TEACHING WISDOM IN MUSIC EDUCATION: SCALE DEVELOPMENT, IRT CALIBRATION, AND INVARIANCE IN PRIMARY RECORDER CLASSES

Abstract

This study develops and validates the Teaching Wisdom in Music Education Scale (TWiMES) as a practice-proximal, fairness-tested measure for primary-school recorder classes. Across three waves and 57 classes (N = 1,460; Grades 3–5), Study 1 generated items from observable classroom routines and established strong content and response-process evidence (I-CVI = .83–1.00; S-CVI/Ave = .92). Study 2 piloted the instrument (n = 512), supporting a three-factor structure—Adaptive Pedagogy, Equitable Orchestration, and Developmental Evaluation—and yielding a 19-item form with good reliability (ω = .86–.91; ω_h = .90). In a new sample (n = 948), Study 3 confirmed the model (WLSMV; CFI = .964, TLI = .958, RMSEA = .041), calibrated items with a graded response item response theory (IRT) model (median a = 1.65; ordered thresholds), and showed high precision across θ ≈ −0.3 to +1.5 (max test information ≈ 16). A 12-item short form preserved ~90% of information in the core range (EAP reliability ≈ .88; r with full form = .98). Multi-group analyses supported configural→metric→scalar invariance across gender, grade, and school locale/SES (largest ΔCFI = .006), with only two items exhibiting small uniform grade DIF and negligible score impact (|Δθ| < .06). TWiMES correlated with rubric-scored recorder proficiency (student r = .23–.29; class r = .34–.41) and with perceived classroom justice/engagement (r = .47/.43); two-level models controlling for grade and SES showed class-level TWiMES predicted proficiency (β = .22–.33, ps < .01). TWiMES thus offers a precise, fair, and transportable measure for equity-relevant comparisons, progress monitoring, and program evaluation in primary music education.