Faculty Interjudge Reliability of Music Performance Evaluation

Abstract

Assessment of music performance in authentic contexts remains an underinvestigated area of research. This study is an examination of one such context, the inter-judge reliability of faculty evaluation of end-of-semester applied music performances. Brass (n = 4), percussion (n = 2), woodwind (n = 5), voice (n = 5), piano (n = 3), and string (n = 5) instructors evaluating a recent semester's applied music juries at a large university participated in the study. Each evaluator completed a criterion-specific rating scale for each performer and assigned each performance a global letter grade not shared with other evaluators or with the performer. Interjudge reliability was determined for each group's rating scale total scores, subscale scores, and the letter-grade assessment. All possible permutations of two, three, and four were examined for interjudge reliability, and averaged correlations, standard deviations, and ranges were determined. Full-panel interjudge reliability was consistently good regardless of panel size. All total score reliability coefficients were statistically significant, as were all coefficients for the global letter-grade assessment. All subscale reliabilities for all groups except Percussion (which, with an n of 2, had a stringent significance criterion) were statistically significant, with the exception of the Suitability subscale in Voice. For larger panels (ns of 4 and 5), rating scale total score reliability was consistently but not greatly higher than reliability for the letter-grade assessment. There was no decrease of average reliability as group size incrementally decreased. Permutations of two and three evaluators, however, tended on average to exhibit more variability, greater range, and less uniformity than did groups of four and five. No differences in reliability were noted among levels of experience or between teaching assistants and faculty members. Use of a minimum of five adjudicators for performance evaluation in this context was recommended.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 89 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.