Information Loss in Nursing Clinical Teaching Evaluations Measured by Shannon Entropy
Serican Journal of Medicine · 28.05.2026 · 3 (2)
DOI: 10.17161/sjm.v3i2.25486 ↗Kurz gefasst
Fachartikel in der Zeitschrift Serican Journal of Medicine, online erschienen am 28.05.2026. Verfasst von Yuan Zhang und 5 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
Background
Nursing student ratings of clinical teachers inform faculty development, mentorship, and accreditation. We applied information theory to quantify the information content and discriminative capacity of routine nursing teaching evaluation scores, with three parallel comparators illustrating how rater-evaluatee relationships modulate compression.
Methods
We analyzed 7,105 evaluations from four instruments at a Chinese university-affiliated teaching hospital. The primary analysis focused on nursing student ratings (n = 3,972; 436 students; 227 nurse educators with ≥5 ratings; 44 departments). Comparators: residency teaching scores (n = 2,271), secretary-to-resident 360 (n = 339), and resident self-assessment 360 (n = 523), all on 0-100 integer scales. We computed normalized Shannon entropy (H/H_max), ICC(1,1) with F-distribution CIs, and normalized mutual information (NMI). Cluster-aware, coarser-bin (10-category), zero-score inclusion, and stratified-ICC sensitivity analyses were performed.
Results
For nursing clinical teaching evaluations, the ceiling rate (scores = 100) was 74.8% (95% CI: 73.5-76.2%); normalized Shannon entropy was 0.131 (0.122-0.139), retaining 13.1% of theoretical information capacity. Teacher-level ICC(1,1) was 0.008 (-0.005 to 0.023); NMI = 0.035. Stratified analyses confirmed near-zero discriminability across strata. Comparators spanned a wider compression range: residency was more compressed (ceiling 91.6%, H_norm = 0.084, ICC = 0.065); the two 360 systems showed less compression (H_norm 0.331-0.494; ICC 0.100-0.280).
Conclusions
Routine nursing student ratings exhibited severe compression and limited single-rating discriminability, consistent with limited stand-alone utility for high-stakes individual-level decisions; they do not preclude use within multi-source evidence frameworks. We are not aware of prior applications of this information-theoretic framework to nursing clinical teaching evaluations.
Zitieren
APA
Zhang, Y., He, H., Cui, Y., Chen, Y., Li, N., & Huang, W. (2026). Information Loss in Nursing Clinical Teaching Evaluations Measured by Shannon Entropy. Serican Journal of Medicine, 3(2). https://doi.org/10.17161/sjm.v3i2.25486
Vancouver
Zhang Y, He H, Cui Y, Chen Y, Li N, Huang W. Information Loss in Nursing Clinical Teaching Evaluations Measured by Shannon Entropy. Serican Journal of Medicine. 2026;3(2). doi:10.17161/sjm.v3i2.25486
BibTeX
@article{Zhang2026,
author = {Zhang, Yuan and He, Hua and Cui, Yajiao and Chen, Yaofei and Li, Nu and Huang, Weiyue},
title = {Information Loss in Nursing Clinical Teaching Evaluations Measured by Shannon Entropy},
journal = {Serican Journal of Medicine},
year = {2026},
volume = {3},
number = {2},
doi = {10.17161/sjm.v3i2.25486}
}
EUVIMED