1. Figure 2/Q13 is internally inconsistent. The legend claims a 4-point Likert scale, but Q13 shows only two bars (267/97). Q8, Q12, and Q13 are factual yes/no/frequency questions, not agree/disagree items. How were they coded, and how is the 73% had not received formal emergency training claim derived?
2. The composite awareness score is questionable. Q2 is self-perceived familiarity and Q7 is perceived prioritization, yet both are scored as awareness. For multiple-choice items, incorrect options receive 0 but no penalty; so selecting all options may maximize score. How was validity established? This undermines conclusions about legal literacy and risk differentiation.
3. Cross-sectional regression cannot support causal language (enhance, linked to). Also, Q11 (familiarity with biosafety policies) overlaps with the outcome. Is this not tautological?
4. Cluster results contradict the Discussion. Results say Cluster 3 had lower current-compliance ratings; Discussion says Cluster 3 demonstrated high compliance. Which is correct? Also, Figure 3C marks add book chapters (), focused lectures (), and case studies () as significant, but the text says only case studies/latter was significant. Reconcile.
5. Figure 1/statistics need clarification. Individual disciplines were not significantly different, and the technical-vs-other difference is very small (~0.5–1 point/18) with Pharmacy n=4. What are effect sizes, adjusted pairwise p-values, and model assumptions? Is the academic/clinical/technical grouping empirically justified or actually WHO-defined?
6. The information-source conclusion is overstated. p=0.086, yet the text concludes university education plays a key role and that media/self-directed learning had weaker awareness. This is not supported by the reported test.
7. Chi-square tests on Q18–Q20 likely violate assumptions. Cluster 1 n=28, multiple-select items, multiple comparisons, and small expected counts. Were corrections or exact tests used?
8. PCA does not reveal clusters; k-means does. Report variance explained, loadings, cluster stability, and whether clusters are driven by one item.
