<![CDATA[Mathematical abstraction is a foundational cognitive competency for junior high school students, yet diagnostic instruments that map students’ positions along the Concrete–Representational–Abstract (CRA) continuum are still limited. This study developed and initially validated a Rasch-based diagnostic instrument for mathematical abstraction in single-variable linear equations. Participants were 222 Grade 8 students from seven junior secondary schools in Sintang Regency, West Kalimantan, Indonesia. The instrument consisted of 18 constructed-response items scored 0–3 and administered using an overlapping multi-booklet design; each student completed 9 items, with valid responses per item ranging from 56 to 166. Data were analyzed with the Partial Credit Model using the TAM package in R. Seventeen of 18 items met the acceptable fit criteria, while item A2 (Outfit MNSQ = 1.519) required further review. EAP Person Reliability was 0.755, WLE Reliability 0.674, and Item Separation Reliability 0.889, indicating adequate psychometric quality for classroom diagnosis. Unidimensionality was supported (MADaQ3 = 0.137). The Wright Map classified students into four diagnostic zones: Zone 1 pre-transition Concrete (14.41%), Zone 2 Concrete-to-Representational transition (53.60%), Zone 3 Representational-to-Abstract transition (26.58%), and Zone 4 established Abstract (5.41%). Local dependency appeared between R1 and R4 (Q3 = 1.000). Sensitivity analysis showed that removing R4 only slightly reduced EAP reliability (0.751) while lowering the maximum Q3 to 0.498. A lower-ability measurement gap was also found, with 11 students (4.95%) having WLE standard errors above 1.0. The instrument is an initial diagnostic prototype that requires refinement through R4 revision, additional easier Concrete items, and validation with a larger sample.]]>
Copyrights © 2026
