A 36-item instrument tested with 257 Indonesian undergraduates used three experts, pilot testing and Rasch analysis. Empirical reliability was moderate despite stronger model-based estimates; some items misfit and residual PCA suggested secondary dimensions. Critical thinking in academic writing may combine cognitive, linguistic and cultural components rather than forming a strictly single score.
Key findings
- Observed-data reliability was moderate while model-based reliability appeared stronger. Several items misfit and residual PCA suggested multidimensional structure, supporting a view of critical writing as intertwined with language and culture.
Why this matters globally
The work warns international assessment designers to test item functioning and dimensionality before using total scores for high-stakes decisions or institutional comparison, especially among EFL learners.
Thai researcher contribution
Aleeyah Masae of Yala Rajabhat University contributed language-and-writing assessment expertise to an Indonesian collaboration. The sample was not Thai, so cross-cultural validation is required before use in Thailand.
Limitations to consider
The purposive sample came from one university. The abstract does not report criterion validation against actual writing, and differential item functioning across gender, language, institution or country was not established.