Measurement invariance of Rosenberg Self-Esteem Scale between British and Chinese college students

Abstract

The present study examined the factor structure and measurement invariance of the Rosenberg Self-Esteem Scale in college students from Britain (N=150) and China (N=205). Confirmatory factor analyses suggested that the two-factor model, which consisted of a positive self-image factor and a negative self-image factor, could fit the data better than single factor structure especially after deleting the eighth item. Furthermore, factorial structure was invariant across groups in configural level and weak factorial level. After releasing several intercepts, partial strong and strict factorial invariance were certified. Subsequently, DIF (differential item functioning) and response patterns were analyzed, evidence indicates that approximately half items are non-invariant on intercept level. However, literature provides little guidance about the implications for the use of the partially invariant scale. In the end, psychological properties of the nine-item RSES was measured, Cronbach’s α were satisfactory and item-total correlations were good for Chinese data, and acceptable for British data. Finally, limitations and implications were discussed.