Measuring School Risks: Exposure, Frequency, Severity, and Data Quality
Measuring School Risks tests how you identify exposures, choose exposure units, calculate frequency and severity, judge data quality, and present results in a risk register or matrix. Expect calculations and questions about which data source or metric is appropriate.
Core rules and concepts
- An exposure is a condition that creates a possibility of loss. An exposure unit is the denominator you measure against: student, employee, mile driven, square foot, event, or building.
- Expected loss equals frequency multiplied by average severity. Never rely on claim counts alone.
- Express frequency as a rate per exposure unit per year: claims per 1,000 students, per 100 full-time employees, per 100,000 miles, or per 1,000 facility-use events.
- Severity is average incurred cost including defense. Track median and tail separately, because averages hide catastrophic claims.
- Data sources include incident and near-miss reports, claim loss runs covering five or more years, workers compensation first reports, property inspections, fire marshal reports, safety audits, fleet telematics, human resources records, surveys, benchmarking, insurer loss control reports, and public data on crime, flood, and weather.
- Test data for completeness, consistent definitions, timeliness, and relevance. Underreporting is the largest bias in school data.
- Qualitative tools include the risk matrix with defined likelihood and impact scales, heat maps, scenario analysis, bow-tie diagrams, expert panels, and structured inspections.
- Register fields: identifier, description in cause-event-consequence form, category, owner, existing controls, inherent score, residual score, treatment, and due date.
- Score inherent risk before controls and residual risk after controls, then compare residual risk with tolerance.
- Scan for emerging risks: cyber, student mental health, artificial intelligence use, extreme heat, vaping, e-bikes, and travel.
Worked example
A district operates 42 buses over 1.1 million miles a year and recorded 18 reportable collisions over three years at an average incurred cost of 9,500 dollars. Frequency equals 18 divided by 3.3 million miles, or 5.45 per million miles, about six collisions per year. Expected annual loss equals six times 9,500 dollars, or 57,000 dollars. A telematics and training program costs 40,000 dollars a year and is projected to cut collisions by 25 percent, saving about 14,250 dollars. The savings do not cover the cost, so negotiate the price or pilot at one depot. If average severity instead rose to 20,000 dollars, the same program would clearly pay, because severity dominates the result.
Traps that cost marks
Comparing raw counts across districts of different size. Using immature policy years or ignoring incurred-but-not-reported claims. Omitting defense costs from severity. Using inconsistent matrix scales. Measuring only what is easy to count. Confusing an exposure with a loss.