Relationship between job control, work pressure and strain : studies in the USA and in The Netherlands

Similar documents
The association between job skill discretion, decision authority and burnout

THE RELATION BETWEEN JOB CHARACTERISTICS AND QUALITY OF WORKING LIFE: THE ROLE OF TASK IDENTITY TO EXPLAIN GENDER AND JOB TYPE DIFFERENCES

IMPACT OF JOB SATISFACTION ON QUALITY WORK LIFE AMONG THE IT EMPLOYEES

Gender and employees job satisfaction-an empirical study from a developing country

MONITORING ABSENTEEISM: GUIDELINES FOR DIAGNOSIS, INTERVENTION AND CHANGE. Peter G W. Smulders TNO Institute of Preventive Health Care, Leiden

Impact of work variables and safety appraisal on well-being at work

The moderating role of task characteristics in determining responses to a stressful work simulation

Stressful jobs and non-stressful jobs: a cluster analysis of office jobs

Chapter 3. Bernadette Willemse Jan de Jonge Dieneke Smit Marja Depla Anne Margriet Pot. International Journal of Nursing Studies 2012: 49(7):

Master thesis Humans Resource Studies. The effect of a job crafting intervention and proactive personality on work engagement

Community Mental Health Journal, Vol. 40, No. 1, February 2004 ( 2004)

QUALITY INSTITUTIONALIZATION IN THE PUBLIC SECTOR: THE WISCONSIN DEPARTMENT OF REVENUE STUDY

Nasrin Arshadi a *, Hojat Damiri a

Dutch monitor on stress and physical load: risk factors, consequences, and preventive action

PSYCHOLOGICAL STRAIN AS THE MEDIATOR IN THE RELATIONSHIPS BETWEEN WORK DESIGN AND WORK ATTITUDES AMONG MALAYSIAN TECHNICAL WORKERS

BURNOUT, LOCUS OF CONTROL AND JOB SATISFACTION. A STUDY ON HIGH SCHOOL TEACHERS

A STUDY ON LINKING ORGANIZATIONAL RESOURCES, WORK ENGAGEMENT AND SERVICE CLIMATE AT FASHION RETAILS OF KOCHI.

A Study on the Relationship Between Job Satisfaction and Contextual Performance of Knowledge Workers

A Note on Sex, Geographic Mobility, and Career Advancement. By: William T. Markham, Patrick O. Macken, Charles M. Bonjean, Judy Corder

SUPPORTIVE LEADERSHIP- A CONCEPTUAL STUDY

Construct, antecedents, and consequences 1

Residential care sector: Working conditions and job quality

Perceived Masculinity/Femininity of Managers and the Feedback Environment

Pay fairness and employee outcomes: Exacerbation and attenuation evects of nancial need

Metal industry: Working conditions and job quality

Effects of Personal Variables of Call Centre Employees on their. Psychological well being

HEA 676 Women s Health Research Seminar Brittney Anderson, Brian McDonald, Casey Buss, Laura Jones, Louise Vincent with Paige Hall Smith

Please respond to each of the following attitude statement using the scale below:

2012 International Symposium on Safety Science and Technology Research on the relationship between safety leadership and safety climate in coalmines

Perception of Women Executives towards Balancing their Work Life and Personal Life

Wholesale sector: Working conditions and job quality

CHAPTER 4 RESEARCH FINDINGS. This chapter outlines the results of the data analysis conducted. Research

The Effects of Workplace Spirituality and Work Satisfaction on Intention to Leave

Author please check for any updations

A STUDY OF RELATIONSHIP BETWEEN EMPLOYEE EMPOWERMENT, AFFECTIVE, NORMATIVE AND CONTINUANCE COMMITMENT IN HOTEL INDUSTRY

Factors affecting organizational commitment of employee s of Lao development bank

PUTTING THE CONTEXT BACK IN JOB CRAFTING RESEARCH

A more specific test. The demands-control model of job strain: Toby D. Wall", Paul R. Jackson, Sean Mullarkey and Sharon K. Parker

Explaining Organizational Responsiveness to Work-Life Balance Issues: The Role of Business Strategy and High Performance Work System

The relationship between empowerment and job satisfaction: Applied study on Jordanian Textile Companies

FACTORS AFFECTING JOB STRESS AMONG IT PROFESSIONALS IN APPAREL INDUSTRY: A CASE STUDY IN SRI LANKA

A STUDY ON EFFECT OF ORGANIZATIONAL ROLE STRESS ON JOB SATISFACTION AMONGST WORKING WOMEN

A Study on Motivational Factors in the Workplace (MODI-Paints), Ghaziabad, UP

THE IMPACT OF EMPOWERMENT AND QUALITY CULTURE ON JOB SATISFACTION

Supervisor and organizational factors associated with supervisors' support for work accommodation in low back pain injured workers

[04] Emotional Labour and Job satisfaction: A Case Study on Bank Tellers in Sri Lanka. Abstract

JOB CONTROL AND BURNOUT ACROSS OCCUPATIONS '

IMPACT OF CORE SELF EVALUATION (CSE) ON JOB SATISFACTION IN EDUCATION SECTOR OF PAKISTAN Yasir IQBAL University of the Punjab Pakistan

Empirical Study on Long Working Hours and Unpaid Working Time in Japan. Summary


Management Science Letters

Journal of Business & Economics Research December 2006 Volume 4, Number 12

Factors Affecting Job Satisfaction of College Lecturers in Ludhiana-An Empirical Study

CHAPTER 5 ANALYSIS OF WORK LIFE BALANCE WITH RESPECT TO OTHER VARIABLES

The Impact of Occupational Stress on Employees Somatic Symptoms, Job Anxiety and Employee s Turnover Intention An Empirical Study

Banking sector: Working conditions and job quality

An Empirical Study on Organizational Climate in Neycer India Limited, Vadalur

Assessing the Business Case for Flexible Work Arrangements

Age differences in coping and locus of control : a study of managerial stress in Hong Kong

Other service activities: Working conditions and job quality

Nathan Hickey. Rest Assured Financial Advisers, New Zealand. Adrian France. Waikato Institute of Technology (Wintec), New Zealand.

The Relationship between Quality of Work Life and Demographic Characteristics of Information Technology Staffs

Perception of Women Executives Towards Balancing Their Work Life And Personal Life

The Impact of Human Resource Management Functions in Achieving Competitive Advantage Applied Study in Jordan Islamic Bank

CHAPTER 2 THEORETICAL FRAMEWORK

Relationship of Demographic Variables and Job Satisfaction among Married Women

DOES AUTHENTIC LEADERSHIP PROMOTE EMPLOYEES ENTHUSIASM AND CREATIVITY?

Individual or collective voice

CHAPTER 5 DATA ANALYSIS AND RESULTS

The Workplace in Recession: Working Conditions and Wellbeing

The Relationships among Organizational Climate, Job Satisfaction and Organizational Commitment in the Thai Telecommunication Industry

CHAPTER 4 METHOD. procedures. It also describes the development of the questionnaires, the selection of the

Stress Coping Strategies Vs Job Satisfaction - An Analytical Approach

A study of relationship between organizational justice and job satisfaction among teachers in Bandar Abbas middle school

On growing your own future leaders: Succession planning practices of hospices

Keywords: Organizational justice; Organizational commitment; Turnover intention; Pharmaceuticals company: Medical representatives

Job involvement in Iranian Custom Affairs Organization: the Role of Organizational Justice and Job Characteristics

Research on the Job Stress and Job Satisfaction of Social Workers at Welfare Institution for the Disabled

Occupational stress and job demand, control and support factors among construction project consultants

Chapter 3. Different Combinations of Job Demands and Resources Predict Burnout

A Study of Component Gender in Job Satisfaction of University Lecturers

Effect of Employees Communication and Participation on Employees Job Satisfaction: An Empirical Study on Airline Companies in Iran

STUDY BACKGROUND AND MOTIVATION

Chinese Validation of the Team Climate Inventory: A Measure of Team Climate for Innovation in R&D Teams

Effect of Emotional Intelligence and Perceived Work Environment on Performance of Bankers

QUALITY OF WORK LIFE: A STUDY OF JAMMU UNIVERSITY TEACHERS

INVESTIGATING THE RELATIONSHIP BETWEEN ORGANIZATIONAL LEARNING AND FLEXIBILITY OF EMPLOYEES IN EDUCATION DEPARTMENT OF SARPOOL-E-ZAHAB CITY

ORGANIZATIONAL COMMITMENT AND ITS IMPACT ON EMPLOYEE BEHAVIOR

Relationship between age and occupational stress among IT employees

UAF Administrative Services Work Environment Survey. Prepared for: University of Alaska, Fairbanks Administrative Services

ORGANIZATIONAL COMMITMENT AMONG PHYSICAL EDUCATION ABSTRACT

A STUDY ON ROLE OF OCCUPATIONAL STRESS ON EMPLOYEES PRODUCTIVITY

The effect of work locus of control on the relationship between inclusive leadership and work engagement.

Western Kentucky University Staff Satisfaction Survey

Topic: Readiness to Organizational Change: The Impact of Employees Commitment to the Organization and Career

CREATING CONSTRUCTIVE WORK ENVIRONMENT TO IMPROVE QUALITY OF WORK LIFE OF COIMBATORE-BASED IT PROFESSIONALS

FEMALE FACULTY ORGANIZATION SUPPORT AND COMMITMENT IN SAUDI ARABIA: THE FOCUS OF HAIL UNIVERSITY

Role of Internet Experience in Moderating Influence of Work Stressor on Cyberloafing

Impact of Work-related Stress on Well-being among Academician in Malaysian Research University

Job Satisfaction of Knowledge Workers in Croatian Companies

Transcription:

work & stress, 1999,vol. 13, no. 1 32± 48 Relationship between job control, work pressure and strain : studies in the USA and in The Netherlands PASCALE CARAYON* Ecole des Mines de Nancy, Parc de Saurupt, 54042 Nancy Cedex, France FRED ZIJLSTRA Work and Organization ResearchCenter, Tilburg University, 5000 LE Tilburg, The Netherlands Keywords : Job control ; Work pressure ; International study. The purpose of this study was to test a model of job control, work pressure and strain in two samples in the USA and in The Netherlands. According to the proposed model, work pressure has a mediating role between job control and worker strain. The present model speci es three types of job control: task or instrumental control, conceptual or resource control, and decision} organization control. Task and resourcecontrol are assumed to reduce the stressor of work pressure which, in turn, is related to high strain. Organization control is assumed to be positively related to work pressure. The model was tested in two samples : one US sample of 273oYce employees and one Dutch sample of 958 employees in a variety of jobs. The same measure of job control was used in both samples. Resultspartially con rm the proposed model. As predicted by the proposed model, high task control was related to low work pressure, whereas high organization control was related to high work pressure. The evects of task control and organization control on strain were mediated by work pressure for the measure of anxiety in the US sample and for the measure of stress in the Dutch sample. However, the mediating evect of work pressure was not found for job satisfaction in both samples, for the measure of mood disturbances in the US sample and for the measure of sick days in the Dutch sample. The proposed model is only partially con rmed by the results. These results con rm the importance of diverentiating between diverent dimensions of job control. 1. Introduction The occupational stress literature has shown that lack of job control can be an important source of strain in various occupations (Karasek, 1979; Sauter, Hurrell, and Cooper, 1989; Spector, 1986).Although Karasek (1979)suggested an interaction evect model between job demands and control on worker health, other models of the evect of job control on worker health have been proposed (Frese, 1989; Sauter et al., 1989). The authors view on job control is based on the `stressor reduction mechanism described by Frese (1989). Job control is seen as an important means to reduce work pressure, which has been shown to be an important work stressor in diverent occupations (Frese, 1989; Karasek, and Theorell, 1990). In many studies on job control, the concept of control has been conceptualized and measured as a single dimension, usually referred to as `decision latitude (Karasek, 1979; * Author for correspondence. e-mail : carayon! mines.u-nancy.fr This research was performed while Pascale Carayon was at the Department of Industrial Engineering of the University of Wisconsin-Madison, USA. Work & Stress ISSN 0267-8373 print} ISSN 1464-5335 online 1999 Taylor & Francis Ltd http :} } www.tandf.co.uk } JNLS} wst.htm http :} } www.taylorandfr ancis.com} JNLS} wst.htm

Job control, work pressure and strain 33 Karasek, and Theorell, 1990). However, more recently, researchers have proposed a multifaceted conceptualization of control (Dwyer, and Ganster, 1991; Jackson, Wall, Martin, and Davids, 1993; Smith, Tisak, Hahn, and Schmeider, 1997). The aim of the present study was to test a model of the evects of diverent dimensions of job control and work pressure on strain in diverent occupations and in diverent countries. The proposed model incorporates three dimensions of job control : task or instrumental control, resource or conceptual control, and decision} organization control. According to the present model, the diverent dimensions of job control have diverent relationships with work pressure and strain. 2. Research model Two work factors are included in the research model : work pressure and job control. Three types of job control are de ned: task or instrumental control, conceptual control and organization control. The research model de nes relationships among the concepts of work pressure, job control and strain. 2.1. Concepts of work pressure and job control Work pressure is de ned as the perception of high job demands that never seem to diminish, that include tight deadlines and that people have a hard time in keeping up with. Work pressure is found in many work environments nowadays (Andries, Kompier, and Smulders, 1996). Work pressure has been found as a critical determinant of worker stress and health, especially in oyce} computer work (Carayon, 1995). Many studies on occupational stress show that lack of job control is an important stressor (Sauter et al., 1989). However, the conceptualization of job control in most of these studies appears to be limited. In the literature, various viewpoints and perspectives on job control can be found. Levinson (1972) used the term of mastery, which is closely related to the concept of skill discretion of the Job Demands-Control Model (Karasek, and Theorell, 1990). For some authors, job control is related to worker participation in the organization, autonomy in the task or decision authority (Ganster, and Fusilier, 1989; Gardell, 1982; Hackman, Oldham, Janson, and Purdy, 1975; Karasek, 1979). Greenberger, Strasser, Cummings, and Dunham (1989) de ne control as a psychological construct re ecting an individual s belief in his} her ability to avect a change, in a desired direction, in the environment. Control, in this latter perspective, is clearly de ned as personal control, and has a strong subjective orientation. It is also clear that personal control is not solely related to work situations. Personal control has also been conceptualized as a personality characteristic determining the way in which people behave in and outside work settings (see, for example, the literature on self-eycacy (Wood, and Bandura, 1989)). Job control is often conceptualized as a one-dimensional concept: one has more or less control or in uence in his} her job. However, an important question to be asked is `control or in uence over what? This question is particularly important in work settings because many diverent aspects of work can be in uenced. In the perspective of (re)designing jobs and organizations for reducing or controlling occupational stress, one needs to pay attention to the type and level of control. The question of the dimensionality of the job control concept has only recently received some attention in the occupational stress literature (Jackson et al., 1993).While some researchers have proposed conceptualizations of diverent facets and levels of job control (see for example Sauter et al., 1989), very little empirical research has examined diverent facets and levels of job control (Hurrell, and McLaney, 1989; Sainfort, 1991; Wall, Jackson, and Mullarkey, 1995).

34 P. Carayon and F. Zijlstra Figure 1. Researchmodel. In an organization, people have diverent positions, according to which they may have more or less authority or formal power. Depending on his} her position and job, and the management and organizational style within the organization, an individual may in uence various aspects of work, such as the physical aspects of the workplace (arrangement of furniture, light, temperature), the task (working method, speed), the group (or organization), procedures and rules, and organizational policies. A conceptual distinction can be made between hierarchically related levels of job control. Three levels of control have been distinguished (Gardell, 1982; Sainfort, 1991): (1) instrumental control, which is primarily related to in uence that can be exerted over the tasks (task order, pace, amount of work) ; (2) conceptual control, which is related to the context in which tasks are accomplished, and the working methods ; and (3) decision control, which is related to in uence over organizational processes, procedures and policies. These three levels of job control represent a hierarchy, where the lowest level of control deals with the task (i.e. instrumental control) and the highest level deals with aspects of the organization (i.e. decision control). Sainfort (1991) examined three diverent levels of job control (instrumental control, conceptual control and decision control) in a group of computer users. She found that the three levels of job control had diverent relationships with strain outcomes. In the context of new manufacturing technologies and practices, Wall et al. (1995)examined two diverent aspects of job control: timing control and method control. Timing control refers to `the individual s opportunity to determine the scheduling of his} her own work, whereas method control refers to `the choice of how to carry out given tasks respectively. Wall and his colleagues tested a new questionnaire of timing and method control in a group of 1691 manufacturing employees, and demonstrated its validity and reliability. The authors concept of instrumental control includes the two dimensions of job control examined by Wall et al. (1995), that is timing control and method control. Greenberger et al. (1989) have developed a questionnaire for measuring job control. McLaney, and Hurrell (1988) have used this questionnaire in a large sample of nurses, and performed a factor analysis. The analysis yielded a four-factor solution that represented the following four dimensions of control: task control (e.g. control over task order, working methods, speed and time), decision control (e.g. control over task assignment and policies), environment control (e.g. control over the physical environment), and resource control (e.g. control over the availability of supplies, materials and equipment). In the present study, the authors used Greenberger s questionnaire to measure diverent dimensions of job control. The concept of instrumental control can be related to the measure of task control, the concept of conceptual control to the measure of resource control and the concept of organization } decision control to the measure of decision control.

Job control, work pressure and strain 35 2.2. Development of the research model Frese (1989) has suggested several mechanisms of the evects of job control on stress and health. One of these mechanisms is called the `stressor reduction mechanism. According to this mechanism, there is a direct evect of control on stressors : control is supposed to be negatively correlated with stressors. This conceptualization was used to build the research model that is proposed and tested in this paper ( gure 1). According to the proposed model, job control is a means for reducing work pressure. In this model, work pressure plays a mediating role between job control and strain outcomes. In addition, the authors diverentiate between three levels of job control. These three levels of job control are assumed to have diverent relationships with work pressure and strain ( gure 1). The rst part of the research model is similar to the `stressor reduction mechanism of Frese (1989).It is hypothesized that job control, in particular task and resource control, can be used to reduce work pressure (i.e. the stressor), therefore reducing stress and improving health. A similar model of the evect of job control on other job stressors has been tested by Carayon and colleagues (Carayon, Jarvenpa$ a$, and Hajnal, 1993; Jarvenpaa, Carayon, and Lim, 1994). According to the stressor reduction mechanism of Frese (1989), task and resource control can be the means for an individual to reduce work pressure. The third level of job control, organization control, is assumed to have a diverent role. Having high organization control means that one can in uence group and organization processes and procedures. It may, for instance, imply that one has authority to decide whether to take on (new or additional) tasks. Typically, an individual with high organization control is responsible for many tasks and projects. Within this category one can nd the so-called `active jobs (Karasek, and Theorell, 1990). People in those positions may have the authority to delegate certain tasks, but they remain responsible for the tasks and need to ensure that the tasks are executed correctly and on time. Therefore, the authors hypothesis is that people with high organization control will experience high work pressure (see the positive relationship between organization control and work pressure in the model in gure 1). Furthermore the authors assume that work pressure is an important mediating variable in predicting evects on health and well-being. This means that the stressor reduction model of Frese (1989)should be modi ed. It is assumed that there is no direct relation between job control and stress evects: the evects of job control on strain are mediated by experiences of work pressure. This model of job control, work pressure and strain was tested in two samples : one US sample of oyce workers and one Dutch sample of diverent occupations. This research is `transnational (Kohn, 1987)in which a model is tested in two countries. This type of crosscultural research assumes that the proposed model is valid in both countries. 3. Methods Two cross-sectional surveys of US and Dutch employees provided the data to examine the proposed research model. Although these surveys were conducted in the context of other research programmes, both used the same measures of job control. This data was used to test the research model. 3.1. Samples 3.1.1. US sample : The US sample comprised 273 oyce workers from one large public service organization in the Midwest. The study participants performed a wide range of

36 P. Carayon and F. Zijlstra oyce jobs, including clerical and administrative jobs, professional jobs (e.g. accountants, computer specialists), and supervisory } management jobs. The majority were female (74%) and married (63%). The mean age was 43 years (SD 5 9.5 years, range : 19± 67 years). The mean tenure within the organization was 13.8 years (SD 5 9.3 years, range : 1 month± 41 years), while the mean experience with current job was 6.7 years (SD 5 5.9 years, range: 1 month± 31 years). A total of 34% had less than high-school education, 18% had completed high school, 22% had some college experience, and 27% had completed college. Twelve per cent of the sample said that they supervised someone. The mean number of persons supervised was seven persons (SD 5 ve persons, range: 1± 26 persons). Two divisions of a large public service organization in the Midwest of the USA were selected for participation in the study. Employees in these two divisions performed a range of oyce tasks, using computers extensively. Agreement was achieved with the management and union representatives about the study procedures. The organization gave each employee on-the-job time to ll out the questionnaire survey. Employees were invited to attend meetings during which the study was explained. At the end of the study presentation by the researchers, employees were asked to participate in the study. Employees who agreed to participate in the study were invited to stay in the meeting room to ll out the questionnaire. All questionnaires were collected at the end of the meetings by the researchers. Using this procedure, a response rate of 85% was achieved. 3.1.2. Dutch sample : The Dutch sample comprised 958 employees from many diverent organizations. The Dutch survey was a cross-sectional survey on work and work-related aspects (in particular leadership) for which a random sample of over 4000 households had been selected from a telephone directory. Potential participants were contacted by telephone. Two-thousand households that had one or more persons actively participating in the labour market and who were willing to participate in the survey received a questionnaire by mail. Within a 2-week period a total of 958 valid questionnaires were returned, resulting in a response rate of 48%. According to Dutch statistics on the national work force (Centraal Bureau voor Statistiek (CBS), 1996), the service sector appeared to be over-represented in the sample, but there was a good t between the sample and the Dutch working population with respect to other economic branches, thus guaranteeing a substantial variation concerning type and size of organizations within the sample. The job categories in the survey included technical} professional employees (21%), clerical workers (15%), middle management (13%), and production} maintenance personnel (15%). The majority of the respondents (70%) were male. The mean age was 39.4 years with a range of 17± 62 years. The mean tenure within the organization (present position) was 8 years (range of from 1 month to 44 years), and 38% of the respondents said that they supervised someone. The mean number of persons supervised was 18.5 (SD 5 51 persons, with a range of 1± 550 persons). About 10% of the respondents had less than high-school education, 48% had completed a professional training, 24% had a college level education, and 8% had a university degree. About 55% of the respondents used a computer during work, and on average the respondents spent 15% of their work time working with computers. 3.2. Measurement 3.2.1. Measures of job control : In both studies the Job Control Questionnaire developed by Greenberger et al. (1989)was used. [The list of questions is available from the authors.] The second author (FZ) translated Greenberger s questionnaire into Dutch. The translation was

Job control, work pressure and strain 37 checked by a native speaker, and ambiguous items were discussed by the two authors. McLaney, and Hurrell (1988)performed a factor analysis of this Job Control Questionnaire in a sample of 675nurses from various hospitals. The analysis yielded a four-factor solution: (1) task control (e.g. task order, work pace: seven items), (2) decision control (e.g. task assignment, policies and procedures: four items), (3) physical environment control (e.g. arrangement of furniture : two items), and (4) resource control (availability of supplies, materials and equipment : two items). The internal consistency of the scales was established with Cronbach a scores. The Cronbach a scores were.85 for the scale of task control,.74 for the scale of decision control,.79 for the scale of environmental control, and.82 for the scale of resource control. High scores on the measures of job control mean more control. 3.2.2. Measures of work pressure : In the US study, the measure of work pressure was a scale developed by Sainfort (1990). The scale is composed of six Likert-type items, asking questions such as `How often do you feel pushed by deadlines? and `To what extent do you face a backlog of work?. The Cronbach a score of the scale was.78. In the Dutch study a scale that focused on con icting work demands was used as the measure for work pressure. It was assumed that when people encountered a lot of con icting demands at work, they would experience high work pressure. The scale consists of eight items with questions such as `I often have to avoid the rules in order to execute my job, `In my job I often have to deal with con icting interests of people, `I seldom get the required materials or supplies that I need to do my job, and `I often lack the manpower to do the job (Rizzo,House,and Lirtzman, 1970).This scale had a Cronbach a score of.75. High scores on the two measures of work pressure mean more pressure. 3.2.3. Measures of worker strain : In the US study, three measures of worker strain were used. First, a scale of anxiety consisting of seven items asked about various anxiety-related symptoms experienced within the past year (Sainfort, and Carayon, 1994; Smith, Cohen, Stammerjohn, and Happ, 1981).TheCronbach a score of the anxiety scale was.83. Second, a score of total mood disturbances was based on the Pro le of Mood States scale (McNair, Lorr, and Droppleman, 1971). The Cronbach a score of the total mood disturbances scale was.89. The third measure of worker strain was a measure of facet-free job satisfaction that was developed in the Quality of Employment Survey (Quinn, Seashore, Kahn, Mangion, Cambell, Staines, and McCullough, 1971). The Cronbach a score for the job satisfaction scale was.83. A high score on anxiety means more anxiety, a high score on mood disturbances means more disturbances, and a high score on job satisfaction means high satisfaction. In the Dutch questionnaire, a short 6-item Likert-type scale focusing on the outcomes of work was used to measure work stress. This scale consisted of items such as `My feeling is that I have to do it all alone, `This work has a real evect on my health, `My job is very demanding and requires a great deal of vigorousness, and `There is so much to do that I often have the feeling that the work outgrows my capacity. The scale had a Cronbach a score of.73. A high score on stress means more stress. Another scale was used to measure job satisfaction (Taillieu, 1987). This 6-item Likerttype of scale had a Cronbach a score of.82, and included items such as `I am satis ed with the variety in my job, `I am satis ed with the degree of responsibility, `I am satis ed with the type of work, and `I am satis ed with the opportunities to use my skills and experiences. A high score on job satisfaction means more satisfaction. Respondents were also asked to indicate how many days they had been on sickness leave in the past year, as an indication of their general health situation.

38 P. Carayon and F. Zijlstra 3.3. Statistical analyses First, a factor analysis of the Job Control Questionnaire was performed for the US and the Dutch data separately. The principal component factor analysis method with Varimax rotation was used. Second, the relationship between the demographic variables of age, gender, and level of education and the measures of work pressure and strain was examined using analysis of variance and correlation analysis. The third step of the statistical analyses consisted of a correlational analysis in which the various scales of job control were correlated with the measures of work pressure and strain. In the fourth step, regression analysis was performed to examine the degree to which the job control scales predicted work pressure, and strain outcomes. Path analyses were performed to test the proposed research model, that is to examine the mediating evect of work pressure in the relationship between job control and strain. 4. Results 4.1. Development of the job control measures 4.1.1. Factor analyses : Factor analyses of the job control questionnaire were performed on the US and Dutch data sets. A principal component factor analysis method with Varimax rotation was used in both data sets. The number of factors was decided on the basis of the Kaiser s rule (eigenvalues " 1), the scree test and the interpretabilit y of the factor solution. In the US study, a four-factor solution seemed to t the data best. The Kaiser s rule gave a four-factor solution, which was con rmed by the examination of the scree plot. In addition, the four-factor solution tted the ndings of the factor analysis performed by McLaney, and Hurrell (1988). The results of the four-factor solution are displayed in table 1. Four scales of job control were constructed in the same way as in McLaney, and Hurrell (1988). (1) Decision control or organization control: four items (in uence over the distribution of tasks, the timing of tasks, policies and training of other employees). This scale corresponds to Factor 1 of the factor analysis. In the US sample, the Cronbach a score for this scale was.90. (2) Instrumental control or task control: seven items (in uence over the variety of tasks, the order of tasks, the amount of work, work pace, work quality, work schedule and rest breaks). This scale corresponds to Factor 2 of the factor analysis. In the US sample, the Cronbach a score was.77. (3) Conceptual control or resource control: two items (in uence over availability of supplies and equipment, and materials). This scale corresponds to Factor 3 of the factor analysis. In the US sample, the Cronbach a score was.57. (4) Environment control: two items (in uence over the arrangement and decoration of work area, and the arrangement of furniture and other work equipment). This scale corresponds to Factor 4 of the factor analysis. In the US sample, the Cronbach a score was.79. The factor analysis of the Dutch data resulted in a four-factor solution, tting the ndings of McLaney, and Hurrell (1988) (table 1). The Cronbach a scores for the four resulting scales were:.87 for organization control,.72 for task control,.83 for resource control, and.82 for environment control. The results of the factor analyses in the US and the Dutch samples are quite similar. In both samples, a four-factor solution emerged that is very similar to the factor solution obtained by McLaney, and Hurrell (1988).However, there were a few diverences between the factor solutions obtained in this study and the factor solution of McLaney, and Hurrell

Job control, work pressure and strain 39 Table 1. Results of the factor analysis.* Factor 1 Factor 2 Factor 3 Factor 4 Communality US Dutch US Dutch US Dutch US Dutch US Dutch Organization Distribution.86.84.17.09.13.15.09.12.84.74 Timing.80.71.19.30.19.17.10.28.73.70 Policies.83.77.18.16.27.20.07.27.79.74 Training.76.78.09.04.17.13.04.07.62.63 2 2 2 2 2 2 2 2 Task Schedule.49.45.28.48.16.06.35.40.47.60 Variety.42.50.54.45.34.31.09.01.58.55 Order.22.27.58.59.37.18.20.14.56.48 Amount.13.24.85.70.05.15.15.04.77.57 Pace.04.02.87.78.05.18.15.09.78.65 Quality.15.01.57.36.07.43.05.08.36.32 Rest.11.13.07.49.24.09.69.52.56.53 2 2 Resource Supplies.16.22.21.16.86.88.01.12.82.78 Materials.35.27.07.14.76.80.16.24.73.80 Environment Decoration.17.27.22.01.14.00.66.76.53.74 Furniture.46.39.06.05.20.22.65.75.68.76 % Variance 24% 39% 18% 10% 12% 8% 11% 7% Total explained variance explained 66% 64% * The factor loadings greater than.50 are in bold characters, while the factor loadings of between.40 and.50 are in italic characters. Distribution 5 in uence on distribution of tasks within the work unit; Timing 5 in uence on timing of tasks within the work unit; Policies 5 in uence on policies, procedures and performance within the work unit ; Training 5 in uence on training of other workers within the work unit; Schedule5 in uence on one s work schedule; Variety 5 in uence on one s variety of tasks ; Order 5 in uenceon task order ; Amount 5 in uenceon one s amount of work ; Pace 5 in uenceon one s work pace ; Quality 5 in uence on one s work quality ; Rest5 in uence on one s rest break schedule; Supplies 5 in uence on availability of supplies and equipment needed in one s work ; Materials 5 in uence on availability of materials needed in one s work ; Decoration 5 in uence on arrangement and decoration of one s work area ; Furniture 5 in uence on the arrangement of furniture and other work equipment in one s work area. (1988).The items of schedule, quality and rest did not t as nicely on the task control factor as they did in McLaney, and Hurrell s study (1988).In both samples, the item of control over schedule had loadings of around.50 on the factor of organization control, and the item of control over rest breaks had high loadings on the environmental control factor (.69 for the US sample and.52 for the Dutch sample). In the Dutch sample, the item of control over quality had a loading of.36 on task control and a loading of.43 on resource control. Given the ambiguity of these three items, it was decided to perform the statistical analyses of the proposed model using two diverent versions of the task control scale: (1) the task control scale similar to McLaney and Hurrell s, and (2) the task control scale without the ambiguous items of schedule, quality and rest. Since both analyses yielded the same results, the authors report here the results of the analyses using McLaney and Hurrell s scales for comparative purposes. The scale of environment control was dropped from any further analyses because it was not included in the proposed model. 4.1.2. Validity checks : In order to examine the validity of the job control scales, people in supervisory jobs and in non-supervisory jobs were compared on the three job control scales,

40 P. Carayon and F. Zijlstra Table 2. Comparison of supervisory and non-supervisory jobsð results of ANOVAs. Non-supervisory Supervisory Mean SD Mean SD ANOVA US sample (n 5 32) (n 5 235) Task control 2.88.72 3.80.50 p!.001 Organization control 1.80.81 4.05.99 p!.001 Resourcecontrol 2.51.97 3.63.79 p!.001 Dutch sample (n 5 335) (n 5 580) Task control 3.22.74 3.89.68 p!.001 Organization control 2.38.82 3.68.83 p!.001 Resourcecontrol 3.27.81 3.82.71 p!.001 using the US and the Dutch data. It was assumed that people with supervisory responsibility would report higher levels of job control than people without any supervisory responsibility. Analysis of variance was used to compare the levels of job control reported by people in supervisory and non-supervisory jobs. The results of the ANOVAs performed on the three job control scales for both samples are reported in table 2. In the US sample, the ANOVAs showed that people in supervisory jobs (n 5 32) reported higher levels of task control, organization control and resource control than people in non-supervisory jobs (n 5 235).The diverences between the two groups were statistically signi cant for all three scales of job control. For the Dutch sample, a similar result was found: people in supervisory jobs (n 5 355) reported signi cantly higher levels of organization control, task control, and resource control than people in non-supervisory jobs (n 5 580). In both samples, the ANOVAs showed that people in supervisory jobs reported higher levels of task control, organization control and resource control than people in non-supervisory jobs. 4.2. Analysis of demographic and background variables In the US sample, the relationship between demographic } background variables and the three measures of worker strain was examined using correlation and ANOVAs. Gender and marital status were related to anxiety, but not to mood disturbances and job satisfaction : women reported a higher level of anxiety than men (ANOVA, p!.001), and married persons reported lower anxiety than non-married persons (ANOVA, p!.05). Age, tenure with the company and experience with current job were not related to anxiety. Age and tenure with the company were related to job satisfaction : older persons reported higher job satisfaction (r 5.20, p!.01) and people with higher tenure reported higher job satisfaction (r 5.14, p!.05). Age was also related to mood disturbances : older employees reported lower mood disturbances (r 5 2.16, p! 0.05). For the Dutch sample, similar analyses were performed. It appeared that age was not related to the number of sick days, work pressure and stress. However, age was related to job satisfaction : older people were more satis ed with their jobs (r 5.14, p!.001). A multivariate analysis of variance was performed with gender and level of education (high, low), and supervisory position (yes, no) as independent variables, and work pressure, sick days, stress, and job satisfaction as dependent variables. Gender was related to work

Job control, work pressure and strain 41 Table 3. Pearson correlations between the study variables. (a) US sample, (b) Dutch sample. Work Mood Job (a) pressure Anxiety disturbances satisfaction 2 2 2 2 2 2 Task control.09.09.22***.21*** Organization control.23***.09.03.17** Resourcecontrol.12.02.07.01 Work pressure.21***.11.12 Work Job (b) pressure Sick days Stress satisfaction 2 2 2 2 2 2 2 Task control.08*.09**.05.41*** Organization control.11***.13***.12***.33*** Resourcecontrol.07.10**.01.41*** Work pressure.03.55***.30*** ***p!.001, **p!.01, *p!.05. 5!!! 5!! 5 pressure: men had signi cantly more work pressure (20.2 versus 17.8 for women; F(1,850) 11.8; p.001). Level of education was related to stress, work pressure, and job satisfaction : people with a higher level of education reported more stress (22.0 versus 20.6 for lower educated people ; F(1,850)5 17.2; p.001) and work pressure (20.3 versus 19.1 for lower educated people ; F(1,850)5 13.3; p.001), and had slightly less job satisfaction (42.3 versus 42.8 for lower educated people ; F(1,850)5 4.6; p.03). Being in a supervisory position was related to work pressure, strain, and job satisfaction : supervisors had more work pressure (20.9 versus 18.7 for non-supervisors ; F(1,850)5 19.1; p.001), more strain (22.2versus 20.4 for non-supervisors ; F(1,850)5 17.9; p.001),and more job satisfaction (43.9 versus 41.8 for non-supervisors ; F(1,850)5 10.7; p.001). An interaction evect between gender and level of education was found with respect to stress : highly educated women reported more stress. All remaining statistical analyses (correlational and regression analyses) were performed with and without the demographic} background variables, and results were similar for both analyses. Therefore, for the sake of simplicity, the results of the analyses are reported without the demographic} background variables. 4.3. Correlational analysis 4.3.1. US study : The Pearson correlations between the study variables in the US data set are shown in table 3. Work pressure was related to one of the three job control scales. As expected, high work pressure was related to high organization control. None of the measures of job control was related to anxiety, whereas work pressure was related to anxiety. Only task control had a signi cant correlation with mood disturbances. High task control was related to low total mood disturbances. Job satisfaction was related to two measures of job control, i.e. task control and organization control. High levels of task and organization control were related to high job satisfaction.

42 P. Carayon and F. Zijlstra 4.3.2. Dutch study : The Pearson correlations between the study variables in the Dutch sample are shown in table 3. The results show positive correlations between, on the one hand, organization control and, on the other hand, stress, work pressure, and job satisfaction, while task control correlated negatively with work pressure and number of sick days, and positively with job satisfaction. Resourcecontrol was not signi cantly correlated with work pressure, but was related to low number of sick days and high job satisfaction. Work pressure correlated quite highly with stress (r 5.55, p!.001). This high correlation was expected: the pressure scale measures the (con icting) demands of work, while the stress scale measures the strains that people experience at work. High work pressure was also related to low job satisfaction. 4.4. Path analyses For the path analyses an identical approach was used for both samples. In the rst step, a multiple regression was performed with work pressure as the dependent variable and the measures of job control as predictors. In the second step, measures of strain were used as the dependent variables and the dimensions of job control as predictors. In the third step, regression analyses were performed with the strain measures as the dependent variables, and job control measures and work pressure as independent variables. The comparison between the beta-coeycients of the measures of job control obtained in the second and third steps can give one an indication of the mediating evect of work pressure. If job control has a signi cant beta-coeycient in the regression without work pressure, and if the coeycient becomes non-signi cant when work pressure is added in the regression analysis, it can be concluded that work pressure has a mediating evect. In the US sample, the measures of anxiety, mood disturbances and job satisfaction were used as indicators of strain. In the Dutch sample, the stress scale, number of sick days and job satisfaction were used as indicators of strain. The results of the path analyses are shown in table 4. As can be seen in table 4, organization control and task control made a signi cant contribution towards predicting work pressure. As expected, for task control, the betacoeycient was negative, and for organization control the beta-coeycient was positive. It is interesting to note that both dimensions of job control, that is task and organization control, made a signi cant contribution towards predicting anxiety, but this contribution disappeared when the variable of work pressure was included in the regression model. Work pressure then appears to have played a signi cant role in predicting anxiety. In the case of anxiety, work pressure played a mediating role between job control and strain. However, this was not true for mood disturbances and job satisfaction. In the case of mood disturbances, task control and organization control contributed signi cantly to the variance explained, even when work pressure was included in the regression model. High task control and low organization control were related to low mood disturbances. Work pressure did not in uence mood disturbances. As for job satisfaction, task control, organization control and work pressure contributed signi cantly to the variance explained. High task control and organization control, and low work pressure were related to high job satisfaction. Resourcecontrol did not contribute to the variance of work pressure and any of the strain measures. The same pattern of results was found in the Dutch data (table 4). Task and organization control made a signi cant (but opposite) contribution in predicting work pressure and stress. Task control was negatively related to work pressure and stress, whereas organization control was positively related to work pressure and stress. When work pressure was entered

Job control, work pressure and strain 43 Table 4. Results of the path analyses. (a) US sample, (b) Dutch sample. (a) Dependent variables Anxiety Mood disturbances Job satisfaction Without With Without With Without With Independent Work work work work work work work variables pressure pressure pressure pressure pressure pressure pressure 2 2 2 2 2 2 2 2 2 2 2 2 Task control.31***.17*.12.31***.29***.20**.17* Organization control.36***.20**.14.21**.19**.13.16* Resourcecontrol.07.04.05.05.05.12.11 Work pressure.18**.05.13* R 12%*** 3%* 6%** 8%*** 8%*** 6%*** 8%*** (b) Dependent variables Sick days Stress Job satisfaction Without With Without With Without With Independent Work work work work work work work variables pressure pressure pressure pressure pressure pressure pressure Task control 2.14** 2.02 2.01 2.19*** 2.12.22***.18*** Organization control.26*** 2.11** 2.11*.16***.03.09*.16*** Resourcecontrol 2.11* 2.02 2.01 2.05.11*.21***.18*** Work pressure.05.54*** 2.30*** R # 5%*** 2%*** 2%** 3%*** 31%*** 20%*** 29%*** ***p!.001, **p!.01, *p!.05. The numbers shown in the table are beta-coeycients, except for the values of R #. in the regression model of stress, the evects of task and organization control on stress disappeared, and work pressure played a signi cant role in predicting stress. As can be seen in table 4, this evect is not found in the prediction of job satisfaction and number of sick days. These ndings suggest that, in the Dutch sample, work pressure had a mediating in uence on stress evects only. The ndings for job satisfaction indicate that all three measures of job control and work pressure were signi cantly related to the outcome. High task, organization and resource control, and low work pressure were related to high job satisfaction. The analysis of the number of sick days showed that only organization control made a signi cant (negative) contribution. The results from the US and Dutch samples show that the various dimensions of job control had diverent evects on diverent types of strain. Resourcecontrol made almost no contribution towards predicting strain, whereas high task control was related to low strain, and organization control was related to high anxiety in the US sample and high stress in the Dutch sample, and to high job satisfaction in both samples. 5. Discussion 5.1. Summary of results The results of this study partially con rm the proposed model. First, the results show that it is important to distinguish between the various dimensions of job control. Although

44 P. Carayon and F. Zijlstra many experts have discussed the importance of diverent levels and facets of job control (Sauter et al., 1989),very few empirical studies have examined diverent dimensions of job control. In the present study, the diverent dimensions of job control had diverent roles in in uencing worker strain. Task control seems to have had both direct and indirect evects on strain. Task control had an indirect evect on anxiety in the US sample and on stress in the Dutch sample. This indirect evect is due to the evect of task control on work pressure which, in turn, is related to strain. High task control was related to low work pressure which, in turn, was related to low strain. Task control also had a direct evect on other measures of strain, i.e. job satisfaction in both samples and mood disturbances in the US sample. This result ts in with the `stressor reduction mechanism described by Frese (1989). Organization control appeared to have a positive contribution in predicting work pressure and some of the strain measures, i.e. anxiety in the US sample and stress in the Dutch sample. However, this positive contribution of organization control disappeared when work pressure was entered in the regression equation. This con rms the mediating role of work pressure in predicting stress and health outcomes. On the other hand, organization control was directly related to high job satisfaction in both samples. Resource control had no evect on strain, except on job satisfaction in the Dutch sample. Resource control was not related to work pressure in the US sample, but was negatively related to work pressure in the Dutch sample. As was expected, task control and organization control in uenced strain indirectly through work pressure. This result was true for anxiety in the US sample and for stress in the Dutch sample. Task control had a negative evect on work pressure and strain, whereas organization control had a positive evect on work pressure and strain. These results show that diverent facets and levels of job control can have very diverent evects on strain, and those results con rm that job control can have diverent types of evect on worker strain (Frese, 1989).The present results suggest that the diverent mechanisms between job control and worker strain depend on the type of job control, as well as the type of worker strain outcome. The results for task control t with the `stressor reduction mechanism proposed by Frese (1989),whereas organization control has diverent relationships with diverent types of worker strain. Task control can be a means for reducing work stressors, in particular work pressure, whereas organization control plays a diverent role. It can be concluded that in order to reduce anxiety in the US sample, and stress in the Dutch sample, it is important to decrease work pressure. Increasing task control can have a positive evect directly and} or indirectly via a reduction in work pressure. However, increasing organization control may actually increase strain because of a resulting increase in work pressure. The results obtained for job satisfaction were diverent from those found for anxiety in the US sample and for stress in the Dutch sample. High job satisfaction was directly in uenced by task control, organization control and work pressure. In addition, resource control was also a signi cant contributor to job satisfaction, but only in the Dutch sample. Providing more task control and organization control and reducing work pressure could have positive evects on job satisfaction. High level of mood disturbance was related to low task control and high organization control in the US sample. Work pressure was not a signi cant contributor to mood disturbance. In order to reduce mood disturbances, task control should be increased, but organization control should be decreased. In the Dutch sample, frequent sick days were explained by low job control, but not by work pressure. In a Dutch cross-sectional study, Houtman, Bongers, Smulders, and Kompier (1994) found that low control was associated with an increased probability of a recent period of sickness absence.

Job control, work pressure and strain 45 It is interesting to note that, even though two diverent measures of work pressure were used in the US and Dutch samples, similar patterns of results were found. The results do not seem to have depended on the measure of work pressure. In addition, the results were relatively similar in the two cultures. Therefore, the hypothesis that the model was valid in both countries seems to have been con rmed. The results appear to be `transnational, that is they are similar in the US and Dutch samples. 5.2. Implications for job redesign The results demonstrate diverent patterns of relationship between job control, work pressure and the diverent measures of strain. These results challenge the widely held belief that increasing job control is an evective means of reducing strain. The results suggest that the evectiveness of such intervention depends on the type and level of job control and on the strain outcome. In general, increasing task control could be an evective means of reducing strain in both US and Dutch samples because of its evect on work pressure and its indirect and} or direct evects on strain. However, increasing organization control could lead to negative outcomes because of increased work pressure. On the other hand, increasing organization control could have a positive evect on job satisfaction. These results suggest that a distinction has to be made not only between dimensions of control, but also between organizational levels (Westman, 1992). The type of job in which it is likely that high organization control will be found are the jobs that t the `active learning type of the Job Strain model (Karasek, 1979; Karasek, and Theorell, 1990).Typically, these are higherlevel jobs in organizations. In the case of these jobs, particular care should be taken in any evorts towards redesigning them because of the potential negative evects of high work pressure. In addition, there is growing evidence that personality characteristics also play a crucial role. Self-eYcacy has been found to be an important moderator in the relation between job demands and control, on the one hand, and strain, on the other hand (Schaubroeck, and Merritt, 1997). Similar results have been found for the need for control and coping style (De Rijk, Le Blanc, Schaufeli, and De Jonge, 1998).This suggests that particular groups of people may even suver more from increases in job control. These characteristics should also be taken into account in job redesign evorts and in future research. The most important nding is that increasing job control may not always be the solution to reducing strain. Attention should be paid to work pressure in particular. Employees must be aware of how much control is at their disposal if they are to cope evectively with job demands (Ganster, and Fusilier, 1989). 5.3. Measurement of job control The Job Control Questionnaire used in this study seemed to be a useful tool for diverentiating between dimensions of job control. The popularity of the Job Strain model has strongly in uenced the conceptualization, operationalization and measurement of job control. Karasek (1979) used a 9-item scale for decision latitude that encompassed the dimensions of skill discretion and decision authority. In most studies these two dimensions are merged and a single score of job decision latitude is computed (for a distinction between these two concepts, see for example Carayon, 1993). The questionnaire used in this study has a conceptually broader scope than most instruments on job control. It has great similarities with a scale proposed by Dwyer, and Ganster (1991),except that they included a scale for predictability of amount of work and results of decisions. The scales of job

46 P. Carayon and F. Zijlstra control used in this study appeared to have satisfactory internal consistencies (except the resource control scale), and are related to organizational and health outcomes. The results of the factor analyses, in both samples, con rm the results obtained by McLaney, and Hurrell (1988). There were, however, a few questionnaire items that were somewhat ambiguous : In uence on schedule, In uence on quality of work and In uence on rest breaks. In the US sample, the item In uenceon schedule had a high loading on the factor of organization control, instead of on task control as in the McLaney, and Hurrell (1988)study. In the Dutch sample, the item In uenceon schedule had high loadings on both factors of task control and organization control. In the Dutch sample, the item In uenceon the quality of one s work had a high loading on the factor of resource control, instead of task control as in the McLaney, and Hurrell (1988) study. In both US and Dutch samples, the item In uence on rest breaks had high loadings on the environment control factor instead of the task control as in the McLaney, and Hurrell study. It is important to examine these items further in future research. 5.4. Study limitations The authors note that only a very small amount of variance of the strain measures is explained. This may be due to the fact that the instruments used in this study were not speci cally designed to test the model. The study has other weaknesses, such as the crosssectional design, which limits the interpretabilit y of the results in terms of causal relationships. Longitudinal studies can be useful in examining the time dependency between independent variables and dependent variables, therefore improving con dence regarding the causality of these relationships (see, for example, the longitudinal study of job control by Carayon, 1993). Longitudinal intervention studies can also provide important information on the actual evects of changes in job control on stress and health (see, for example, the study by Jackson, 1983). Another weakness of the study was the reliance on a single data-collection method for measuring all study variables. The measures of job control and work pressure were based on questionnaire data, and therefore included an element of `subjectivity (for a discussion of objectivity } subjectivity in the measurement of psychosocial work factors, see, for example, Kasl (1987), Frese, and Zapf (1988) and Carayon, and Hoonakker (1999)). However, the results show statistically signi cant diverences between supervisory jobs and non-supervisory jobs on the three measures of job control. Therefore, the measures of job control are related to the `objective reality of job characteristics. Whereas multiple data-collection methods are preferred, in particular in the study of occupational stress, this study had features that compensated for the potential problems related to reliance on a single data-collection method, i.e. questionnaire survey. In the study, already-collected questionnaire data from two large samples in two diverent countries were used. The sample characteristics, i.e. size and country of origin, represented a strength of the study. 6. Conclusion The results of this study have practical and some theoretical implications. The authors think that the current models specifying the relationship between job control and strain should be adapted to take into account the diverent dimensions of job control and the diverent types of stress and health outcomes. The results of this study, which partially con rm the proposed model, need to be validated. A new study that will be speci cally designed to test these hypotheses will have to be carried out in order to con rm the ndings of this study.