Presenteeism among Japanese IT Employees : Personality , Temperament and Character , Job Strain and Workplace Support , and Mental Disturbance

To study correlates of presenteeism among Japanese IT employees, a cross-sectional study in 440 employees of four IT companies was studied with a net-based questionnaire survey. A structural regression analysis revealed that presenteeism (measured by the Work Limitation Questionnaire) was directly associated with common mental symptoms (measured by the Clinical Outcomes in Routine Evaluation-Outcome Measure) and job strain (measured by the Job Strain Questionnaire), common mental symptoms mediated the effects of job strain and poor workplace social support as well as high Harm Avoidance and low Self-directedness and Cooperativeness (measured by the Temperament and Character Inventory-Revised) upon presenteeism. IT company workers presenteeism may be a result of complicated effects of common mental symptoms, job strain and poor workplace social support, and personality traits.

Considering such research on the association between high Harm Avoidance and low Self-directedness and a different variety of common mental disorders and perceptive nature of job strain and social support, we hypothesized that such personality traits would be associated with job strain and workplace social support.Hence the effects of high Harm Avoidance and low Self-directedness on mental disorders may be mediated by perceived job strain and low social support.We were also interested in whether such personality traits would directly predict presenteeism.
Our research hypotheses in this study are thus (Figure 1); 1) Presenteeism is predicted by job strain as well as lack of workplace social support; 2) This effect is mediated by common mental symptoms; 3) High Harm Avoidance and low Self-directedness and Cooperativeness predict presenteeism and common mental symptoms directly and; 4) The effects High Harm Avoidance and low Self-directedness and Cooperativeness on presenteeism and common mental symptoms are mediated by job strain as well as lack of workplace social support.

Participants
We solicited the participation in the present web-based questionnaire survey a total of 2360 employees of four information technology companies located all in Tokyo.We chose four companies based on the fact that they had different managerial and occupational backgrounds.handling the workload).Each scale implies the percent of work time in the previous two weeks that an employee's physical and or emotional health problems interfered with his or her ability to perform each group of job demand.Item scores within each scale is averaged and transformed to a 0 (no limitation) to 100 (limited 100% of the working time) scale.This instrument's good validity and reliability was reported by its original authors (Lerner, Amick, Lee, Rooney, Rogers, Chang, & Berndt, 2003;Lerner, Reed, Massarotti, Wester, & Burke, 2002).The WLQ was translated into Japanese and it showed god validity (Ida, Nakagawa, Miura, Ishikawa, & Yakura, 2012).In the present participants, internal reliabilities expressed by Cronbach's α were .83,.89,.91,and .87 for TMM, PHD, MID, and OPD, respectively.

Mental Disturbance
Different symptoms of common mental disorders were measured by the Clinical Outcomes in Routine Evaluation-Outcome Measure (CORE-OM: Evans, Mellor-Clark, Margison, Barkham, Audin, Connell et al., 2000).This is a self-report on common psychiatric symptoms and psychosocial functioning.The CORE-OM is with 34 items.It covers four areas including Symptomatic Problems (i.e., anxiety, depression, physical symptoms, and trauma), Risk (i.e., risks to others and self), Functioning (i.e., close, general, and social functioning), and Psychological Well-being.The reliability and validity of the CORE-OM was confirmed by its original authors (Evans, Connell, Barkham, Margison, McGrath, Mellor-Clark et al., 2002).The CORE-OM was translated into Japanese and was shown excellent in its psychometric properties (Uji, Sakamoto, Adachi, & Kitamura, 2012).In the present participants, internal reliabilities expressed by Cronbach's α were .90,.84,.87,and .69for Symptomatic Problems, Risk, Functioning, and Psychological Well-being, respectively.The template is used to format your paper and style the text.All margins, column widths, line spaces, and text fonts are prescribed; please do not alter them.You may note peculiarities.For example, the head margin in this template measures proportionately more than is customary.This measurement and others are deliberate, using specifications that anticipate your paper as one part of the entire journals, and not as an independent document.Please do not revise any of the current designations.

Job Strain and Workplace Support
The job stress and workplace social support was measured by the Job Strain Questionnaire (JSQ: Steptoe, Cropley, & Joekes, 1999).Based on Karasek's (1979) job demand-control theory, the JSQ measures Job Demand (three items, e.g., "My job is hectic"), Job Control (three items, e.g., "I have freedom to decide what I do in my job"), and Skill Utilization (four items, e.g., "My job involves me learning new things").(Steptoe et al, 1999) defined the Job strain Index as.

(
) The JSQ also includes five items to measure Workplace Social Support for example "I have a good relationship with my supervisors".The JSQ was translated into Japanese and used in a working people population study (Takagishi, Sakata, & Kitamura, 2011).In the present participants, internal reliabilities expressed by Cronbach's α were .78,.79,and .67 for Job Demand, Job Control, and Skill Utilization, respectively.

Data Analyse
Of 441 participants, 434 (98%) responded data without missing values.After calculating means and SDs of all the variables used in this study, we collated all of them.In a simple correlational matrix, we obtained variables that were correlated significantly with the four WLQ scale scores.We constructed a structural regression model using a structural equation modelling (SEM: Klein, 2005).In the original model (Figure 1), we posited that 1) two latent variables Work Limitation and Mental Disturbance consisted of the scale scores of the WLQ and CORE-OM, respectively, 2) Mental Disturbance predicted Work Limitation, 3) Job Strain and Workplace Social Support as well as temperament and character scale scores predicted both Mental Disturbance and Work limitation, 4) temperament and character scale scores predicted Job Strain and Workplace Social Support, and 5) temperament scale scores predicted character scale scores.The fit of each model with the data was examined in terms of chi-squared (CMIN), comparative fit index (CFI), and root mean square error of approximation (RMSEA).According to conventional criteria, a good fit would be indicated by CMIN/df <2, CFI > 0.97, and RMSEA < 0.05; an acceptable fit by CMIN/df < 3, CFI > 0.95, and RMSEA < 0.08.
Then SEM models were "trimmed" by deleting the least significant path from the model one by one as far as χ 2 (d) did not reach a statistically significant level (Klein, 2005).Because of the cross-sectional nature of the research design, we examined the goodness-of-fit of alternative models.
All the statistical analyses were conducted using the Statistical Package for Social Science (SPSS) version 20.0 and Amos 20.0.

Procedure
The participants were asked via e-mail to participate in the net survey.Anonymity was confirmed.The participants were asked via e-mail to participate in the net survey.Each assessment took about three quarter of an hour.Anonymity was confirmed.

Results
Means and SDs of all the variables used in this study is in Table 1.The four WLQ scale scores were correlated with each other.None of the WLQ scale scores were associated with gender.Age was correlated slightly with TMM (r = −.24,p < .001),MID (r = −.23,p < .001),and OPD (r = −.22,p < .001).
The WLQ scale scores were also correlated with almost all of the CORE-OM and JSQ subscale scores.Also they were correlated with Harm Avoidance, and low Self-directedness and Cooperativeness scores.All the CORE-OM subscales were correlated with each other.They were also correlated with high JSQ Job Satisfaction and low Workplace Social Support scores as well as with Harm Avoidance and low Self-directedness and Cooperativeness scores.JS and SOCS scores were inversely correlated with each other.High Job Satisfaction and low Workplace Social Support were correlated with low Self-directedness and Cooperativeness (Table 2).
Because only Harm Avoidance, Self-directedness and Cooperativeness were correlated with the WLQ, CORE-OM, and JSQ subscale scores, we constructed a SEM structural regression model.The original path model showed an acceptable fit with data: χ 2 /df = 3.3, CFI = .960,and RMSEA = .072(Figure 2).In this model, Harm Avoidance predicted low Self-directedness and Cooperativeness as well as Mental Disturbance.Low Self-directedness and Cooperativeness predicted high Job Strain and poor Workplace Support whereas low Self-directedness but not Cooperativeness predicted Mental Disturbance.Job Strain and poor Workplace Support predicted Mental Disturbance whereas Job Strain but not poor Workplace Support predicted Work limitation directly.
Then we "trimmed" the model by deleting one path by another examining whether the increase of χ 2 did not go beyond 5% significance level (df = 1) no non-significant path could be deleted.In the final model, all the non-significant paths in the original model (Figure 1) could be successfully deleted without a significant increase of χ 2 .
In order to consider another possible model in which Work Limitation would rather predict Mental Disturbance, we added a path from Work Limitation to Mental Disturbance (Figure 3).This non-recursive model showed that it was Mental Disturbance that predicted Work Limitation but not the other way round.About 50% of the variance of Work Limitation was explained by this non-recursive model.

Discussion
As predicted from our research hypotheses, presenteeism measured by the WLQ was associated directly with common mental disturbance measured by the CORE-OM and job strain measured by the JSQ.The final SEM model demonstrated that 53% of the variance of the presenteeism could be explained by these two variables.
Because this was a cross-sectional study, we examined the possibility that presenteeism would predict common mental symptoms rather than the other way round (Figure 3).The path from presenteeism toward common mental symptoms was not significant but the path from common mental symptoms toward presenteeism was significant.Thus we presumed that presenteeism was the result of common mental symptoms.
Of clinical and research interest is the lack of direct path from workplace social support towards presenteeism although workplace social support decreased the severity of common mental symptoms directly.Thus job strain and workplace social support, though sharing a substantial variance between each other acted upon presenteeism differentially.
Another unique finding of this study is the attention paid to the effects of temperament traits, Harm Avoidance in particular, and character traits, Self-directedness and Cooperativeness, on job strain, workplace social support, common mental symptoms, and presenteeism.As indicated by many previous investigations, high Ham Avoidance and low Self-directedness were associated directly with common mental symptoms (cited in Introduction).They were also related to mental symptoms via higher job strain and poorer workplace social support.They were never associated directly with presenteeism.Low Cooperativeness, often reported as being linked to common mental symptoms, lost its direct link toward common mental symptoms but indirectly predicted common mental symptoms through higher job strain and poorer workplace social support.To our best knowledge, this is the first report to refer to the mediating effects of job strain and poor workplace social support on the association between low Self-directedness and Cooperativeness and common mental symptoms.People low in Self-directedness and Cooperativeness may be more likely to cope with the stressful job situations with malad-justive styles and less likely to create supportive interpersonal relationships with peers.The latter leads to common mental symptoms creating presenteeism at workplace.Harm Avoidance, a temperament trait, though linked to common mental symptoms, had no direct paths toward job strain and poor workplace social support.
The clinical implications of this study in regard occupation mental health are three-fold.First, this study indicated that presenteeism in a IT employee population in Japan was moderately linked to common mental symptoms.More emphasis should be paid to mental health of employees in workplace where presenteeism is an important issue.Second, job strain, though having a direct path toward presenteeism, exerts its effects on presenteeism via common mental symptoms.This is coupled with the negative effects of poor workplace social support.Highly strained work situations are thus at risk of mental disorders as well as presenteeism.Third, the present study highlighted importance of personality and character in particular in understanding the mechanism of presenteeism.When coping with presenteeism at workplace, mental health professionals must carry out careful scrutiny of employees' personality.This may lead to better method of not only approaching mental health problems and presenteeism but also creating effective way to prevent such mental health problems.
Limitations of this study include many issues.First, this is a cross-sectional study.We should exercise care when discussing causality.Prospective follow-up studies of employees are essential next steps in the future.However, use of SEM in this study made it possible to speculate such causality.Particularly, a non-recursive model seems to have made such perspective study promising that common mental symptoms are not the result but the mediators of workplace presenteeism.Second, the attrition rate was about 80%.We sent a reminder mail to the eligible employees.However, this did not recruit much more responders.Furthermore we heavily relied on employees' self-report for assessing all the variables.Although web-based investigation was the only option we obtained within a limited resources, replication of the results of this study should conducted in interviewbased studies.In addition, we were unable to make psychiatric diagnosis in workplace participants although the CORE-OM is a reliable and valid measure of general severity of psychological maladjustment.Use of diagnostic structured interview may be a strong method that should be used in future studies.

Conclusion
Our first hypothesis "presenteeism was predicted by job strain as well as lack of workplace social support" was partly supported: Presenteeism was predicted by job strain but indirectly by lack of workplace social support.Our second hypothesis "this effect is mediated by common mental symptoms" was supported.The third hypothesis "high Harm Avoidance and low Self-directedness and Cooperativeness predict presenteeism and common mental symptoms directly" was only partially supported: High Harm Avoidance and low Self-directedness predicted only common mental symptoms that fully mediated the effects on presenteeism.The effect of Cooperativeness on common mental symptoms was mediated via high job strain and pure workplace social support.The last hypothesis "The effects High Harm Avoidance and low Self-directedness and Cooperativeness on presenteeism and common mental symptoms are mediated by job strain as well as lack of workplace social support" was also partly supported: The effects of low Self-directedness and Cooperativeness but not that of high Harm Avoidance were mediated by job strain and poor workplace support.
Taking into consideration the methodological drawbacks, this study suggested that workplace presenteeism in Japanese IT companies could be explained to certain extent by common mental symptoms and job strain and that common mental symptoms mediated the effects of job strain and poor workplace social support as well as high Harm Avoidance and low Self-directedness and Cooperativeness upon presenteeism.

Table 1 .
Means, SDs of all variables used in this study (N = 434).

Table 2 .
Inter-correlations of all variables used in this study (N = 434).