Construct Validity and Reliability of the Malay Version of Rosenberg Self Esteem Scale (RSES) among youth in Southern Malaysia: A Confirmatory Factor Analysis

Abstract

Introduction:

The Rosenberg Self-Esteem Scale (RSES) is widely used to measure self-esteem among adults and youth. This study aims to determine the construct validity and reliability of the Rosenberg Self-Esteem Scale Malay version (RSES-M) among Malaysian youth using Confirmatory Factor Analysis (CFA).

Methods:

We administered the Malay-language RSES to 378 Form Four students in the Kota Tinggi District, selected through multistage sampling. The construct validity of RSES-M was assessed using confirmatory factor analysis (CFA), while internal consistency was measured using Cronbach alpha. AMOS version 26 and SPSS version 20 were used for statistical analysis. We compared three measurement models of the RSES-M for the best relative fit: one uni-dimensional model and two different two-domain models (with different items assigned to each domain for each model).

Results:

The findings indicate that the best model for the RSES-M was a two-domain model, with domain one representing positive self-esteem and domain two representing negative self-esteem. The item “I wish I could respect myself more” demonstrated a strong fit within the CFA model when included under the positive domain of self-esteem (Model 3) compared to negative domain ((Model 2) (Chi-Square/degree of freedom (df) = 3.341, goodness of fit (GFI) = 0.967, Comparative Fit Index (CFI) = 0.905, Incremental Fit Index (IFI) = 0.906, and the Root Mean Squared Error of Approximation (RMSEA) = 0.079 and substantial reliability (Cronbach's alpha for domain one = 0.765, and domain two = 0.648).

Conclusion:

This finding diverges from the original RSES developed by Morris Rosenberg in 1965, which conceptualised the RSES as a unidimensional construct, and other studies that categorised the item "I wish I could respect myself more" under the negative self-esteem domain.