Main conclusions and meaning of the results
The CENSOPAS-COPSOQ presented evidence of validity and reliability for the medium version and the short version, therefore its use could be recommended in the Peruvian population to assess occupational psychosocial risks. The goodness of fit indices supports the evidence of the validity by internal structure, while the internal consistency coefficients of alpha and omega support the reliability of the measurement for both versions.
Contrasting the findings with the existing literature
Different studies evaluated the measurement properties of the COPSOQ, using heterogeneous analysis methods and providing results that included different factorial solutions both in the number of items and in the number of dimensions.
Two studies were identified that present inadequate analysis methods, which suggested unstable factor solutions. In Spain, the validity of the COPSOQ-ISTAS-21 in workers was analyzed through a factor analysis with varimax rotation, and the internal consistency of the model was evaluated through the Alpha coefficient [24]. However, this rotation method used was not adequate since it assumed that the items and dimensions are independent, which does not happen within the psychological variables since they are closely related to each other [25], even more in an instrument with such factorial complexity (i.e., with a large number of dimensions and subdimensions). In this Spanish study, 27 subdimensions were found, possibly due to the use of an analysis method that does not conform to the nature of the variables (categorical-ordinal). The short version of the instrument they presented had 38 items [24], which represents 7 more items than the one presented in our study. On the other hand, a study carried out in Brazil validated the average version of the COPSOQ-ISTAS-21 II in university workers [26]. In this study, the authors state that they used confirmatory factor analysis, exploratory factor analysis, and reliability analysis by internal consistency [26], managing to identify a factorial model composed of 21 subdimensions that grouped 70 items. However, in the analysis plan of their article, they point out that they used analysis by principal components, varimax rotation, and the number of dimensions was determined with the eigenvalue. This method of analysis is not recommended as it is highly subjective and can overestimate the number of dimensions [25].
Two other studies identified four-dimensional models, which diverges from our findings where six dimensions were identified in the Peruvian population. A Persian study by COPSOQ evaluated the short version of the instrument [27], finding through confirmatory factor analysis a total of 4 dimensions (32 items): quality of leadership, social support from supervisors, rewards, justice and respect, trust, and predictability (dimension 1), self-rated health, burnout, stress, work-family conflict and emotional demands (dimension 2), the meaning of work, commitment to the workplace, influence at work and role clarity (dimension 3), and offensive behavior (dimension 4). Unlike the CENSOPAS-COPSOQ (our study), the Persian study by Aminian et al., collapses dimensions to achieve a more stable factorial solution. However, conceptually they maintain the same indicators as our study. On the other hand, a study validated the French version of the COPSOQ based on the short version in the Danish language [28]. Through exploratory factor analysis, only 4 dimensions were presented ("interpersonal relationships and leadership", "influence and development", "tension", and "demands") with 32 items. This version is similar in the number of items that are proposed in our study for the short version.
Two other studies argued that a five-dimensional version is more stable compared to other factorial solutions. One of them adapted the medium version of the COPSOQ to Persian and concluded that the items were grouped into 20 subdimensions[29], just like in our model. However, in this study, these subdimensions were grouped into only 5 dimensions (Type of production and tasks, Organization and content of work, Interpersonal relationships and leadership, Work-individual interaction, and Health and well-being) [29]. Likewise, a study that evaluated the validity of the COPSOQ in professional drivers in Spain also reported a better factor solution with 5 dimensions (“Demands”, “Influence and development”, “Interpersonal relationships and leadership”, “Job Insecurity” and “Strain ”)[30]. These two studies present similar results to each other and by what our research identified since their dimensions are very similar to those proposed in our study.
A Chilean study carried out a validity and reliability analysis for the short version of the COPSOQ-ISTAS-21, which they called SUSESO-ISTAS21 [31]. This study proposed a factorial structure of 5 dimensions ("Psychological demands"; "work and skills development"; "social support in the company and quality of leadership"; "compensation" and "double presence") with a total of 20 items. However, this structure composed of 5 dimensions did not present adequate adjustment indices (CFI = 0.795; TLI = 0.762; RMSEA = 0.080)[19]. Despite presenting problems in its validity, it did achieve adequate internal consistency coefficients (α > 0.70)[23]. Therefore, although the instrument presents evidence of reliability, its factorial structure is unstable, and a correlated five-dimensional solution might not be the most appropriate.
The COPSOQ studies have presented heterogeneous factor structures and each one has obtained different dimensions and different numbers of items, according to the characteristics of their populations.
Also, measurement properties have been evaluated in various ways such as exploratory and confirmatory factor analyses, and some studies have used poorly recommended analytical methods that could lead to unstable results [24, 26].
It should be noted that our study provides the use of second-order models, which allows a global score to be obtained by dimension and considers the 20 sub-dimensions within its structure. Therefore, we consider that our study provides a new approach to the available evidence from COPSOQ-ISTAS21 and is renamed CENSOPAS-COPSOQ.
The evidence presented corresponds to the second edition of COPSOQ; however, it should be noted that the third edition of COPSOQ was published in December 2019 [32]. The third edition adds a set of mandatory core items that must be included in all versions of the instrument (short, medium, or long), regardless of the country or context that adapts the scale. This will allow future studies conducted with the third edition of COPSOQ to have a constant set of items in all versions, which will allow comparisons to be made between countries and versions. This represents an important advance for the measurement of psychosocial risks because it will allow knowing the most appropriate items to the context but respecting a set of mandatory core items.
Strengths and limitations
Our CENSOPAS-COPSOQ study had a representative sample of workers from different economic activities and regions of Peru, allowing the representativeness of workers with different characteristics. However, the present study is not without limitations. First, evaluations using bifactor models, ESEM, or analysis of invariance could not be carried out, since there sample size was insufficient to make the models converge; so, there could be other factorial solutions that could be more stable using these models. Second, although it collects information from different participants, the selection strategy was non-probabilistic, so the results could not be generalized to the entire Peruvian territory. Third, CENSOPAS-COPSOQ collects the information on the psychosocial risks at work perceived by the worker, however, an identification of the psychosocial risk at the workplace was not carried out, to corroborate the perception of the worker. Fourth, it was not possible to apply other instruments that assess occupational psychosocial risks, which allows other validity evidence to be presented as evidence of relationship with other variables of the CENSOPAS-COPSOQ.
Implications for public health and decision-making
The COPSOQ is an instrument used in different contexts and countries to assess psychosocial risks at work. Therefore, it is a valid and reliable tool that would allow directing public policies and periodic evaluations for work centers of 25 people or less with the short version (31 items), as well as for work centers with more than 25 workers with the medium version (69 items). Other countries have used COPSOQ to design and evaluate their labor policies [10], therefore, this instrument can be used in Peru as a tool to direct decision-making in occupational and occupational health.
The CENSOPAS-COPSOQ could be used to evaluate the effect of organizational or labor interventions since it would allow seeing if the safety and risk management strategies applied in the workplace have a positive or protective effect on the health of the workers. workers, by carrying out evaluations before and after the interventions.