PSYCHOMETRIC PROPERTIES OF THE THAI VERSION OF THE 22-ITEM AND 45-ITEM KARASEK JOB CONTENT QUESTIONNAIRE

Transcription

1 International Journal of Occupational Medicine and Environmental Health 2008;21(4): DOI /v PSYCHOMETRIC PROPERTIES OF THE THAI VERSION OF THE 22-ITEM AND 45-ITEM KARASEK JOB CONTENT QUESTIONNAIRE PITCHAYA PHAKTHONGSUK 1 and NUALTA APAKUPAKUL 2 1 Prince of Songkla University, Hatyai, Songkhla, Thailand Occupational Health Unit, Department of Community Medicine 2 Prince of Songkla University, Hatyai, Songkhla, Thailand Epidemiological Unit Abstract Objectives: To evaluate the validity and reliability of the Thai version of the job content questionnaire (TJCQ) both 22-item and 45-item versions. Materials and Methods: The study encompassed 10,450 employees in Songkhla province, Thailand. A 45-item job content questionnaire (JCQ) was translated and back translated complying with the JCQ usage policy. TJCQ covers seven scales as follows: psychological demand, decision latitude, supervisor support, coworker support, physical demand, job security and hazard at work. Internal consistency was examined using Cronbach s alpha coefficient. Factor validation was tested using both exploratory and confirmatory factor analysis. Results: Both 22-item and 45-item versions of TJCQ demonstrated acceptable internal consistency in nearly all scales except for psychological demand. The exploratory and confirmatory factor analysis result gave support to 5-factor model in 22-item version and 8-factor model in 45-item version of TJCQ. The psychological demand scale tended to split into two subscales. Other scales were consistent with the expected dimensions though some items extracted differently along the theoretical dimensions. Conclusions: Our study provided evidence for the reliability and validity of the TJCQ among Thai employees and supported the ongoing debate on multifaceted psychological demand and its complexity, making it feasible to extend the TJCQ beyond its original structure. Key words: Job content questionnaire (JCQ), Job strain, Factor analysis, Reliability, Psychometric properties INTRODUCTION The job content questionnaire (JCQ) has been the most popular self-administered instrument for measuring the psychosocial work environment since formulated by Karasek in 1979 [1,2]. The JCQ is based on the job demandcontrol-support (JDCS) model, which is composed of three major components that describe psychosocial work characteristics: psychological demand, job control or decision latitude and social support. Psychological demand refers to the measure of stress factors involved in accomplishing the work load, organizational constraints on task completion and job-related conflicting demand. Job control or decision latitude relates to the freedom permitted to the worker in how to perform tasks and how to meet the job demand and is assessed as a composite of skill discretion and decision authority. According to the model, a high level of skill gives the worker control over which specific skills to use to accomplish the task and make decisions that reduces possible adverse effects of psychological demand. Social support refers to overall levels of helpful social interaction available on the job both from coworkers and supervisors. For the past thirty years, researchers in various countries have used different versions of the JCQ to study the Received: June 9, Accepted: November 24, Address reprint requests to P. Phakthongsuk, Occupational Health Unit, Department of Community Medicine, Faculty of Medicine, Prince of Songkla University, Hatyai, Songkhla province 90110, Thailand ( pphakthongsuk@gmail.com). Unauthenticated IJOMEH 2008;21(4) 331

2 P. PHAKTHONGSUK AND N. APAKUPAKUL negative effects from high levels of job strain in agreement with Karasek s hypothesis on various health outcomes, including cardiovascular disease [3 7], musculoskeletal disease [8,9], psychiatric disorder [10 12] and reproductive health [13 16]. The JCQ has also been translated into over a dozen languages and nationally standardized by detailed occupations in several countries in Europe, America, and in Japan [17]. There have been increasing concerns over occupational stress which is becoming world phenomenon following the development of globalization, rising competition and economic plight. In Thailand, most studies of job stress used ad-hoc measures, which lack any clear theory or adequate validation. Researchers in Occupational Health are concerned to implement a standardized instrument to asses job stress based on strong theory and validation. Although the validity of the JCQ in various languages has been studied in the industrialized countries in North America, Europe and Japan and recently, among the newly industrialized countries of Eastern Asia, it remains uncertain whether it is applicable to populations like Thai who have a different socioeconomic structure, culture and more labor-intensive industries. The stage of validation seems crucial for the examination of the psychometric properties of an instrument and allows international comparisons between studies. With permission from the JCQ center, we translated 45 items of the JCQ into Thai measuring seven scales of psychological demand, decision latitude, physical demand, job security, supervisor support, coworker support and hazard at work. This version also included 22 core items assessing psychological demand, decision latitude, supervisor support and coworker support. The purpose of this present study was to measure the internal consistency and construct validity of the Thai version of both the 22 and 45 item JCQ among Thai heterogeneous workers. the Ethics Committee of the Faculty of Medicine at Prince of Songkla University. Songkhla province is the largest trade and commercial centre in Southern Thailand with an approximate population size of 1.3 million. The study population comprised about 100 subjects from each 3-digit occupational code according to international standard classification of occupations (ISCO-88) [18]. The sampling frame was based on the Provincial Registry on public administration organizations and the Industrial Registry. Random samples made during field visits included subjects in the informal sector outside the registry such as skilled agricultural and fishery workers (code ), craft and related trades workers (code ) and elementary occupations (code ). Of the 16,920 questionnaires distributed, 10,450 were collected of which 93.7% had been completed by the subjects and 6.3% were interviewed. The response rate was 61.6%. For practical purposes, invitation letters were sent to the selected work organizations and internally distributed within those organizations. After an in-house training course, six research assistants were divided into two teams. They met the subjects at their workplaces during working hours. Each subject was then distributed a questionnaire attached with a covering letter guaranteeing confidentiality and informed consent for the research. Participants filled in the questionnaire anonymously. The research assistants then collected the questionnaires either on the same day or two weeks later. For those who were unable to complete the questionnaire, assistance was given or an interview made by the research assistants. Onsite editing of the questionnaires was made and double-checked at the research center. On a daily basis the principle investigator coded the occupations according to the 3-digit code of ISCO-88. Double entry was conducted to achieve quality control of data. METHOD Design and Study Population A cross-sectional study was conducted in Songkhla province, in southern Thailand. The research was approved by Job Content Questionnaire The 45-item JCQ [19] was translated into Thai and then retranslated into English by a bilingual speaker who was unaware of the original English. This was then submitted 332 IJOMEH 2008;21(4) Unauthenticated

3 PSYCHOMETRIC PROPERTIES OF THE JOB CONTENT QUESTIONNAIRE ORIGINAL PAPERS to the JCQ center, University of Massachusetts, Lowell for the JCQ center approval. The Thai version JCQ (TJCQ) contained 45 items and consisted of a full set of questions for assessment of seven scales, such as decision latitude (9 items), psychological demand (9 items), job security (6 items), coworker support (4 items), supervisor support (4 items), physical exertion (5 items) and hazard at work (8 items). Each item has a response set of four-point Likert scales that range from 1 or strongly disagree to 4 or strongly agree. The 22 item TJCQ was a subset of 45-item TJCQ under study and consisted of 4 scales as follows: decision latitude (9 items), psychological demand (5 items), coworker support (4 items) and supervisor support (4 items). The questionnaires were self-administered including 45 item JCQ items, demographics, job title and its details, workshift, working hours and income. Answer choices were presented on a four-level Likert-type scale varying from strongly disagree to strongly agree. Statistical Analysis For each of the scales, their internal consistency was evaluated by Cronbach s alpha coefficients. The construct validity of TJCQ was first studied by exploratory factor analysis (EFA) using multiple squared correlations to estimate commonalities. The principle component method was used to extract the factor which was followed by a Varimax rotation method. To determine the number of meaningful factors to retain for rotation, the scree plot method was used. A factor loading of 0.3 or greater for a given factor was considered in the study. Confirmatory factor analysis (CFA) was then made to examine the factorial validity of TJCQ by testing the fit of four model as: a) 22-item, 4 factors, b) 22-item, 5 factors, c) 45-item, 7 factors, d) 45-item, 8 factors. In order to determine the overall fit of the hypothetical model to the observed data, the following fit indices were reported: chi-square statistic, goodness of fit index (GFI), adjusted goodness of fit index (AGFI), the root mean square error of approximation (RMSEA), non-normed fit index (NNFI), comparative fit index (CFI). The first four measures of fit were absolute fit indices to assess how well the hypothesized model covariance matrix estimates the sample covariance matrix. The chi-square statistics were a measure of overall fit of the model to the data. It tested the null hypothesis that the sample covariance matrix perfectly fitted the hypothesized model. A small chi-square corresponded to a good fit and a large chi-square to a bad fit. The root mean squared error of approximation (RMSEA) focused on the discrepancy between the sample covariance matrix and the hypothesized model covariance matrix but did also account for the degree of freedom. A value of less than 0.05 was indicative of a good fit; between 0.05 and less than 0.08 a reasonable fit and greater than 0.1 a poor fit. The goodness of fit index (GFI) was indicative of the relevant amount of the hypothesized model s covariances and demonstrated how close the model was to the sample covariance matrix. The adjusted goodness of fit (AGFI) was the GFI adjusted for the degree of freedom in the model. Values of GFI and AGFI should range between 0 and 1 and any values exceeding 0.9 reflected acceptable fits. The last two measures of fit were the comparative fit indices, which showed how much better the model fit compared to the basic model. The non-normal fit index (NNFI) and comparative fit index (CFI) measured how much better the hypothesized model fitted a null model that did not specify any relationships between the variables. The NNFI and CFI ranged from 0 to 1, with values exceeding 0.9 indicating a good fit. RESULT Baseline Characteristics Table 1 summarizes the characteristics of the study population. The subjects comprised 5,231 males and 5,096 females. Nearly all of the respondents belonged to young age group from 19 to 45 years old. About 60% had education below or equal to college diploma level. The large majority had an income of less than 500 dollars per month and had regular working hours of more than 40 hours Unauthenticated IJOMEH 2008;21(4) 333

4 P. PHAKTHONGSUK AND N. APAKUPAKUL Table 1. Characteristics of the Study Population (N = 10415) Population N % Sex Male Female Missing Age (years) > Missing Education (years) > Unclassified Missing Income (dollars per month) > Missing Working hours per week > Missing Major occupational group 1. Legislators, senior officials and managers Professionals Technicians and associate professionals Clerks Service workers and shop and market sales workers 6. Skilled agricultural and fishery workers Craft and related trades workers Plant and machine operators and assemblers Elementary occupations Armed force per week. The occupational groups according to ISCO system are also shown. Reliability The Cronbach s alpha coefficients of the 22 and 45-item TJCQ are presented in Table 2. For the 22-item TJCQ, the internal consistency coefficients were fairly high for decision latitude, supervisor support and coworker support but quite low for psychological demand which consisted of five items. The values of all item-total correlation of the supervisor and coworker support scales were greater than 0.3, indicating that each of the items had an acceptable correlation with other items comprising the overall scale score. However, for the psychological demand scale, the three items of not enough time, excessive work and conflicting demand had an item-total correlation value of less than 0.2 indicating a low correlation of these items for this scale. Dropping these three items changed the alpha level of 5 question version of psychological demand substantially to 0.52; and in 9 question version, it was moderately changed to In addition, the three items of little decision freedom, repetitive work, and variety connected with decision latitude scale also yielded a low item-total correlation. Regarding the 45-item TJCQ, fairly high values for Cronbach s alpha coefficients were found in decision latitude, supervisor support, coworker support, physical demand and hazard at work, indicating acceptable levels of internal consistency for each of the scales. However, the Cronbach s alpha levels were only moderate for the psychological Table 2. Internal Consistency of the 22-item and 45-item TJCQ Scale Cronbach s Alpha Coefficients 22 Item 45 Item Decision latitude Psychological demand # Supervisor support Coworker support Physical exertion 0.76 Job security 0.60 Hazard at work 0.88 # 5 items in 22-items TJCQ and 9-items in 45 item TJCQ. 334 IJOMEH 2008;21(4) Unauthenticated

5 PSYCHOMETRIC PROPERTIES OF THE JOB CONTENT QUESTIONNAIRE ORIGINAL PAPERS demand and job security scales. Low values of item-total correlation were as follows: two items of little decision freedom and repetitive in decision latitude scale; three items of not enough time, excessive work and conflicting demand in psychological demand scale; and one item of skill value in the next five years in job security scale. As might be expected, low correlation between these items across these scales was also found. EXPLORATORY FACTOR ANALYSIS 22-item TJCQ Table 3 presents the EFA result using principle component analysis and Varimax rotation of 22-item TJCQ for the four and five factor models consecutively. Initially, the sampling adequacy measure of Kaiser s criterion was first examined to determine how appropriate the matrix was for a factor analysis. The value of 0.83 suggested sufficient correlation for conducting the analysis. A scree test suggested five meaningful factors with eigenvalues of more than 1, explaining 51.1% of the total variance, while the 4-factor model according to JDCS model explained 45.3% of the total variance. Factor 1 grouped items connected with decision latitude except little decision freedom, repetitive, and variety. These three items unexpectedly revealed adequate factor loading scores on Factor 4, which was most likely connected with psychological demand. Of interest was that the factor extraction pattern, including its items distribution Table 3. Exploratory Factor Analysis with Varimax Rotation of 22-item TJCQ (4 factors and 5 factors solution) Dimensions Items Factor Factor F1 F2 F3 F4 F1 F2 F3 F4 F Decision latitude Learn new things Creativity High skill Develop own abilities Allow own decisions A lot of say Little decision freedom Repetitive work Variety Supervisor support Concerned Pay attention Good organizer Helpful Coworker support Competent Psychological demand Interested Friendly Helpful Work fast Work hard Excessive work Not enough time Conflicting demands.57 Unauthenticated IJOMEH 2008;21(4) 335

6 P. PHAKTHONGSUK AND N. APAKUPAKUL of Factor 4, showed similar solutions in both the 22 and 45-item TJCQs. In addition, the item of variety similarly shared loadings on the two factors relevant to decision latitude and the psychological demand in both TJCQs. In the 5-factor model, Factor 2 and 3 were unambiguously connected with supervisor and coworker support, respectively. A solution with 4-factor model also revealed a clear meaning for Factor 3 regarding theoretical model, but not for Factor 2. As seen, Factor 2 comprised not only four items connected to supervisor support but also two other items of excessive work and not enough time from the psychological demand scale with ambiguously low loadings. Considering the 5-factor model, Factor 5 contained three items of excessive work, not enough time and conflicting demand from the psychological demand with significant factor loadings of more than 0.5 and modest variance of 6.2%. It seemed that the psychological demand scale in TJCQ might appropriately separate into two subscales. 45-item TJCQ The result of factorial validity of the 45 items TJCQ are shown in Table 4 and 5. The scree test based on the eigenvalues favored an eight factor solution, explaining 46.3% of total variance as shown in Table 4. Factor 1 was clearly associated with hazard at work and all items theoretically connected with this scale had high loadings of more than or equal to Factor 2 comprised of items mainly connected with decision latitude and nearly all items loadings on this scale ranged from 0.43 to 0.69 except for little decisions freedom, repetitive work and variety. Interestingly, these three items were loaded in Factor 3 relevant to psychological demand. Furthermore, EFA revealed that three items from other scales according to the theoretical model were unexpectedly extracted in this factor as follows: high concentration from the psychological demand scale; career possibilities and skill valuable in next 5 years from the job security scale. Factor 3 grouped items relevant to psychological demand with factor loading of more than 0.4 except for waiting on others (0.36). All the items belonging to this scale loaded on this factor except for excessive work, conflicting demand, not enough time and intense concentration. Instead, the item intense concentration loaded on the factor relevant to decision latitude, as previously mentioned. Of note was that the other three items belonging to the psychological demand scale in JDCS model were separately extracted into Factor 8. One interpretation could be that in the Thai context these three items point to a conflicting mental load beyond stress-related routine work. Factor 4 and 5 were unequivocally connected to the theoretical scales of supervisor support and coworker support, respectively with high factor loadings. Factor 6 grouped three items most likely connected with physical demand except for physical effort and rapid and continuous. These two items loaded in Factor 3 relevant to psychological demand instead. The first two items of this factor awkward body position and awkward head and arm position loaded on this factor were clearly relevant to the static physical demand and showed satisfactory loadings of more than 0.8. However, the third item yielded comparatively low loadings ( lift heavy loads = 0.42). Factor 7 grouped four items clearly associated with job security except for skill valuable in next 5 years and career possibilities. These two items were loaded on the factor associated with decision latitude as previously mentioned. That the results of an examination of the factor structure fit to the seven dimensions of the original JCDS model is also demonstrated in Table 5. Compared to the eight factor solution, five out of seven factors resembled the extraction patterns including their item distribution except for some degree of difference across Factor 3 connected with psychological demand and Factor 5 connected with physical demand as follows: Firstly, all 5 items belonging to the physical demand scale as used by Karasek loaded on Factor 5, which clearly coincided with physical demand. In addition, one item of little decision freedom from the decision latitude scale also loaded on this factor. Secondly, regarding the three items which separated into Factor 8 in the eight-factor model, two items excessive work and conflicting demand loaded on Factor 5 which is connected to physical demand while one item not enough time 336 IJOMEH 2008;21(4) Unauthenticated

7 PSYCHOMETRIC PROPERTIES OF THE JOB CONTENT QUESTIONNAIRE ORIGINAL PAPERS Table 4. Exploratory Factor Analysis with Varimax Rotation of 45-item TJCQ (8 factors solution) Dimensions Items Factor F1 F2 F3 F4 F5 F6 F7 F Hazard at work Chemicals.73 Air polluted.72 Things stored dangerously.73 Badly maintained area.64 Fire/burns/shocks.74 Catching diseases.71 Dangerous tools/machinery.80 Dangerous work method.76 Decision latitude Learn new things.60 Creativity.69 High skill.62 Develop own abilities.68 Allow own decisions.59 A lot of say.60 Little decision freedom.45 Repetitive work.49 Variety Psychological demand Work fast.57 Work hard.67 Excessive work.54 Conflicting demands.48 Not enough time.55 Intense concentration.53 Waiting on others.36 Work interrupted.46 Hectic work.66 Supervisor support Supervisor concerned.82 Supervisor pay attention.73 Supervisor good organizer.79 Supervisor helpful.76 Coworker support Coworkers competent.64 Coworkers interested in me.56 Coworkers friendly.75 Coworkers helpful.76 Physical demand Physical effort Rapid and continuous.50 Awkward body.83 Awkward head and arm.86 Lift heavy loads Job security Career possibilities.43 Skill value in next 5 yrs.47 Job security.59 Steady job.61 Recent layoff.72 Future layoff.70 Unauthenticated IJOMEH 2008;21(4) 337

8 P. PHAKTHONGSUK AND N. APAKUPAKUL Table 5. Exploratory Factor Analysis with Varimax Rotation of 45-item TJCQ (7 factors solution) Factor Dimensions Items F1 F2 F3 F4 F5 F6 F Hazard at work Chemicals.73 Air polluted.72 Things stored dangerously.73 Badly maintained area.64 Fire/burns/shocks.73 Catching diseases.71 Dangerous tools/machinery.80 Dangerous work method.76 Decision latitude Learn new things.57 Creativity.67 High skill.59 Develop own abilities.71 Allow own decisions.59 A lot of say.63 Little decision freedom.37 Repetitive work.41 Variety Psychological demand Work fast.48 Work hard.63 Excessive work.34 Conflicting demands.53 Not enough time.35 Intense concentration Waiting on others.41 Work interrupted.57 Hectic work.71 Supervisor support Supervisor concerned.82 Supervisor pay attention.73 Supervisor good organizer.79 Supervisor helpful.76 Physical demand Physical effort Rapid and continuous Awkward body Awkward head and arm.64 Lift heavy loads Coworker support Coworkers competent.63 Coworkers interested in me.52 Coworkers friendly.73 Coworkers helpful.76 Job security Career possibilities.48 Skill value in next 5 yrs.49 Job security Steady job.61 Recent layoff.71 Future layoff IJOMEH 2008;21(4) Unauthenticated

9 PSYCHOMETRIC PROPERTIES OF THE JOB CONTENT QUESTIONNAIRE ORIGINAL PAPERS Table 6. Confirmatory Factor Analysis of 22-item and 45-item TJCQ; Two Models for Each Version Model χ 2 df GFI AGFI RMSEA NNFI CFI 22-item TJCQ Model 1 four-factor Hypothesized Final Model 2 five factor Hypothesized Final Δχ 2, df 5 = ; p-value < item TJCQ Model 1 seven-factor Hypothesized Final Model 2 eight-factor Hypothesized Final Δχ 2, df 14 = ; p-value < 0.05 loaded on Factor 3 connected to psychological demand. Of interest to note was that their factor loadings were obviously low. Confirmatory Factor Analysis Both the hypothesized and final versions of the four models by CFA are shown in Table 6. Each model specified that each of the proposed items were connected to only one common factor. Some correlated error items were specified to improve the fit of the models, based on the modification indices provided. Regarding the 22-item TJCQ; high fit indices indicated a reasonably good fits of both the 4-factor and 5-factor models. The comparison of both model, based on the discrepancy in χ 2 fit values, indicated that the 5-factor model provided a significantly better fit. Likewise, the 8-factor model also showed a better fit compared to the 7-factor model. However, the absolute fit measures of the 45-item TJCQ showed a modest fit in both 7-factor and 8-factor models. As seen, the value of both the GFI and AGFI is smaller than 0.9, which means that the models were moderately fitted, though other fit measure values were acceptable. DISCUSSION The purpose of the study was to evaluate the Thai version of JCQ for both the 22-item and 45-item versions. This study, based on a large sample and variety of occupations showed that the psychometric properties of TJCQ of decision latitude, supervisor support, coworker support, physical exertion and hazard at work were satisfactory for internal consistency except for the psychological demand and job security scale. The internal consistency of psychological demand in our study was the lowest compared to other studies with Cronbach s alpha of 0.23 for five questions of 22-item TJCQ and 0.54 for nine questions of 45-item TJCQ. Some western studies have reported acceptable Cronbach s alpha values of more than 0.7 for the psychological demand scale [20 24] while some have reported borderline alpha values around 0.6 [17,25]. The psychological demand scale, with five questions, was borderline in all Asian studies with alpha values ranging from [26 29]. To the authors of some Asian studies, it was suggested that five questions of this scale might not be sufficient for various psychological demands encountered at work and need to be modified in order to improve the scale reliability [26,27]. Karasek (1998) has also Unauthenticated IJOMEH 2008;21(4) 339

10 P. PHAKTHONGSUK AND N. APAKUPAKUL recommended that the correlation of the nine questions of the psychological demand scale with skill discretion was higher than that of five questions and might be better off in assessing cognitive demand [17]. However, after dropping the three items with low itemtotal correlation of enough time, excessive work and conflicting demand, the alpha coefficients increased substantially. This indicated that these three items were poorly correlated with other items in the psychological demand scale and supported further findings under study that these three items separated into new factor with adequate factor loadings around 0.5. There were also some studies from the United States, Canada and Netherlands where psychological demand was split into two factors as our study, but the items and their Cronbach s alpha were not presented in detail [17]. The low item-total correlation of the item conflicting demand was usually criticized in many previous studies [17,24,26,27,30]. The Chinese study among industrial workers [27] also reported a low value of the item not enough time but the item excessive work was seldom mentioned in previous studies. Considering the reliability estimates of job security scale, the value under study was about the same as the values reported in previous studies [20,22] and better than the coefficient of the Korean study [28] and the Netherlands s sample in an internationally comparative study [17]. The reduction of the initial items into the meaningful scales by EFA showed that the psychometric properties were satisfactory for the 5-factor model in the 22-item TJCQ and the 8-factor model in the 45-item TJCQ, respectively. The factor extraction pattern of the 22-item version including its item distribution was similar to that of the 45-item version. CFA suggested that the best model consisted of five latent factors for the 22-item TJCQ and eight latent factors for 45-item TJCQ, though the 8-factor model may be unable to provide an acceptable value in all goodness of fit indices. In both TJCQs, nearly all the items of skill discretion and decision authority loaded on the same factor relevant to decision latitude, except for little decision freedom, repetitive work and variety. According to Karasek and Theorell [31], decision latitude consisted of two distinct constructs skill discretion and decision authority. These two dimensions formed complementary relationships because job skill provided people with a sense of authority or control over their work activities. Skill discretion and decision authority should then be regarded as separate scales of job control. Our study, however, has not indicated a clear distinction between these two subscales. That decision latitude appears as a separate factor in our occupationally heterogeneous sample is in agreement with previous studies [21,22,25,32,33]. In the Belstress study by Pelfrene et al. [21], he has reported a high correlation between skill discretion and decision latitude without a clear distinction between the two scales in heterogeneous samples, similar to the results of Sanne et al. [33]. Actually, it has also been demonstrated by Sanne that decision latitude split into skill discretion and decision authority when the samples were divided into two homogeneous occupational groups and concluded that the considerable variation in latitude scores between different occupations may be restricted in specific occupational groups. However, the clear distinction between these two subscales was not consistent in most Asian validation studies which had been conducted among occupationally homogeneous samples. Three studies reported decision latitude with two subscales of skill discretion and decision authority [27,30,34] while another two reported only one scale [26,28]. In the 45-item TJCQ, one item belonging to the psychological demand scale, high concentration, loaded on decision latitude factor. This result was in line with that of Brisson [35] who found a shared loading of this item on decision latitude and explained that this item seemed to relate to the concept of cognitive/mental load or demandlearning. This leads to the conjecture that cognitive job demand might lead to problem solving and thus build up skill discretion and influence over decisions at work. Moreover, the other two items, career possibilities and skill valuable in next 5 years, from the job security scale loaded on the decision latitude factor scale in our study. The broad area of job security measured two points of job security/career prospect and layoff/work instability. These two items unloading on job security scale were in agreement with the Dutch study [22]. Nevertheless, further 340 IJOMEH 2008;21(4) Unauthenticated

11 PSYCHOMETRIC PROPERTIES OF THE JOB CONTENT QUESTIONNAIRE ORIGINAL PAPERS comparison with any studies could not be made since most validation studies rarely included a job security scale. Almost all the Asian studies did not include this scale in their studies except one Chinese study among petrochemical workers but only three items of this scale were recruited [34]. The Thai economy is heavily dependent on exports and on the transition to a knowledge-based economy. Meanwhile, the human resource development is inadequate toward this new economy. Key obstacles include weak educational development, inadequate researches and insufficiently skillful manpower in science and technology and administration [36]. In such situation, jobs involving high intellectual demands and significant decision making abilities could result in career promotion and job security of the employees. EFA showed that, in the 22-item and 45-item TJCQs, nearly all the items belonging to the psychological demand scale loaded on two factors. The separated new factor in both TJCQs consisted of the identical three items of not enough time, excessive work and conflicting demand. The grouping of three similar items was also observed in a Japanese study among employees in telecom and electric power companies [30] and was criticized that the measure of job demand included several different aspects in one scale. In another study which included five questions of the psychological demand scale [32], the EFA also extracted these three items uniformly on the psychological demand factor while the other two items of work fast and work hard loaded on physical demand. Our CFA finding that psychological demand was appropriately separated into two distinct subscales was in agreement with Sale s study among health care workers [23] which provided a better goodness of fit for the two-factor model when compared against the original one-factor psychological demand model. The two factors connected with the psychological demand scale in our study were associated with time constraints and conflicting demand at work and may be redundant with each other. In consideration of the Thai economic development, we speculate that the new factor might involve a higher level of demand besides routine work requirements. Thailand s rapid transformation from an agriculturally dominant structure to an industrially dominant one became apparent in late 1980s. The rapid economic growth in the 1990s depended on natural-based and labor-intensive industries, in particular the preserved and canned food, textile, gemstone and jewelry sector. Towards the end of the 1990s and into the 2000s, new industries such as automobile, machinery, electrical goods, petroleum product and service sectors expanded. By the early and mid 2000s the Thai labor market, however, has still not shown any significant improvement in the quality of labor, which has limited the country s move towards higher economic activities [37]. As previously mentioned, labor-intensive manufacturing and the lack of highly skilled labor coexist with long working hours and compulsory over-time work, which is considered normal in Thailand. A recent survey on working hours in over 50 countries has demonstrated that 46.7% of the Thai labor force has worked more than 48 hours per week. and was in the third rank among the top three countries with long working hours in the world besides Peru (50.9%) and the Republic of Korea (49.5%) [38]. It is possible that this new factor might actually describe the emergent and unavoidable qualitative and quantitative work demand created by organizational constraints set by a competitive world market, loss of comparative advantage to new labor intensive countries in Asia and a struggle towards the new era of knowledge-based economies. Of interest was that the three items of little decision freedom repetitive work and variety unexpectedly loaded on the psychological demand factor. Conceptually, these items seemed to capture psychological demand in Thai subjects instead of decision latitude as theoretically expected. Previous studies have reported that items belonging to decision latitude and psychological demand shared loadings or even mixed [24,26,27,30]; though items reported are different across the studies. The operationali- zation of items in decision latitude has been criticized for reflecting job autonomy/control as well as the complexity or difficulty of the task. Thus, too much skill discretion/ decision authority can also be a source of stress [39 41]. According to the international study by Karasek [17], the most troublesome item consistently found was repetitive Unauthenticated IJOMEH 2008;21(4) 341

12 P. PHAKTHONGSUK AND N. APAKUPAKUL work. He explained that it was the result of the non-normal distribution, as repetitive work is much more common among the lowest skilled workers. Likewise, in all Asian studies, this item had a low loading and more likely loaded on psychological demand [26 28,30]. Kawakami [30], in a study of Japanese female workers found that little decision freedom loaded on psychological demand while in the Korean study by Eum [28], the item variety, like in our study, loaded on psychological demand factor. The EFA also indicated that the items belonging to physical demand scale loaded on psychological demand. As seen, the two items of physical effort and rapid and continuous activity which originally was intended to capture general physical demand and dynamic workload consecutively loaded on the psychological demand scale under study. We postulated that these two items might be perceived as assessment of overall work exhaustion or psychophysiological costs at work in Thai employees. The interpretative ambiguities between physical and psychological demand could occur, since high physical demand could arouse high levels of autonomous nervous system and certainly contribute psychologically. Our findings support the debate among temporary researches [32,40,42,43] on multidimensional psychological character of the demand scale and its complexity, which open way to extending it beyond its original domain. The other two items body in awkward position and hand/ arm in awkward position that originally measured statically work activity loaded on the factor connected with physical demand scale with the item lift heavy load. In this study with its heterogeneous set of occupations, the result as such may be explained by its skewed distribution, as statically and dynamically physical exertion was much more common among blue collar workers under study. EFA clearly showed that supervisor support, coworker support and hazard at work were loaded as theoretical construct. Most previous studies from North America, Europe and Asia [21,24,26,27,29,34] reported a factorial discrepancy between supervisor support and coworker support, though some did report social support as one scale [27,28]. Unlike our study, very few studies have included hazard conditions and toxic exposure scales. A Dutch study of nine different organizations has, however, reported that items of the hazardous conditions and toxic exposure did also load onto the same factor as our finding [22]. CONCLUSION The psychological environment at work has been receiving increasing attention as there is growing evidence of its relationship to various health outcomes. A large number of occupational stress studies have been based on the JCQ, which has been now widely used in several countries. We performed this cross-sectional study to thoroughly examine both the core and full version of the TJCQs. Our result of the scale reliability and exploratory and confirmatory factorial analysis were satisfactory for the 5-factor model in the 22-item TJC and the 8-factor model in the 45-item TJCQ, respectively. This study also suggests that psychological demand is better treated as two distinct subscales and some items might distribute differently in Thai population. Generally, the factor solution in TJCQ supported the priori structure of JDCS model, although the psychological demand scale split into two factors and some items extracted in each factor lined up somewhat differently from those in the original scale. Further research needs to be conducted to confirm and expand our findings in a national sample and specific occupational groups including addressing the test-retest reliability of these scales. ACKNOWLEDGEMENTS The authors would like to thank the Thai Research Fund (TRF) for financial support on this research (Grant No. RDG3/10/2544). References 1. Hurrel JJ, Nelson DL, Simmons BL. Measuring job stressors and strains: where we have been, where we are, and where we need to go. J Occup Health Psychol 1998;3: Benavides FG, Benach J, Muntaner C. Psychosocial risk factors at the workplace: is there enough evidence to establish reference values? J Epidemiol Community Health 2002;56: IJOMEH 2008;21(4) Unauthenticated

13 PSYCHOMETRIC PROPERTIES OF THE JOB CONTENT QUESTIONNAIRE ORIGINAL PAPERS 3. Alfredsson L, Karasek R, Theorell T. Myocardial infarction risk and psychosocial work environment: an analysis of the male Swedish working force. Soc Sci Med 1982;16: Karasek R, Theorell T, Schwartz JE, Schnall PL, Pieper CF, Michela JL. Job characteristics in relation to the prevalence of myocardial infarction in the US Health Examination Survey (HES) and the Health and Nutrition Examination Survey (HANES). Am J Public Health 1988;78: Theorell T, Tsutsumi A, Hallquist J, Reuterwall C, Hogstedt C, Fredlund P, et al. Decision latitude, job strain, and myocardial infarction: a study of working men in Stockholm. Am J Public Health 1998;88: Kuper H, Marmot M. Job strain, job demands, decision latitude, and risk of coronary heart disease within the Whitehall II study. J Epidemiol Community Health 2003;57: Wamala SP, Mittleman MA, Horsten M, Schenck-Gustafsson K, Orth-Gomér K. Job stress and the occupational gradient in coronary heart disease risk in women. The Stockholm female coronary risk study. Soc Sci Med 2000;51: Cole D, Ibrahim S, Shannon H, Scott F, Eyles J. Work correlates of back problems and activity restriction due to musculoskeletal disorders in the Canadian national population health survey (NPHS) data. Occup Environ Med 2001;58: Hoogendoorn WE, Bongers PM, de Vet HC, Houtman IL, Ariëns GA, van Mechelen W et al. Psychosocial work characteristics and psychological strain in relation to low-back pain. Scand J Work Environ Health 2001;27: Kawakami N, Araki S, Kawashima M, Masumoto T, Hayashi T. Effects of work-related stress reduction on depressive symptoms among Japanese blue-collar workers. Scand J Work Environ Health 1997;23: Sanne B, Mykletun A, Dahl AA, Moen BE, Tell GS. Testing the Job Demand-Control-Support model with anxiety and depression as outcomes: the Hordaland Health Study. Occup Med 2005;55: Ylipaavalniemi J, Kivimäki M, Elovainio M, Virtanen M, Keltikangas-Järvinen L, Vahtera J. Psychosocial work characteristics and incidence of newly diagnosed depression: a prospective cohort study of three different models. Soc Sci Med 2005;61: Homer CJ, James SA, Siegel E. Work-related psychosocial stress and risk of preterm, low birthweight delivery. Am J Public Health 1990;80: Brandt LP, Nielsen CV. Job stress and adverse outcome of pregnancy: a causal link or recall bias? Am J Epidemiol 1992; 135: Klonoff-Cohen HS, Cross JL, Pieper CF. Job stress and preeclampsia. Epidemiology 1996;7: Landsbergis PA, Hatch MC. Psychosocial work stress and pregnancy-induced hypertension. Epidemiology 1996;7: Karasek R, Brisson C, Kawakami N, Houtman I, Bongers P, Amick B. The Job Content Questionnaire (JCQ): an instrument for internationally comparative assessments of psychosocial job characteristics. J Occup Health Psychol 1998;3: ILO. International Standard Classification of Occupations: ISCO-88. Geneva: International labour office; Karasek R. Job content questionnaire and user s guide. Lowell: University of Massachusetts Lowell, Department of Work Environment; Henriksen TB, Hedegaard M, Secher NJ. The relation between psychosocial job strain, and preterm delivery and low birthweight for gestational age. Int J Epidemiol 1994;23: Pelfrene E, Vlerick P, Mak R, de Smet P, Kornitzer M, de Backer G. Scale reliability and validity of the Karasek «Job-Demand-Control-Support model» in the Belstress study. Work Stress 2001;15: Storms G, Casaer S, de Wit R, van den Bberg O, Moens G. A psychometric evaluation of a Dutch version of the job content questionnaire and of a short direct questioning procedure. Work Stress 2001;15: Sale JE, Kerr MS. The psychometric properties of Karasek s demand and control scales within a single sector: data from a large teaching hospital. Int Arch Occup Environ Health 2002;75: Niedhammer I. Psychometric properties of the French version of the Karasek Job Content Questionnaire: a study of the scales of decision latitude, psychological demands, social support, and physical demands in the GAZEL cohort. Int Arch Occup Environ Health 2002;75: Unauthenticated IJOMEH 2008;21(4) 343

14 P. PHAKTHONGSUK AND N. APAKUPAKUL 25. Mausner-Dorsch H, Eaton WW. Psychosocial work environment and depression: epidemiologic assessment of the demand-control model. Am J Public Health 2000;90: Cheng Y, Luh WM, Guo YL. Reliability and validity of the Chinese version of the Job Content Questionnaire in Taiwanese workers. Int J Behav Med 2003;10: Li J, Yang W, Liu P, Xu Z, Cho SI. Psychometric evaluation of the Chinese (Mainland) version of job content questionnaire: A study in university hospitals. Ind Health 2004;42: Eum KD, Li J, Jhun HJ, Park JT, Tak SW, Karasek R, et al. Psychometric properties of the Korean version of the job content questionnaire: data from health care workers. Int Arch Occup Environ Health 2007;80: Hadi AA, Naing NN, Daud A, Nordin R. Reliability and construct validity of the Malay version of the Job Content Questionnaire (JCQ) among secondary school teachers in Kota Bharu, Kelantan, Malaysia. Southeast Asian J Trop Med Public Health 2006;37: Kawakami N, Kobayashi F, Araki S, Haratani T, Furui H. Assessment of job stress dimensions based on the job demands-control model of employees of telecommunication and electric power companies in Japan: Reliability and validity of the Japanese version of the Job Content Questionnaire. Int J Behav Med 1995;2: Karasek R, Theorell T. Healthy work: stress, productivity, and the reconstruction of working life. New York: Basic Books; Muntaner C, Schoenbach C. Psychosocial work environment and health in U.S. metropolitan areas: a test of the demandcontrol and demand-control-support models. Int J Health Serv 1994;24: Sanne B, Torp S, Mykletun A, Dahl AA. The Swedish Demand-Control-Support Questionnaire (DCSQ): factor structure, item analyses, and internal consistency in a large population. Scand J Public Health 2005;33: Li W, Zhang JQ, Sun J, Tan PF, Wang S. Reliability and validity of Job Content Questionnaire in Chinese petrochemical employees. Psychol Rep 2007;100: Brisson C, Blanchette C, Guimont C, Dion G, Moisan J, Vezina M. Reliability and validity of the French version of the 18-item Karasek job content questionnaire. Work Stress 1998;12: Makishima M, Suksiriserakul S. Human resources development toward a knowledge-based economy: The case of Thailand. [cited 2008 November 25]. Available from: ide.go.jp/english/publish/download/asedp/066.html. 37. Kaosa-ard M. Economic development and institutional failures in Thailand. Thailand Dev Res Inst Q Rev 1998;13: Lee S, McCann D, Messenger JC. Working time around the world: Trends in working hours, laws, and policies in a global comparative perspective. Geneva: Routledge; Kristensen TS. The demand-control-support model: methodological challenges for future research. Stress Med 1995;11: De Jonge J, Kompier MA. A critical examination of the Demand-Control-Support model from a work psychological perspective. Int J Stress Manage 1997;4: Smith CS, Tisak J, Hahn SE, Schmieder R. The measurement of job control. J Organ Behav 1997;18: Soderfeldt B, Soderfeldt M, Muntaner C, O Campo P, Warg L, Ohlson C. Psychosocial work environment in human service organizations: A conceptual analysis and development of the demand-control model. Soc Sci Med 1996;42: De Jonge J, Mulder MJ, Nijhuis FJ. The incorporation of different demand concepts in the job demand-control model: effects on health care professionals. Soc Sci Med 1999;48: IJOMEH 2008;21(4) Unauthenticated