The study evaluates a culturally reworded Thai Rosenberg Self-Esteem Scale in 443 boarding-school students. Confirmatory factor analysis supported a bifactor structure with a dominant general self-esteem factor and residual wording-related variance. Rasch analysis supported essential unidimensionality, acceptable item fit, strong reliability, and ordered response thresholds. No item exceeded the prespecified threshold for sex differential item functioning. Scores correlated positively with resilience (r=0.589) and negatively with depressive symptoms (r=-0.601), but precision declined among adolescents with higher self-esteem and broader replication remains necessary.
Key findings
- CFA supported a bifactor model in which most score variance represented general self-esteem, with some residual variance associated with positive and negative wording.
- Rasch modeling supported essential unidimensionality, acceptable item fit, strong reliability, and ordered category thresholds.
- No item showed substantial sex DIF by the prespecified criterion, although the authors call for replication in a larger and more sex-balanced sample.
- Scores correlated positively with resilience (r=0.589) and negatively with depressive symptoms (r=-0.601), both p
- Item targeting provided less precision at higher self-esteem. Revised item 8 performed well, whereas revised item 7 was weaker but remained acceptable.
Why this matters globally
The study supports culturally informed rewording as a way to preserve construct coverage rather than automatically deleting troublesome items, while CFA and Rasch provide complementary structural and item-level evidence. If replicated, the scale may support Thai adolescent surveys and program evaluation. Scores should not be used as a depression diagnosis or high-stakes individual judgment without further validation and decision thresholds.
Thai researcher contribution
All five authors are affiliated with Chiang Mai University: Justin DeMaranville, Peerapat Naprung, Supisara Prompila, Nahathai Wongpakaran, and Tinakon Wongpakaran, spanning the Mental Health Program, Faculty of Medicine, and Department of Psychiatry. The work directly develops a Thai-language, Thai-adolescent measure rather than merely importing foreign norms, although a boarding-school sample cannot represent all Thai adolescents.
Limitations to consider
The boarding-school sample and limited sex balance may not generalize to day schools, out-of-school youth, clinical groups, or adolescents across regions and socioeconomic settings. Cross-sectional correlations with resilience and depressive symptoms establish neither causality nor diagnostic performance. The abstract reports no test–retest reliability or sensitivity to change. Precision is weaker at high self-esteem, item 7 remains relatively weaker, and independent replication should account for possible school clustering.