Factor structure and measurement invariance of the Turkish COA-IIIA among pre-service teachers in a centralized education system

This study examined the factor structure and measurement invariance (MI) of the Turkish adaptation of the Conceptions of Assessment III-Abridged (COA-IIIA) among pre-service teachers within Türkiye's centralized, exam-driven education system. Using data from 594 participants, confirmatory and exploratory factor analyses revealed that the original four-factor model was inadmissible, while a theoretically coherent two-factor model, Accountability & Improvement and Irrelevance, provided superior fit and interpretive clarity. Multi-group analyses established configural, metric, and scalar invariance across gender, grade level, disciplinary branch, and completion of a measurement and evaluation course, supporting valid latent-mean comparisons. Results indicated higher Accountability & Improvement scores among course completers and senior students, and lower Irrelevance scores among males. The findings suggest that Turkish pre-service teachers interpret assessment as serving intertwined formative and accountability functions, reflecting policy-driven conceptions shaped by a high-stakes testing culture. The study extends the cross-cultural validation of the COA-IIIA and demonstrates that the dimensionality of teachers' assessment beliefs is context-dependent. Implications highlight the need for practice-oriented, feedback-rich teacher education and coherent policy alignment to foster learning-oriented assessment cultures that balance improvement with accountability.