دانلود رایگان مقاله نظر سنجی پرسشنامه فرسودگی شغلی مازلاک - عمومی

ABSTRACT

This study tested the dimensionality of the Maslach Burnout Inventory – General Survey (MBI-GS) on a sample of 1190 Romanian healthcare professionals from three county hospitals. Data provided evidence to support the hypothesised three-factor model after removing one item from the cynicism scale: 2(86) = 432.29, CFI = .94, GFI = .95, NFI = .93, and RMSEA = .05. Results of multigroup analysis confirmed the invariance of the 15 items model across professional role, gender, age, and organisational tenure. Structural equation modeling results proved specific relations between occupational factors and burnout dimensions. Our results have practical implications for future research on burnout using the MBI-GS among samples of healthcare professionals.

4. Discussion

The first aim of this research was to investigate the factorial validity of the Maslach Burnout Inventory – General Survey among a sample of Romanian healthcare professionals. Results confirmed the originally proposed three-factor structure with exhaustion, cynicism, and professional efficacy as distinct yet correlated dimensions. Confirmatory factor analysis showed that the fit of the expected three-factor model obtained a significantly superior fit over one and two-factor models. Still, the hypothesised three-factor model proved a modest fit to the data. The analysis of modification indices suggested two improvements: (1) to eliminate item 13 of the cynicism’ scale as it cross-loaded on exhaustion and professional efficacy scales and (2) to allow two residual errors to correlate. After introducing those two modifications, the model significantly improved as indicated by 2(df). Correlations between the residual errors of the two items from the cynicism scale may be a result of their position in the questionnaire and thus the shared variance might be the result of their successive ordering. Other studies which confirmed the original three-factor model found an improved fit if correlations were allowed between residual errors (Langballe et al., 2006; Taris et al., 1999).