4.1. Main Findings
The results of the current study support the validity and usefulness of the Swedish version of PSC-4 as an instrument for identifying workplaces having poor, fair, and good OSH practices. PSC scores 8 or lower indicate poor OSH practices that might need urgent actions; scores higher than 8 and up to 12 are classified as fair, suggesting more attention should be paid for OSH practices; and scores higher than 12 indicates a good level of OSH practices. Further, concurrent validity of the three risk levels in relation to central aspects of work environment, strain, and motivational outcomes was corroborated. In sum, the current study confirms the Swedish version of PSC-4 as being a valid and reliable measure for both research and practical use for risk assessment at workplaces.
4.2. PSC in Relation to Background Factors
The average level of PSC was higher for managers, especially for those with staff responsibility, than for other employees. This is in line with findings from a study on PSC among police officers [30
]. At the workplace level, differences in perceptions of psychosocial risks call for a social dialogue between managers and employee representatives for promoting psychosocial risk management [47
]. Previous research has shown that PSC works invariantly in relation to related constructs for public and private sector employees, but it has not analyzed differences in PSC levels between sectors [48
]. In the present study, private sector employees on average reported higher PSC than public sector employees. However, this result was confounded by gender as men from the public sector rated PSC higher than men from the private sector. A consistent finding among both women and men was that the more subordinates the nearest leader had, the lower the average level of PSC. In Sweden, the public sector employs around 1.5 million people and the majority (circa 70%) is women [26
]. While women in the public sector mainly work within care, social work, and schools, the men are employed in more technical jobs [27
]. The managerial structures differ so that managers in the female-dominated jobs have a larger span of control than managers in the male-dominated jobs [28
]. As a line manager, being able to enact good PSC leadership requires, besides commitment and support from senior management, necessary prerequisites, including a reasonable span of control.
4.3. Benchmarks Indicating PSC Risk Levels
The national benchmarks indicating PSC risk levels at Swedish workplaces are lower than the Australian ones [31
]. The Australian benchmarks are, for low-risk PSC, >13.7; for moderate risk, 12.3–13.7; and for high risk, <12.3 (risk levels reported on a scale 12–60 converted to 4–20) [31
]. If we had applied the Australian benchmarks, 66% of the Swedish national sample would have been categorized as having a high risk and only 26% would have been considered at low risk. The overall national average of PSC-4 was also lower for the Swedish than for the Australian populations (11.5 for Sweden compared to 13.2–13.4 reported for the years 2009–2015 from the Australian Workplace Barometer study (p 393 in [36
])). While the difference between Australian and Sweden regarding cutoff for low risk corresponds to the difference in population averages, the difference is larger concerning a high risk. In other words, the interval for moderate risk is larger in Sweden than in Australia. This lack of proportional correspondence between cutoff values for risk levels could be expected since the present Swedish study used different external criteria for developing the benchmarks than the Australian study [31
]. However, the remarkably lower average level of PSC in Sweden compared to Australia calls for a need for further cross-cultural validation including analyses of measurement invariance across international translations of the instrument.
In the present study, we analyzed concurrent validity of PSC risk levels in relation to constructs measured by the COPSOQ questionnaire, i.e., quantitative demands, quality of leadership, job satisfaction, commitment to the workplace, work engagement, stress, and burnout. An earlier version of COPSOQ was also used in one of the first Australian studies placing PSC in its nomological framework [9
]. Our approach was to evaluate how the PSC risk levels related to the Swedish national benchmarks for COPSOQ scales. We found solid support for the established PSC risk benchmarks for use in the Swedish context, both at the individual and workplace levels. First, the mean values of the investigated COPSOQ scales were close to the Swedish national COPSOQ benchmarks [34
] for respondents categorized as having a moderate PSC risk. Next, we found statistically significant differences not only across risk levels in the expected direction but also at a size that can be considered important according to the criterion of minimum 5–10 point for COPSOQ scales used for interpretation of workplace survey results [41
]. As expected, we found good reliability of PSC group means when aggregated to the workplace level [46
]. Around 15 percentage of the variance was attributed to the workplace, which is a little lower than Dollard and Bakker found in their study [9
]. However, 15 percentage can be considered a strong effect [45
], and while Dollard and Bakker investigated a more homogenous sample of schools [9
], the sample in the current study consisted of a wide variety of workplaces.
We have followed the Australian tradition of using aggregated mean scores for workplaces and applied the same cutoff values for the group level and individual level. However, according to Dollard and Bailey, it is important not only to focus on the average score for the group but also to consider that there might be individuals who are at risk and need attention [12
]. An alternative could be to define workplace risk levels based on the proportion of employees being at risk rather than using the aggregated mean score. However, we find it somewhat problematic from a practical as well as an ethical perspective. Usually, workplace risk assessment surveys report mean values for the entire organization and for subgroups of employees (e.g., the way COPSOQ results are reported for the organization, departments, and occupational groups [34
]). We believe it is easier to calculate and implement PSC by following this tradition. Besides, reporting the proportion of employees at risk at smaller workplaces may jeopardize the principle of anonymity, which is considered essential when managing psychosocial risks [49
]. Previously, a vision that all individuals in an organization should have an individual PSC score indicating low risk has been suggested [31
]. If this vision was followed, almost all workplaces would be considered being at risk in the present study. Worth noticing, only a small proportion of employees reporting a low PSC score (indicating a personal high risk) were found at workplaces classified as low risk based on the mean score for all employees.
4.4. Strengths and Limitations
Our study has several major strengths. First, it is an advantage that the benchmarks are based on a random sample of employees and therefore representative for employees working in Sweden. Next, the cross-validation of concurrent validity at the workplace level including a test of appropriateness of aggregation of PSC scores was corroborated. This is especially important for practical use of risk assessment at workplaces and for inclusion in multilevel research. Finally, the study builds on a thorough adaptation process using cognitive interviews and cross-cultural validation conducted prior to the present study [21
However, the study also has some limitations which need to be taken into consideration. The cross-sectional study design does not allow causal claims or testing of predictive validity. There is an increased risk of inflated results due to common method bias as the study is based on questionnaire data only. The response rate of the national study was somewhat low, and despite the use of weighting procedures, this might have induced bias. Finally, the workplace sample is not representative of the whole workforce, which limits the possibilities to generalize results regarding the workplace level.
Using central aspects of the Swedish legislative framework for the criteria-based approach while developing the benchmarks for risk levels is an asset. Research from numerous studies have pointed out that PSC is a precursor for several factors linked to both the health deteriorating and the motivating paths of the Job Demands-Resources Model, not merely mental health. Further, it adds to the literature, placing PSC as a true upstream factor since managerial and organizational priority for stress prevention is closely linked to enacting the legislative framework Still, additional validation of prospective validity of PSC in its Swedish version is recommended, for example, how PSC risk levels predict register-based measures such as mental health, sickness absence, or staff turnover.
4.5. Implications for Practice and Research
There is evidence that three out of four patients diagnosed with a stress-related exhaustion disorder have decreased stress tolerance and other symptoms 7 years after seeking care [50
]. This makes it crucial to find ways of efficiently preventing work-related stress diseases. However, a large-scale regional public sector project reveals that the vast majority of work environment problems are identified to have their roots at the organizational level [51
]. Such findings underline the importance of being able to screen workplaces to identify those in most need for organizational level interventions. As such, PSC-4 can be used as a practical tool for monitoring and identifying risk levels of work conditions that can affect employees’ health and productivity. Thus, the short version of PSC could be included in existing projects, e.g., at The Swedish Association of Local Authorities and Regions, or for use as a screening tool for workplace inspections conducted by The Swedish Work Environment Authority. Further, PSC can be used as an evaluation tool for organization-based psychosocial risk prevention and intervention strategies.
Even though research on PSC is increasingly conducted in different countries, there is a need for further studies applying PSC across different cultures and contexts [19
]. In 2015, Bailey, Dollard, and Richards pointed to a need for future research to ascertain the levels of PSC required in other countries to set standards for worker health [31
]. While the principles behind PSC are found to be cross-culturally transferable [21
], our finding that the average level of PSC was remarkably lower in Sweden compared to Australia supports the need for more international studies to better understand cultural differences of potential importance determining relevant benchmarks for the local context. For the Swedish context, we recommend further validation of the PSC, the benchmarks, and the suggested use of the instrument as a screening tool.