2013 Volume 55 Issue 2 Pages 47-55
Objectives: Several European studies showed that low organizational justice (i.e., procedural justice and interactional justice) was associated with major depressive disorders. In these studies, however, the diagnosis of major depressive disorders may be underestimated because they identified only individuals who visited a doctor and received a diagnosis. Moreover, these studies did not consider neurotic personality traits, which can affect the occurrence of major depressive disorders. The purpose of the present study was to investigate the cross-sectional association of organizational justice with major depressive episodes in the past 12 months more precisely in Japanese employees. Methods: A total of 425 males and 708 females from five branches of a manufacturing company in Japan completed self-administered questionnaires measuring organizational justice, other job stressors (i.e., job strain, social support at work, and effort-reward imbalance), neuroticism, and demographic characteristics. A web-based self-administered version of the computerized Composite International Diagnostic Interview Version 3.0 (CIDI 3.0) was used to assess major depressive episodes. Logistic regression analyses were conducted. Results: In the univariate analysis, low procedural justice and low interactional justice were significantly associated with major depressive episodes in the past 12 months. After adjusting for other job stressors and demographic characteristics, only the association of interactional justice remained significant. The moderating effect of neuroticism on the association of organizational justice with major depressive episodes in the past 12 months was not significant. Conclusions: Low interactional justice may be associated with major depressive disorders regardless or other job stressors or neurotic personality traits.
(J Occup Health 2013; 55: 47–55)
In the past decade, epidemiological research linking organizational justice and employee health has increased1). Organizational justice consists of three components: procedural justice, interactional justice, and distributive justice. Procedural justice refers to voice during a decision-making process, influence over the outcome2), or adherence to fair process criteria, i.e., consistency, lack of bias, correctability, representation, accuracy, and ethicality3). Interactional justice refers to frequency with which supervisors treat subordinates with dignity and respect and provide rationales for decisions4). Distributive justice refers to outcomes that are consistent with implicit norms for allocation, such as equity or equality5). Because the definition of distributive justice essentially overlaps with the effort-reward imbalance (ERI) model6), which is one of the well-known psychosocial job stress models, research in the occupational health field has focused mainly on procedural justice and interactional justice7).
Previous prospective studies in Europe (mainly in Finland, the UK, and the Netherlands), which were detailed in a recent systematic review8), reported that lack of procedural and interactional justice was associated with psychiatric disorders, such as major depressive disorders9–11), as well as self-reported psychiatric morbidity12–14). Some previous studies in Japan reported cross-sectional association of organizational justice with self-reported psychological distress (i.e., depression and anxiety) 15, 16). However, no study has investigated the association of organizational justice with major depressive disorders in Japan. All the previously mentioned studies in Europe9–11) identified major depression based on reports from respondents who visited a doctor and received a diagnosis. Because there is a large proportion of people who suffered from major depressive disorders, but did not seek treatment17), this would greatly underestimate the prevalence/incidence of major depressive disorders, as well as the association of organizational justice with the disorders. Major depressive disorders should be identified among both those who visited a doctor and those who did not visit a doctor, for instance, using a standardized assessment schedule to diagnose major depressive disorders. None of these previous studies considered neurotic personality traits, which have been reported as having a moderating effect on the association of organizational justice with sickness absence18). Therefore, the moderating effect of neuroticism on the association of organizational justice with major depressive disorders should be examined.
As a first step to investigate the association of organizational justice with major depressive disorders more precisely, the purpose of the present study was to investigate the cross-sectional association of procedural justice and interactional justice with major depressive episodes in the past 12 months as assessed by a structured diagnostic instrument in Japanese employees. It was hypothesized that low procedural justice and low interactional justice might be associated with major depressive episode in the past 12 months. In addition, we examined the same associations, adjusting for job demands-control (JD-C) (or demand-control-support [DCS]) 19, 20) and ERI models, which are well-known job stress models and are theoretically thought to be affected by procedural justice and interactional justice21, 22), to determine if procedural justice and interactional justice have unique effects on major depressive disorders independently of JD-C (or DCS) and ERI models. Furthermore, we also examined the moderating effect of neuroticism on the association of organizational justice with major depressive disorders to determine whether the association is greater for individuals with higher levels of neuroticism.
A cross-sectional study of employees from five branches of a manufacturing company in Japan was conducted in August 2009 using both a paper-based self-administered questionnaire and a web-based questionnaire. All employees (N=1,279) were invited to participate in this study. A total of 1,228 employees completed both paper-based and web-based questionnaires (response rate, 96%). The paper-based questionnaire, which included scales on job stressors, demographic characteristics, and neuroticism, was mailed to employees, and they were asked to complete it. At the same time, they were also asked to access a website to complete the web-based questionnaire, which was used to assess whether they had experienced major depressive episodes in the past 12 months. After excluding 95 employees who had at least one missing response on the questionnaires, the final number of respondents was 1,133 (425 males and 708 females). Compared with the final sample (n=1,133), the excluded sample (n=95) was significantly older, while other demographic characteristics were not significantly different. Detailed characteristics of participants and Pearson's correlation coefficients among job stressors are shown in Tables 1 and 2, respectively. Study purposes and procedures were explained to the employees, and written informed consent was obtained from them prior to initiation of the study. The Ethics Committee of the Graduate School of Medicine, The University of Tokyo reviewed and approved the study procedures (No. 2580).
| Male (N=425) | Female (N=708) | |||
|---|---|---|---|---|
| Average (SD) | n(%) | Average (SD) | n(%) | |
| Demographic and work-related characteristics | ||||
| Age (years) | 41.5(8.5) | 33.9(6.4) | ||
| 50–64 years old | 60(14.1) | 13(1.8) | ||
| 40–49 years old | 213(50.1) | 115(16.2) | ||
| 30–39 years old | 103(24.2) | 377(53.2) | ||
| 20–29 years old | 49(11.5) | 203(28.7) | ||
| Education | 15.3(1.7) | 14.9(1.5) | ||
| More than 12 years | 348(81.9) | 646(91.2) | ||
| 12 years or less | 77(18.1) | 62(8.8) | ||
| Marital status | ||||
| Currently married | 328(77.2) | 243(34.3) | ||
| Never married | 78(18.4) | 415(58.6) | ||
| Divorced/widowed | 19 (4.5) | 50 (7.1) | ||
| Life events in the past 12 months | ||||
| None | 134(31.5) | 223(31.5) | ||
| Any | 291 (68.5) | 485 (68.5) | ||
| Chronic physical conditions | ||||
| None | 400(94.1) | 691(97.6) | ||
| Any | 25 (5.9) | 17(2.4) | ||
| Occupation | ||||
| Administrator/clerk | 112(26.4) | 135 (19.1) | ||
| Quality assurance/after service | 33 (7.8) | 7 (1.0) | ||
| Adviser | 2 (0.5) | 244 (34.5) | ||
| Sales/sales engineer | 244 (57.4) | 53 (7.5) | ||
| Call talker | 30 (7.1) | 260 (36.7) | ||
| Others | 4 (0.9) | 9 (1.3) | ||
| Employment contract | ||||
| Manager | 180 (42.4) | 16 (2.3) | ||
| Regular employee | 196 (46.1) | 160 (22.6) | ||
| Group businesses employee | 21 (4.9) | 13 (1.8) | ||
| Contract employee | 18 (4.2) | 230 (32.5) | ||
| Temporary employee | 2 (0.5) | 288 (40.7) | ||
| Others | 8 (1.9) | 1 (0.1) | ||
| Working hours in the past month | 195.3 (27.8) | 172.7 (24.4) | ||
| 191 hours or more | 191 (44.9) | 107 (15.1) | ||
| 173–190 hours | 163 (38.4) | 291 (41.1) | ||
| 172 hours or less | 71 (16.7) | 310 (43.8) | ||
| Major depressive episodes | ||||
| Within the past 12 months | ||||
| Not depressed | 419 (98.6) | 675 (95.3) | ||
| Depressed | 6 (1.4) | 33 (4.7) | ||
| More than 12 months ago | ||||
| Not depressed | 414 (97.4) | 678 (95.8) | ||
| Depressed | 11 (2.6) | 30 (4.2) | ||
| Neuroticism (EPQ-R score)† | 5.25 (3.4) | 5.74 (3.2) | ||
| High (8–12) | 128 (30.1) | 218 (30.8) | ||
| Moderate (5–7) | 108 (25.4) | 232 (32.8) | ||
| Low (0–4) | 189 (44.5) | 258 (36.4) | ||
| Job stressors‡ | Average (SD) | Cronbach's α | Average (SD) | Cronbach's α |
| Procedural justice (OJQ) | 3.57 (0.64) | 0.86 | 3.40 (0.71) | 0.88 |
| Interactional justice (OJQ) | 4.11 (0.70) | 0.93 | 3.87 (0.81) | 0.94 |
| Job demands (JCQ) | 34.0 (5.6) | 0.73 | 33.2 (5.3) | 0.65 |
| Job control (JCQ) | 73.8 (8.8) | 0.72 | 67.1 (9.7) | 0.71 |
| Job strain (job demands/control ratio) (JCQ) | 0.47 (0.09) | 0.50 (0.10) | ||
| Social support at work (JCQ) | 24.7 (3.3) | 0.87 | 24.4 (3.6) | 0.85 |
| Effort (ERIQ) | 14.0 (4.8) | 0.90 | 13.3 (4.5) | 0.86 |
| Reward (ERIQ) | 46.3 (7.3) | 0.89 | 46.2 (7.2) | 0.87 |
| Effort-reward imbalance (effort/reward ratio) (ERIQ) | 0.59 (0.33) | 0.56 (0.29) | ||
| Job stressors | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 |
|---|---|---|---|---|---|---|---|---|
| 1. Procedural justice | ||||||||
| 2. Interactional justice | 0.607** | |||||||
| 3. Job demands | −0.084** | −0.058* | ||||||
| 4. Job control | 0.286** | 0.281** | 0.196** | |||||
| 5. Job strain (job demands/control ratio) | −0.279** | −0.262** | 0.654** | −0.589** | ||||
| 6. Social support at work | 0.498** | 0.686** | 0.028 | 0.311** | −0.208** | |||
| 7. Effort | −0.188** | −0.143** | 0.594** | 0.079** | 0.407** | −0.101** | ||
| 8. Reward | 0.467** | 0.484** | −0.225** | 0.191** | −0.324** | 0.488** | −0.416** | |
| 9. Effort-reward imbalance (effort/reward ratio) | −0.308** | −0.307** | 0.496** | −0.015 | 0.404** | −0.282** | 0.850** | −0.744** |
Organizational justice was assessed using the Japanese version of the organizational justice scale21, 23, 24), sometimes known as the Organizational Justice Questionnaire (OJQ) 25). Moorman21) developed the original OJQ, and its modified version23) has been used widely in research on organizational justice and health9–12, 25, 26). The OJQ consists of a seven-item scale assessing procedural justice and a six-item scale assessing interactional justice both measured on a five-point Likert type scale ranging from 1 (strongly disagree) to 5 (strongly agree). The former scale measures the degree of provision of relevant information to employees and the consistency of decision-making policy in the workplace, while the latter scale measures the degree of fairness and consideration of respondents' supervisors. A total score for each OJQ subscale was calculated by averaging item scores. The modified English version of the OJQ was translated into Japanese, and the internal consistency reliability and factor and construct validity have been reported to be acceptable for this version24). In this sample, Cronbach's alpha was 0.86 for procedural justice and 0.93 for interactional justice for male participants; and 0.88 for procedural justice and 0.94 for interactional justice for female participants (Table 1). The scores of procedural justice and interactional justice were reasonably distributed across workplaces (intraclass correlations [ICCs]=0.39 and 0.63, respectively, among five branches of the company). According to previous studies10, 13), the participants were classified into tertiles (high, moderate, and low) based on each subscale score.
2) Major depressive episodesMajor depressive episodes were assessed using the web-based self-administered version of the depression section of the computerized Composite International Diagnostic Interview Version 3.0 (CIDI 3.0)27). The CIDI is a comprehensive, fully structured diagnostic interview for the assessment of mental disorders, which provides, by means of standard scoring algorithms, lifetime and current diagnoses according to the accepted definitions and criteria of the International Classification of Diseases (ICD-10) and the American Psychiatric Association's Diagnostic and Statistical Manual (DSM-IV). A previous study reported that the computerized version of the CIDI in its self-administered form was an acceptable substitute for the CIDI delivered by human interviewers28). The Japanese version of the CIDI 3.0 showed an acceptable concordance with clinical and structured interview-based diagnosis of DSM-IV major depressive disorders when it was administered as a face-to-face interview29). A preliminary study showed that the self-administered version had moderate sensitivity (71.4% among 14 cases with clinically diagnosed major depressive episode) and high specificity (100% among nine cases without clinically diagnosed major depressive episode). Also, an acceptable degree of one-year test-retest reliability was shown among our sample (prevalence-adjusted bias-adjusted kappa [PABAK]30) =0.87, AC1 statistic31) =0.93). In this study, we used the DSM-IV criteria to determine whether participants suffered from major depressive episodes in the past 12 months and more than 12 months previously.
3) Other job stressorsOther job stressors included job strain (i.e., combination of job demands and job control), social support at work, and ERI. Job demands, job control, and social support at work were assessed using the 22-item Japanese version of the Job Content Questionnaire (JCQ) 32, 33), while ERI was assessed using the Japanese version of the Effort-Reward Imbalance Questionnaire (ERIQ)34, 35). The JCQ includes a job demands scale (five items, response range 12–48, Cronbach's alpha=0.73 and 0.65 for males and females, respectively), a job control scale (nine items, response range 24–96, Cronbach's alpha=0.72 and 0.71 for males and females, respectively), and a social support at work scale (i.e., coworkers and supervisors) (eight items, response range 8–32, Cronbach's alpha=0.87 and 0.85 for males and females, respectively). We used a job demands/control ratio (range 0.125–2.000) to measure the degree of job strain with a continuous measure36). The ERIQ34) consists of a six-item extrinsic effort scale (response range 6–30, Cronbach's alpha=0.90 and 0.86 for males and females, respectively) and an 11-item extrinsic reward scale (response range 11–55, Cronbach's alpha=0.89 and 0.87 for males and females, respectively). We used an effort/reward ratio (response range 0.2–5.0) to measure the degree of ERI. The participants were classified into tertiles according to their job demands/control ratio, social support at work score, and effort/reward ratio, respectively.
4) Demographic and work-related characteristicsDemographic characteristics included age, education, marital status, life events in the past 12 months, and chronic physical conditions. These were assessed using the paper-based questionnaire. Age was classified into four groups: 50–64 years old, 40–49 years old, 30–39 years old, and 20–29 years old. Education was dichotomized into some college or higher (i.e., more than 12 years) and senior high school or less (i.e., 12 years or less). Marital status was classified into three groups: currently married, never married, and divorced or widowed. Life events in the past 12 months were defined as experiencing work-related, individual, or family life events. Chronic physical conditions were defined as a presence of any of the 22 chronic physical conditions, including circulatory, gastrointestinal, neurological, musculoskeletal, and malignant diseases.
Work-related characteristics included occupation, employment contract, and working hours in the past month. These were obtained from the personnel data of the surveyed company. Occupation was classified into six groups: administrator/clerk, quality assurance/after service, adviser, sales/sales engineer, call talker, and others. Employment contract was also classified into six groups: manager, regular employee, group businesses employee, contract employee, temporary employee, and others. Working hours in the past month was classified into three groups based on the tertiles.
5) NeuroticismNeuroticism was assessed using the Japanese version of the Eysenck Personality Questionnaire-Revised (EPQ-R)37, 38) consisting of 12 items measured on a scale ranging from 0 to 12. In this sample, Cronbach's alpha was 0.81 for both males and females. The participants were classified into tertiles according to their EPQ-R neuroticism score.
Statistical analysisLogistic regression analyses were conducted to estimate the odds ratios (ORs) and 95% confidence intervals (CIs) of major depressive episodes in the past 12 months in groups classified according to procedural justice, interactional justice, or other job stressors (i.e., job strain, social support at work, and ERI). The procedure was as follows. We first conducted univariate analyses (i.e., each job stressor was entered in the analyses separately) (Model 1) followed by multivariate analyses (i.e., job stressors were entered in the analyses simultaneously) (Model 2), and finally, we adjusted the multivariate model for demographic (i.e., age, sex, education, marital status, life events in the past 12 months, and chronic physical conditions) and work-related characteristics (i.e., occupation, employment contract, and working hours in the past month) (Model 3). In a series of analyses, a linear trend was also tested to see the dose-response relationship between each job stressor and major depressive episodes in the past 12 month. Furthermore, interaction terms of neuroticism with procedural justice, interactional justice, and other job stressors were added in the analysis to test the potential moderating effect of neuroticism on the associations of each job stressor with major depressive episodes in the past 12 month. Similar analyses were conducted after excluding participants who experienced major depressive episodes more than 12 months previously (i.e., 414 males and 678 females were analyzed). The level of significance was 0.05 (two-tailed). The statistical analyses were conducted using PASW Statistics 18 for Windows.
In the present sample, 3.4% (39 employees: six males and 33 females) were diagnosed as suffering from major depressive episodes in the past 12 month using the self-administered version of the CIDI 3.0 (Table 1). In the univariate logistic regression analyses (Model 1), low procedural justice and low interactional justice as well as high job strain, low social support at work, and high ERI were significantly associated with major depressive episodes in the past 12 months (p for trend<0.05) (Table 1). In the multivariate logistic regression analyses (Model 2), however, only low interactional justice was significantly associated with major depressive episodes in the past 12 months (OR=5.08 [95% CI, 1.47–17.5], p for trend=0.006). This association was still significant after additionally adjusting for demographic and work-related characteristics (Model 3) (OR=4.93 [95% CI, 1.32–18.4], p for trend=0.012). When we entered interaction terms of neuroticism with job stressors in the model, no significant moderating effect of neuroticism on the associations of any job stressor with major depressive episodes in the past 12 month was observed (p>0.05) (data available upon request).
These patterns were unchanged after excluding participants who experienced major depressive episodes more than 12 months previously, although the associations were attenuated (data available upon request).
The 12-month prevalence of major depressive disorders was lower in the present study compared with studies conducted in other industrialized countries17), although it was within the range of the prevalence reported from a community-based survey in Japan (2.1–3.7%)29). In the present study, both low procedural justice and low interactional justice were significantly associated with major depressive episodes in the past 12 months in the univariate analyses. The association of interactional justice with major depressive episodes in the past 12 months remained significant in the multivariate analyses as well as after additionally adjusting for demographic and work-related characteristics. The results revealed no significant moderating effect of neuroticism on the association of organizational justice with major depressive episodes in the past 12 month.
In the univariate analyses (Model 1), low procedural justice and low interactional justice were significantly associated with major depressive episodes in the past 12 months. These findings are consistent with a systematic review of prospective studies in Europe8), which reported that low procedural justice and low interactional justice were associated with doctor-diagnosed psychiatric disorders and major depressive disorders9–11) as well as psychiatric morbidity12–14), and are also consistent with cross-sectional studies in Japan, which reported negative association of procedural justice and interactional justice with psychological distress15, 16). Although it was a cross-sectional study, the present study replicates and supports the evidence of association of low organizational justice with major depressive disorders.
In the multivariate analyses that adjusted for job strain, social support at work, and ERI (Model 2) and additionally for demographic and work-related characteristics (Model 3), the association of interactional justice with major depressive episodes in the past 12 months remained significant. This finding is consistent with the previous study, which reported that interactional justice was significantly associated with doctor-diagnosed depression after adjusting for ERI11). It is suggested that interactional justice affects major depressive disorders more directly instead of being mediated by job strain, social support at work, and ERI. However, this finding is inconsistent with the theoretical notion that interactional justice as well as procedural justice focuses on the basic elements of the social structure in which task-level job characteristics (i.e., job demands, job control, and social support at work) are operating22) and affects distributive justice21), which is almost a synonym of the ERI7). The OJQ, which we used in this study, has a limitation that the interactional justice scale may not distinguish supervisors' attitudes toward fairness as a resource from supervisor support given to an individual employee24). Thus, these findings may be attributable to multi-collinearity between interactional justice and social support at work (r=0.686) (see Table 2), which may lead to less accurate estimation of OR (i.e., wide 95% CI) for the low interactional justice group.
When we entered interaction terms of neuroticism with job stressors in the model, no significant moderating effect of neuroticism on the association of job stressors with major depressive episodes in the past 12 month was observed. This finding suggests that interactional justice is associated with major depressive disorders independently of neuroticism. Low interactional justice, such as unfair treatment by supervisors in daily working life, may increase distress and decrease self-esteem and motivation for employees, which could affect the occurrence of major depressive disorders more directly. Despite its cross-sectional nature, the present study adds more precise evidence of the association between organizational justice and major depressive disorders even after accounting for neurotic personality traits.
| Job stressors | n | No. of cases (%) | Odds ratio (95% confidence interval) | ||
|---|---|---|---|---|---|
| Model 1† | Model 2‡ | Model 3§ | |||
| Procedural justice | |||||
| High (4.00–5.00) | 309 | 7 (2.3) | 1.00 | 1.00 | 1.00 |
| Moderate (3.29–3.86) | 505 | 12 (2.4) | 1.05 (0.41–2.70) | 0.62 (0.22–1.78) | 0.53 (0.17–1.63) |
| Low (1.00–3.14) | 319 | 20 (6.3) | 2.89 (1.20–6.93)* | 0.84 (0.27–2.64) | 0.63 (0.18–2.20) |
| Test for linear trend | p=0.007 | p=0.979 | p=0.653 | ||
| Interactional justice | |||||
| High (4.50–5.00) | 398 | 6 (1.5) | 1.00 | 1.00 | 1.00 |
| Moderate (4.00–4.33) | 390 | 8 (2.1) | 1.37 (0.47–3.98) | 1.81 (0.55–5.95) | 1.89 (0.55–6.52) |
| Low (1.00–3.83) | 345 | 25 (7.2) | 5.10 (2.07–12.6)** | 5.08 (1.47–17.5)* | 4.93 (1.32–18.4)* |
| Test for linear trend | p<0.001 | p=0.006 | p=0.012 | ||
| Job strain (job demands/control ratio) | |||||
| High (0.517–2.000) | 375 | 7 (1.9) | 3.29 (1.39–7.81)** | 2.02 (0.80–5.11) | 1.83 (0.66–5.07) |
| Moderate (0.440–0.516) | 381 | 10 (2.6) | 1.43 (0.54–3.78) | 1.16 (0.42–3.18) | 1.21 (0.42–3.52) |
| Low (0.125–0.439) | 377 | 22 (5.9) | 1.00 | 1.00 | 1.00 |
| Test for linear trend | p=0.003 | p=0.094 | p=0.203 | ||
| Social support at work | |||||
| High (26–32) | 364 | 10 (2.7) | 1.00 | 1.00 | 1.00 |
| Moderate (24–25) | 381 | 7 (1.8) | 0.66 (0.25–1.76) | 0.45 (0.16–1.29) | 0.49 (0.16–1.47) |
| Low (8–23) | 388 | 22 (5.7) | 2.13 (0.99–4.56) | 0.71 (0.27–1.89) | 0.70 (0.25–1.95) |
| Test for linear trend | p=0.028 | p=0.692 | p=0.622 | ||
| Effort-reward imbalance (effort/reward ratio) | |||||
| High (0.612–5.000) | 380 | 22 (5.8) | 2.85 (1.25–6.48)* | 1.55 (0.62–3.87) | 2.05 (0.75–5.61) |
| Moderate (0.413–0.611) | 374 | 9 (2.4) | 1.14 (0.44–3.00) | 1.02 (0.38–2.76) | 0.99 (0.35–2.81) |
| Low (0.200–0.412) | 379 | 8 (2.1) | 1.00 | 1.00 | 1.00 |
| Test for linear trend | p=0.007 | p=0.294 | p=0.124 | ||
Some possible limitations of the present study must be considered. First, participants who perceived greater job stress and experienced major depressive episodes in the past 12 months may have been less likely to participate in this study, even though the response rate in the present study was relatively high (96%). Furthermore, some participants with major depressive episodes in the past 12 months may have been excluded from the final analysis due to at least one missing response on the questionnaires because those with major depressive disorders exhibit poor concentration, which may cause missing responses on questionnaires. These selection biases could result in underestimation of the true findings. Second, as described above, the response rate was much higher in the present study (96%) compared with previous studies in Japan. Because supervisors collected questionnaires in sealed envelopes, it is possible that some participants might have felt forced to participate in the study even though all employees were told that their participation was voluntary and that supervisors were not authorized to open the sealed envelopes. Such a process of data collection may provide imprecise information or represent lack of due consideration. Third, the sample size in the present study was relatively small, which may have reduced the statistical significance of the results. Fourth, our sample came from one particular manufacturing company in Japan; therefore, generalization of the findings should be done with caution. Fifth, job stressors, including organizational justice and major depressive episodes were assessed by self-report, which may result in bias due to a common response style; however this bias was relatively controlled, since our study assessed independent and dependent variables separately (i.e., paper-based questionnaire and web-based questionnaire, respectively) 39). Sixth, the self-administered version of the CIDI had 100% specificity, while the sensitivity was about 70% from the preliminary study, which may lead to a high rate of false-negative diagnoses for major depressive episodes in the past 12 months and thus underestimate the true association. Seventh, while we conducted conservative analyses by entering neurotic personality traits in the model, it is also, to some extent, controversial because a previous study reported that poor psychosocial working conditions predicted deterioration of personality40). Therefore, the present findings should be interpreted with caution. Finally, a causal relationship cannot be determined due to the cross-sectional nature of the study. The present findings may be interpreted that participants who experienced major depressive episodes in the past 12 months may have a more negative view of the work environment and thus evaluate organizational justice as low.
The present study was supported by a Grant-in-Aid for Scientific Research (A) 2009 (No. 20240062) from the Ministry of Education, Culture, Sports, Science and Technology, Japan. Preparation of the manuscript was partially supported by a Grant-in-Aid for Young Scientists (B) 2012 (No. 24790599) from the Ministry of Education, Culture, Sports, Science and Technology, Japan.
