Medición del burnout: Estructura factorial, validez y confiabilidad en trabajadores argentinos

Abstract

Over last decades increasing globalization,privatization and liberalization caused significant changes at work, including demands of learning new skills, the need to adopt new types of works, higher pressure of productivity and quality of work, and time pressure, which, in turn, increased rapidly work stress-related phenomenon such as burnout. To date, burnout has become one of the most active research areas in Occupational Health Psychology, spreading attention from researchers, practitioners and policymakers. Despite thousands of investigations on burnout, there is currently an ongoing debate surrounding its conceptualization and measurement, particularly the dimensions of burnout syndrome. In addition, although several studies have been conducted in Argentina, no study so far has demonstrated adequate psychometric properties of any measure of burnout that supports its use. Consequently, the current study sought to examine the internal structure, validity and reliability of a composite measure of burnout compounded by the three scales from Maslach Burnout Inventory-General Survey (MBIGS) and the depersonalization subscale from Maslach Burnout Inventory-Human Services. A sample of 1903 Argentinean workers from public, private and non-profit sectors participated in the study. To ascertain the more appropriate factor structure of burnout several competing models were tested, specifically: (M1) A onefactor model that assumes burnout as a one-dimensional construct; (M2) A two-factor correlated model with only exhaustion and cynicism (core model); (M3) A two-factor correlated model with exhaustion and cynicism combined into a latent variable and professional inefficacy into another; (M4) A three-factor correlated model with cynicism and depersonalization collapsed into one factor (mental distance), exhaustion and professional inefficacy; (M5) A three-factor correlated model including exhaustion, cynicism and depersonalization; and (M6) A fourfactor correlated model with exhaustion, cynicism, depersonalization and professional inefficacy as latent variables. Confirmatory factor analyses indicated that the two-factor correlated model including the core dimensions of burnout (exhaustion and cynicism) provided the best fit to the data. Reliability analyses demonstrated that both exhaustion and cynicism have good internal consistency and adequate construct reliability, which coefficient values upper .70. As expected, exhaustion and cynicism were significantly and negatively related with engagement and positive affect, and positively related with negative affect, supporting for concurrent validity. Overall, these results are in agreement with recent studies suggesting a more parsimonious conceptualization of burnout that includes only exhaustion and cynicism as more appropriate. We did not find support for professional inefficacy as a component of burnout. Thus, in accordance with some researchers, we suggest that professional inefficacy might be more appropriately conceptualized as a predictor or a consequence –or even both- rather than as a component of burnout. However, due to cross- sectional design used in this study, longitudinal study to test such contention is needed. Furthermore, the current findings support the use of exhaustion and cynicism subscales of MBI-GS for assessing burnout in Argentina, making available a useful tool for practitioners and researchers interested in prevention and treatment of job burnout. Nonetheless, it would be worthwhile to conduct further research in order to examine additional relevant psychometric properties such as test-retest reliability and predictive, convergent and discriminant validity. In addition, it would be important for future research to test the invariance of the two-factor model across different occupations, which might strengthen the findings obtained herein and provide evidence that support the validity of conclusions based on comparison between occupational groups. Finally, future investigation should also investigate the robustness of MBIGS against social desirability bias.