A coordinator is asked how Class 9B did on the mid-term. They open the sheet, read off the average, and say 61%. Somebody else, asked the same question the same week, opens a different system and says 53%.
Neither of them is wrong, neither system is broken, and the marks are identical in both. The difference is what each one did with four students who were not in the room.
One paper, thirty students, four absences
Class 9B has thirty students on the roll. The mid-term is out of 50. Twenty-six students sat it, four were absent, and the marks of the twenty-six add up to 790.
That is the entire input. Here is what comes out of it.
| Convention | Divided by | Average mark | Class average |
|---|---|---|---|
| Absent excluded | 26 | 30.4 | 60.8% |
| Absent counted as zero | 30 | 26.3 | 52.7% |
| Difference | — | 4.1 | 8.1 points |
Same 790 marks. The only variable is the denominator.
Eight percentage points is not a rounding difference. It is the gap between a class that looks comfortably above the pass mark and one that looks like it is struggling. On a report that goes to a head of department, it is the difference between no action and an intervention.
The pass rate moves too, and for the same reason:
| Convention | Passed (20/50 or more) | Out of | Pass rate |
|---|---|---|---|
| Absent excluded | 23 | 26 | 88.5% |
| Absent counted as zero | 23 | 30 | 76.7% |
Twenty-three students passed under both conventions. Nobody's mark changed. The reported pass rate moved twelve points.
Why counting an absence as zero is the wrong default
A zero is a statement about performance. It says a student attempted the paper and earned nothing. An absence is a statement about attendance, and it carries no information about the student's ability at all.
Collapsing the second into the first does two separate kinds of damage. It understates the class, which is the visible problem. And it puts whoever was ill at the bottom of every ranked list in the system, which is the one that costs you a teacher's trust. A list of students needing support that is topped by a child who had chickenpox is a list nobody reads twice.
Excluding the absence is the honest default. The average then means what a reader assumes it means: the average of the students who actually sat the paper.
A blank cell is a third thing, and it is the dangerous one
Absent and zero are at least both decisions. A blank cell is the absence of a decision, and it can mean three different things:
- the student was absent
- the paper has not been marked yet
- somebody was interrupted halfway down the column
Those need different actions. The first needs nothing, the second needs a teacher to finish marking, and the third needs someone to find out where they stopped. A system that treats all three as the same thing cannot tell you which one you are looking at.
This is why it is worth distinguishing a result that is absent from one that is not yet scored. A class average calculated while half the written answers are still ungraded is not a wrong number so much as a premature one, and the honest thing for a report to do is say so rather than publish a figure that will move.
Where the two halves of the record disagree
All of the above is manageable inside one system. The trouble starts when the term's marks live in two.
A paper record stores an absence explicitly. There is a row for the student, and that row says they were not there. An online assessment usually has no such thing: a student who did not attempt it simply has no row at all. The absence is represented by nothing.
Those two are easy to merge incorrectly. Read a paper record for a class of thirty and you get thirty rows, four of them marked absent. Read an online quiz for the same class and you might get twenty-six rows, with the four missing students indistinguishable from students who were never assigned the quiz in the first place.
Merge them without deciding what the gaps mean and the four absences silently become either zeros or nothing at all, depending on which side of the join they landed on. The resulting average is a number nobody chose.
What to settle before the next test cycle
This is cheap to fix in advance and awkward to fix afterwards, because reconstructing who was absent three weeks ago means asking a teacher to remember.
- Decide the convention once, in writing. Absences excluded from averages is the sensible default, but the important part is that it is written down and the same everywhere, not that it matches this article.
- Record absence as a letter, never as a blank or a zero. The mark sheet format that supports this takes no extra effort at entry time.
- Keep "not yet marked" separate from "absent". They need different people to do different things.
- Check what your reports do with an absence before you quote one to a parent. The figure is only as trustworthy as the denominator behind it.
None of this makes a class perform better. It makes the number describing the class mean one thing rather than three, which is the precondition for anybody acting on it sensibly.
What the full record looks like for one learner across both halves of a term is worked through in keeping one student record across paper and online tests, and the reporting questions this feeds are in the twelve questions a reports section should answer. The reporting itself is on the assessment software for schools page.