A Multidimensional Chinese Reading Ability Scale for Non-Chinese-Speaking Minority Students: Development and Validation

Abstract To address the lack of multidimensional, psychometrically validated Chinese reading assessments for non-Chinese-speaking minority learners in China, this study developed and validated a 12-item Chinese Reading Ability Scale. Data came from 360 primary-school students (180 boys; 180 per school) in 10 classes from two ethnic-minority schools in Xinjiang. Analyses included parallel analysis, EFA, CFA, reliability, convergent/discriminant validity, measurement invariance across gender and school, common method bias, DIF, PCA-based unidimensionality/local independence, and Rasch item fit. Results showed that parallel analysis and EFA supported four factors (Items 1–3, 4–6, 7–9, 10–12), with excellent fit, loadings .811–.866, 71.1% variance, and low factor correlations (.080–.169). CFA confirmed the structure (χ²(48)=29.385, CFI=1.000, TLI=1.011, RMSEA=.000, SRMR=.019; loadings .808–.873). AVE .690–.726, CR .870–.888, HTMT .083–.175, and Fornell–Larcker supported validity; subscale α/ω .869–.888. Invariance was strict across gender and school. No gender DIF; no overall school DIF (Item 3 ETS B but nonsignificant). Common method bias was absent. PCA did not support strict unidimensionality (eigenvalue ratio 1.460), and local dependence was confined within clusters (max |Q3|=.691); all Rasch Infit/Outfit MNSQ .884–.996, ZSTD<2, PT-measure .360–.437. Theoretically, this study extends L2 Chinese reading models by confirming a multidimensional, measurement-invariant structure and challenging strictly unidimensional proficiency assumptions for an underrepresented minority population. Practically, the validated 12-item scale offers a population-sensitive diagnostic tool for identifying reading strengths and difficulties, enabling fair gender/school comparisons and informing differentiated assessment and instruction.