The Dundee Ready Educational Environment Measure: A Confirmatory Factor Analysis in a Sample of Malaysian Medical Students
Muhamad Saiful Bahri Yusoff

Abstract
Introduction: The validity and reliability of the DREEM inventory was established across educational settings. It has been translated into various languages and claimed as a ‘cultural-free tool’ to measure the educational climate at educational institutions. To the author’s knowledge, none of the articles reported its validity and reliability among Malaysian medical students.
Objective: This study aims to evaluate the construct validity of DREEM using confirmatory factor analysis, as well as its internal consistency in a sample of Malaysian medical students.
Method: A cross-sectional study was carried out on 656 medical students from first, third and fifth year. Confirmatory factor analysis was done using AMOS 19 to assess the construct validity. Item reduction was carried out, based on the modification indices, standardized residual covariance, and standardized factor loadings to select which items are fit to remain in the best model fit. Reliability analysis was performed using SPSS 18 to assess internal consistency of DREEM.
Result: A total of 511 (77.9%) completely responded to the DREEM inventory. The proposed five-factor structure of DREEM failed to demonstrate model fit (X2 = 4650.79, RMSEA = 0.076, RMR = 0.057, GFI = 0.693, AGFI = 0.667, CFI = 0.710, NFI = 0.648, RFI = 0.633, IFI = 0.711, TLI = 0.698). The five-factor structure of the shortened DREEM demonstrated model fit (X2 = 297.90, RMSEA = 0.058, RMR = 0.027, GFI = 0.935, AGFI = 0.909, CFI = 0.953, NFI = 0.928, RFI = 0.910, IFI = 0.953, TLI = 0.941). The overall Cronbach’s values for the original and shortened DREEM were 0.936 and 0.921 respectively. The Cronbach’s alpha values for subscales of the original DREEM ranged between 0.58 and 0.82 while the shortened DREEM ranged between 0.53 and 0.82.
Conclusion: This study did not support the proposed five-factor structure of the DREEM. The shortened version demonstrated good degree of goodness of fit with the proposed structure and was found as reliable as the original DREEM. Continued research is required to verify and maximize the psychometric credentials of the DREEM across institutions and nationalities.

Full Text: PDF