Reading your class results: the numbers that mean something
The class average is the least informative number on your result sheet. Four readings that tell you what to teach next.
27 August 2026 2 min read
After an exam, most tutors calculate one number: the class average. It is the easiest to compute and the hardest to act on. An average of 62% is compatible with a class that all scored around 62%, and with a class where half scored 85% and half scored 40%, two situations requiring opposite responses.
Four readings worth the ten minutes
- The spread, not the centre. How far apart are the top and bottom thirds? A wide spread means one lesson is serving two different classes and needs differentiated practice; a narrow low spread means the teaching, not the students, is the variable to change.
- Per-question success rate. The single most useful number you can produce. Which question did the most students lose marks on? That is your next session, and it is often not the question you would have guessed.
- Movement per student, not position. A student who went from 40% to 55% is doing better than one who went from 80% to 78%, and a ranked list shows you the opposite. Rank by change and you find who is actually responding to your teaching.
- The questions everyone got right. These cost exam time and measured nothing. Two or three are fine as a settling-in; ten means the paper spent half its length confirming what you already knew.
The trap of comparing across papers
A class average that rose from 58% to 71% between two different exams tells you almost nothing, because the two papers were not equally hard, and nobody calibrated them. Comparisons are only meaningful across the same paper, which is the argument for reusing a diagnostic at the end of term rather than writing a new one.
What to do with a bad result
Resist the instinct to reteach the whole topic. Take the per-question numbers, find the two questions with the worst success rates, and work out what they have in common. Frequently they share a single prerequisite skill, and one session on that skill fixes both, where a full reteach would have spent three sessions to achieve the same thing.
Sources
Frequently asked questions
What is the most useful statistic after an exam?
The per-question success rate, the percentage of students who got each question right. It tells you exactly what to teach next, and it is usually not the question you would have guessed.
Why is the class average not enough?
An average of 62% describes both a uniform class and one split between 85% and 40%, and those two situations require opposite responses. The spread carries the information the average hides.
Can I compare averages between two different exams?
Not meaningfully, unless the two papers were calibrated to the same difficulty, which they almost never are. Comparisons are only solid across the same paper given twice.
Topics in this article
Related posts
When students use AI: what to change about the homework
Your students have access to AI, they know how to use it, and some of their homework is already produced by it. Three responses are available and two of them…
Using AI to give feedback faster without giving it away
Marking is the largest recoverable block in most tutors' weeks, not because it is hard, but because it is long, and because the useful part of it is a small…