Testing Measurement Invariance of an Islamic Character Assessment Scale Using Rasch Differential Item Functioning

Abstract

This study examined the measurement invariance of an Islamic Character Assessment Scale (ICAS) across gender and grade levels using Rasch differential item functioning (DIF) analysis. A cross-sectional design with 327 santri (Islamic boarding school students) from pesantren in Polewali Mandar, West Sulawesi, Indonesia, employed stratified random sampling to ensure proportional representation. Rasch Rating Scale Model analysis using Winsteps 5.8.2 evaluated item fit, reliability, and DIF across demographic variables. All 25 scale items demonstrated acceptable fit (Infit/Outfit MNSQ 0.5–1.5), with person reliability of 0.89 and item reliability of 0.96. Gender DIF analysis identified 3 items with statistically significant functioning differences (p < .05), representing 12% of the scale, while grade-level analysis revealed 4 items with significant DIF (16%). All DIF contrasts remained below 1.0 logits, indicating modest practical magnitude. The Islamic Character Assessment Scale demonstrated substantial measurement invariance across gender and grade levels, supporting equitable score interpretation for comparative assessment in Islamic boarding schools. The findings extend psychometric evidence for character assessment in Islamic educational contexts and demonstrate that Rasch DIF analysis provides a robust methodology for evaluating fairness in cultural-specific assessment instruments.