Abstract
With the proliferation of information technology, big data analysis has found wide applications in today's safety management. Based on a Pearson correlation analysis between the incidence of unsafe student behavior and the coverage of unsafe student behavior by class safety benchmarking evaluation, this study demonstrates how expanding evaluation scope can significantly reduce safety risks in vocational college environments.
Introduction
As higher vocational colleges expand enrollment, student safety management has become increasingly complex. While students in these settings are often proactive thinkers, they frequently lack robust safety awareness, leading to behaviors that violate school discipline. Traditional safety management often suffers from a "narrow evaluation coverage" gap. At Campus B of D College, initial safety evaluations only tracked serious violations (e.g., public intoxication), covering only 13.2% of actual unsafe behaviors. This lack of comprehensive benchmarking provided little incentive for student cadres to supervise common "minor" risks, resulting in an unsafe behavior incidence rate of 9.71%—well above the institutional target of 6%.
Methodology
The researchers implemented a two-phase strategy involving a correlation study and a subsequent management intervention.
1. Pearson Correlation Analysis
To determine the impact of evaluation coverage, 32 classes (1,636 freshmen) were divided into four experimental groups. Each group was subjected to varying levels of safety evaluation coverage, ranging from "serious violations only" to "all non-accidental unsafe behaviors."
2. Benchmarking Evaluation System
Following the analysis, a "multiple two-way" reward and penalty system was established. This system linked individual student behavior to the performance scores of class cadres and the class as a whole, effectively creating a community of shared responsibility.
Statistical Analysis and Insights
The core finding of the study was the identification of a strong negative correlation between evaluation coverage and the incidence of unsafe behavior.
As shown in the statistical analysis, the Pearson coefficient ($R$) reached -0.9652. This indicates that as more types of behavior are included in the benchmarking process, the occurrence of those behaviors drops almost linearly. Furthermore, the data suggests a 90% threshold: once coverage exceeds 90%, the incidence rate stabilizes, suggesting that near-total coverage is essential for maximizing campus safety.
Implementation Results
The transition to a comprehensive benchmarking system yielded significant quantitative improvements across all safety metrics.
Key outcomes included:
- Incidence Reduction: The average monthly incidence of unsafe behaviors dropped from 9.71% to 5.63%, meeting the school's safety target.
- Supervision Boost: The rate of supervision by student cadres for critical tasks (such as temperature checks) rose from 51.2% to 94.8%.
- Serious Incident Decline: Year-over-year data showed serious unsafe behaviors dropped from 48 to 11, while accidental injuries were halved.
- System Coverage: The final implemented system achieved 96.8% coverage of all non-accidental unsafe behaviors.
Conclusion and Future Work
By leveraging Pearson correlation analysis to optimize benchmarking, the college successfully transformed student cadres into active safety monitors. However, the authors note that the index system requires ongoing optimization to maintain fairness. Future research should focus on the generalizability of this high-intensity benchmarking model in different cultural and educational contexts where peer-led supervision may be perceived differently.
Summary updated