1.6 1.8 2.0 Intention to Leave 0 10 20 30 40 50 60 Frequency SD = 0.41 M = 1.4 N = 109 1.0 1.8 2.6 3.4 4.2 5.0 Work-related Well-being 0 10 20 30 40 50 60 Frequency SD = 0.78 M = 3.76 N = 109

Figure 8: Bar chart for independent and dependent variables with mean M , standard deviation SD and sample size N .

(30)

2 Research design

Two version of the SPSS data are saved, one with the original ‘raw’ data and the other one with the ‘adjusted’ data.

To demonstrate the distribution of the data for each variable Figure 8 shows the bar charts. The answer categories are on a two-, four- and five-point scale, just like the scales of the variables. The bar charts shows that the respondents score above average on the scales for autonomy and work-related well-being. So, answers for autonomy and work-work-related well-being scored relatively high. As opposed to workplace bullying, role conflict and role ambiguity, which score below the scale mean. Workplace bullying, role conflict and role ambiguity were experienced to a lesser degree. Autonomy and work-related well-being have the largest standard deviation. The mean and the standard deviation of all variables are shown in chapter three in the correlation Table 3.1.

The normal distribution of the scale is checked by skewness and kurtosis levels, around zero a variable is normal distributed (Field, 2009). In Table 2.1 the skewness and kurtosis levels of the variables are given. Except for ‘burnout’, all the other the variables have a normal distribution. The distribution of ‘burnout’ shows a kurtosis of 1.74. A positive kurtosis distribution has high frequen-cies in the tail and is pointy. The answers are given in a smaller range for burnout. Burnout has a skewness of 1.07, the distribution is asymmetrical because the most frequent scores are grouped towards the left of the distribution. This means that a majority of the respondents have less burnout symptoms.

Table 2.1: Skewness and kurtosis levels of the variables.

Variables Skewness Kurtosis

Workload .49 -.36 Role Conflict .49 .84 Workplace Bullying .64 -.18 Autonomy -.31 -.12 Role Ambiguity -.28 -.43 Burnout 1.07 1.78 Job Satisfaction -.76 .31 Work-related Well-being -.10 -.87

The relationships between job demands, job resources, burnout and the different work-related attitudes are examined by means of Pearson’s correlation analysis. There is a significance effect if p ≤ .05 (Boer et al., 1994).

The research model consist of five dependent variables: workplace bullying, burnout, job satis-faction, intention to leave and work-related well-being. In which workplace bullying also serves as

(31)

2.3 Data analysis

independent variable related to burnout, and burnout served as independent variable for job satis-faction, intention to leave and work-related well-being. In order to measure the conceptual model, five hierarchical regression analysis are undertaken. Regression analysis provides information on the effect of the predictor β and shows what the proportion of explained variance in the dependent variable is, measured by R2. There is a significance effect if p≤ .05 (Boer et al., 1994).

Before the regression analysis for the moderators were undertaken, all moderating variables are standardized into z-scores. Furthermore, product terms of the standardized independent variables workload and workplace bullying with the standardized moderating variables autonomy and role ambiguity were made.

The first regression analysis is conducted with the dependent variable burnout and as indepen-dent variables: workload, autonomy, role conflict, role ambiguity and workplace bullying. In step one, the standardized control variables tenure, gender, age, manager and education are added to the model. In the second step the standardized independent variables workload, autonomy, role conflict, role ambiguity and workplace bullying were are to the model. Finally, in step three prod-uct terms of the standardized independent variables with the standardized moderating variables are added to the model. The steps one and two were also undertaken for the regression model with workplace bullying as the dependent variable. And the steps one and two were performed for the regression model with job satisfaction, intention to leave or work-related well-being as the dependent variables.

The regression analysis for burnout is checked for multicollinearity between the predicting vari-ables workload, role conflict, autonomy, role ambiguity and workplace bullying and it is also mea-sured for the moderating variables. If the Variance Inflation Factor (VIF) was smaller than 10 and the Tolerance was greater than .10 than there are no collinearity issues (Field, 2009). The outcomes of the regression analysis for burnout showed for the collinearity statistics of the predicting vari-ables had a VIF that was smaller than 2 and it showed that the Tolerance of all varivari-ables were greater than .50. So, there are no collinearity issues found between the predictor variables or moderators for burnout.

(32)

(33)

3 Research results

This chapter presents an analysis of the results from the survey. It starts with the findings of the preliminary study, after which a treatise is given on the processing of the raw data, followed by a description of the characteristics of the respondents. The reliability tests of the used scales and the results of the correlations between the variables are shown. The chapter ends by presenting the results of the regression analysis for each dependent variable in order to test the hypotheses.

3.1 Preliminary study

The research was done in one organization. Before conducting any research, the absence- and turnover data has been analyzed over a period of two years in order to gain more knowledge about of this particular organization and to obtain a reference value for comparison of the variables ‘burnout’ and ‘intention to leave’.

The preliminary study showed a mean for absenteeism of 4.93 percent, which is 1.53 percent higher when compared to the national average in this industry (3.4 percent) in 20131.

Based on the turnover rates over the past two years (2012–2014) the organizational turnover of the organization has an average of 20 percent a year, one out of five of the employees left the orga-nization. This was partly voluntary and a part involuntary as a consequence of the reorganization in 2012. National turnover rates in the Netherlands of 2013 showed an average of 6 percent2.

By comparing those numbers to the national average rates, the organization in this research scores high on absenteeism and extremely high on turnover. This organization seems suitable for the research compared to organizations where there are no or lower turnover and absenteeism rates.

2_Retrieved _from http://www.penoactueel.nl/HR-Strategie/Algemeen/2013/6/Nederland-laagste-personeelsverloop-van-Europa-1276224W on April 8, 2015.

(34)

3 Research results

3.2 Processing the raw data

The descriptives and frequencies check showed some input errors of the data. This was corrected by checking the case numbers in SPSS with the identical numbered questionnaires and by taking over the answers for these items. The respondents of this research overall felt satisfied with their jobs. This is shown in Figure 9.

1 2 3 4 5 Job Satisfaction 0 10 20 30 40 50 60 Frequency SD = 0.84 M = 3.95 N = 109

Figure 9: Bar chart of job satisfaction.

The items as shown in Appendix B .1 were recoded before analyzing the data. After recoding, for each variable the mean of all the items were conducted into a total scale. The analysis was performed on the total scale of each variable.

3.3 Characteristics of the respondents

Out of 120 employees approached, 109 filled out and returned the paper-and-pencil questionnaire. The sample size is N = 109 with a the response rate of 91 percent. Based on the information of the control variables from the questionnaire the following general facts are known about the respondents.

The gender of the respondents was for 51 percent men and 49 percent women. The mean (M ) age was 34.8 years and the standard deviation (SD) of the age was 13.5, ranging between 20–64 years. Figure 10 shows the distribution of age and gender in a bar chart. The biggest category of

(35)

3.3 Characteristics of the respondents

respondents were women between the 26 and 35 years old.

< 25 26 − 35 36 − 45 46 − 55 > 56 Age [yr] 0 5 10 15 20 25 30 Frequency N = 108 _men women

Figure 10: Bar chart displaying age and gender distribution of the sample.

The educational background of the respondents was as follows; two percent primary school and 18 percent had an education background with a secondary educational program. Forty of the respondents studied an applied education at the middle and higher level (MBO 17 percent, HBO 23 percent) and 40 percent followed an academic education.

The mean of the organizational tenure of the respondents was four years, with a SD of 5.8 years and the organizational tenure ranged in between 0–37 years.

The extremes of organizational tenure are caused by the 54 percent of the employees who work three years or less for this organization. The origin of this situation lies in the expanding of this organization for 50 percent over the last three years. All departments are represented in the survey therefore the results are representative for the total organization.

Compared to the Dutch working population, the sample population counts more female than male workers and is overrepresented in the age group of 26–35 years, also the educational level is higher than average. 3

3_Retrieved _from _{http://www.cbs.nl/nl-NL/menu/themas/arbeid-sociale-zekerheid/} publicaties/barometer-beroepsbevolking/barometer-werkzame-beroepsbevolking-art.htm on April 8, 2015.

(36)

3.4 Reliability test research instruments

In this research, a scale is regarded reliable if it has a Cronbach’s α in the range .70–.80. The higher the α, the more reliable the scale is. However, when a scale scores an α below .60, it is regarded as unreliable and no conclusions can be drawn on the results. As a last resort, a scale will be corrected by deleting an item from the scale (Boer et al., 1994). To see which item can be deleted, the corrected item-total correlation was checked. The corrected item-total correlation shows the differences of correlation between the items of one scale. This number needs to be lower than 0.20 in order to increase α (Boer et al., 1994). The lowest scores represents the items responsible for the lower α and by deleting such an item it is possible to increase α and gain a higher reliability of the scale.

The results of the reliability test are presented in Table 3.1. Most of the reliabilities of the scales scored alpha’s ranging between .74 and .81. The exception was workplace bullying, scor-ing α = .59, which is too low and was corrected in the followscor-ing way. Workplace bullyscor-ing was measured by nine items in total, and by deleting item six: “Are you reminded by others about your mistakes or errors?”, the reliability of the scale increased to α = .62, which was acceptable. The other reliabilities for the job demands were workload α = .77 and role conflict α = .75. The job resources scored for autonomy α = .81 and the scale for role ambiguity had an α = .74. For mea-suring burnout, the UBOS was used and scored α = .81. The scales for the work-related attitudes scored the following reliabilities; work-related well-being α = .80 and intention to leave α = .79. For the dependent variable job satisfaction there was no reliability test possible, because the scale had one item. All the α are presented in Table 3.1 on the diagonal between parentheses.

3.5 Correlation analysis

The correlation analysis of Table 3.1 revealed that the job demands (workload, role conflict, work-place bullying) and the job resources (autonomy, role ambiguity) correlated significantly with burnout. As expected, workload (r = .13, p < .05), autonomy (r = .-37, p < .05) and role con-flict (r = .46, p < .05) correlated significantly with burnout. The negative correlation for autonomy means that in situations of less autonomy, more burnout was measured.

Also workplace bullying (r = .66, p < .05) correlated significantly with burnout. The analysis shows a significant positive effect of role ambiguity (r = .44, p < .05) on burnout. This means that a higher role ambiguity was accompanied with higher burnout.

Role ambiguity was expected to correlate positively with workplace bullying, because bullying

(37)

3.5 Correlation analysis T able 3.1 : Mean M ,standard de viation S D ,Pearson’ s r, and Cronbach’ s α for all variables. V ariable M S D 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 Gender 1.49 .50 (-) 2 Age 2.87 1.33 -.20 (-) 3 Education 5.61 1.65 .16 -.36 (-) 4 Organizational Tenure 2.04 1.47 -.11 .63 -.17 (-) 5 Manager 1.26 .44 -.07 -.01 .10 -.07 (-) 6 W orkload 2.07 .49 .21 .01 .25 .05 .30 (.77) 7 Autonomy 3.36 .85 -.09 .06 .19 .15 .41 .27 (.81) 8 Role Ambiguity 1.81 .48 .13 -.16 .03 -.16 .15 .35 -.07 (.74) 9 Role Conflict 1.71 .39 .11 -.07 .13 -.10 .10 .23 -.26 .41 (.75) 10 Intention to Leave 1.40 .41 .17 .-43 .29 -.25 .11 .09 -.17 .22 .29 (.79) 11 W orkplace Bullying 1.72 .50 .12 .-25 .11 -.23 .14 .27 -.30 .49 .50 .44 (.62) 12 Burnout 2.04 .48 .13 -.14 .03 -.11 -.03 .13 -.37 .44 .46 .36 .66 (81) 13 W ork-related W ell-being 3.76 .78 -.25 .19 -.05 .07 -.08 -.34 .17 -.45 -.40 -.44 -.50 -.67 (.80) 14 Job Satisfaction 3.95 .84 -.12 .09 -.11 -.04 -.07 -.19 .29 -.37 -.46 -.48 -.60 -.67 .60 (-) Note: N = 109. Correlations similar to or smaller than r = -.10 and correlations similar to or greater than r = .10 are significant at p < .05. Gender coded as 1 = men, 2 = women. Age is coded as 1 = < 25; 2 = 26–35; 3 = 36–45; 4 = 46–55; 5 = > 56. Education is coded from 1 meaning ‘Primary’, 2 = ‘MA VO’, 3 = ‘VMBO’, 4 = ‘HA VO/VWO’, 5 = ‘mbo’, 6 = ‘hbo’, 7 = ‘WO’, 8 = ‘Other’. Organizational Tenure is coded as 1 = smaller than three years; 2 = between 4 and 7 years; 3 = between 8 and 11 years; 4 = between 12 and 15 years and 5 = 16 years or more. Manager coded as 1 = ‘No’ and 2 = ‘Y es’. Intention to Leave coded as 1 = ‘No’ and 2 = ‘Y es’. Reliabilities are positioned on the diagonal.

(38)

behavior has a chance if work roles are not clear. In the correlation analysis role ambiguity (r = .49, p < .05) was positively and significantly related to workplace bullying. This means higher role ambiguity resulted in higher workplace bullying.

The correlation analysis shows that autonomy (r = -.30, p < .05) had a direct significant effect on workplace bullying. This negative effect means that lower autonomy was associated with higher workplace bullying.

Two other not hypothesized significant correlations were found. The first is between workplace bullying and workload (r = .27, p < .05) and the second is between workplace bullying and role conflict (r = .49, p < .05). These effects were not predicted, although the results for workload and role conflict to correlate significantly with workplace bullying is not new. Workload, role conflict and workplace bullying were already measured in research of Notelaers et al. (2010) to correlate significantly.

As expected, the correlations between burnout and the different work-related attitudes were sig-nificant. Between burnout and job satisfaction (r = -.67, p < .05) a significant correlation was found. The negative correlation means that less job satisfaction was experienced in situation of more burnout. Also significant correlations of burnout were established with work-related well-being (r = -.67, p < .05) and intention to leave (r = .36, p < .05). In case of more burnout symptoms to occur less work-related well-being and a more intention to leave was experienced. Figure 11 shows the scatter plots of three strong correlations for workplace bullying and three strong corre-lations for burnout.

Table 3.1 shows the mean M , standard deviation SD, Pearson’s r and Cronbach’s α of all vari-ables. Correlations similar to or smaller than r = -.10 and correlations similar to or greater than r = .10 are considered significant p < .05. Reliabilities for each scale were positioned on the diag-onal.

Control variables

Except for the dependent and independent variables, the control variables also demonstrated sig-nificant correlations. The five control variables were tenure, gender, age, manager and education. Only the significantly correlated control variables for the dependent variables are discussed below. The correlation analysis showed that the control variables age, gender and organizational tenure correlated significantly with burnout. Someone’s age (r = -.14, p < .05) and organizational tenure (r = -.11, p < .05) correlated negatively. This means that the younger employees who worked shorter at the organization, experienced more burnout symptoms. For gender (r = .13, p < .05)

(39)

3.5 Correlation analysis 1 2 3 4 5 Workplace Bullying 1 2 3 4 5 Burnout

r

= 0.66

p

= 0.000 1 2 3 4 5 Workplace Bullying 1 2 3 4 Role Ambiguity

r

= 0.49

p

= 0.000 1 2 3 4 5 Workplace Bullying 1 2 3 4 Role Conflict

r

= 0.5

p

= 0.000 1 2 3 4 5 Burnout 1 2 3 4 5 Autonomy

r

= -0.37

p

= 0.000 1 2 3 4 5 Burnout 1 2 3 4 Role Conflict

r

= 0.46

p

= 0.000 1 2 3 4 5 Burnout 1 2 3 4 5 Work-related Well-being

r

= -0.67

p

= 0.000

Figure 11: Scatter plots of six strong correlations for workplace bullying and burnout.

a positive correlation was found with burnout. In the survey the answer category one stood for men and category two for women. This result showed that higher scores on gender, or the women

The effect of job demands and job resources on burnout and its impact on work–related attitudes : a study of workload, role conflict, workplace bullying, autonomy and role ambiguity related to burnout and the influence

MSc thesis

The eﬀect of job demands and job resources on

burnout and its impact on work–related attitudes

drs. H.A. Trompetter

10687807

April 8, 2015

Contents

Abstract

Introduction

Research question

1 Theoretical framework

1.1 Job design for reduced staﬀ turnover and absenteeism

1.2 The concept of burnout at work

1.3 The JDC-model for workload, role conﬂict and autonomy

+

-

+

+

1.4 The JD-R model for workplace bullying, autonomy and role ambiguity

+

+

+

+

+

+

1.5 Burnout and job satisfaction, intention to leave and work-related

well-being

-

+

-

2 Research design

2.1 Method of data collection

2.2 Description of the research instruments

2.3 Data analysis

3 Research results

3.1 Preliminary study

3.2 Processing the raw data

3.3 Characteristics of the respondents

3.4 Reliability test research instruments

3.5 Correlation analysis

r

p

r

p

r

p

r

p

r

p

r

p