Almost every institution that uses an online assessment tool still runs a good share of its assessments on paper. Practicals, half-yearlies, anything with diagrams, anything the board requires in a particular format. The online tool holds one set of marks, a spreadsheet holds the other, and once a term somebody has to combine them into a single number.
That combining step is where records get damaged, and it is almost never the arithmetic that does it. It is the assessments that have no score. This article follows one student through one term and shows the same set of marks producing anything from 17% to 75% depending purely on how three non-scores are handled.
One student, one term, five assessments
Ananya, Class 9, one term, two subjects. Two assessments online, three on paper - a completely ordinary mix.
| Assessment | Mode | Marks | Out of | State |
|---|---|---|---|---|
| Science unit test | Online | 16 | 20 | Graded |
| Science periodic test | Paper | 14 | 20 | Graded |
| Mathematics class test | Online | - | 25 | One written answer not yet graded |
| Mathematics half-yearly | Paper | - | 80 | Student was absent |
| Science practical | Paper | - | 30 | Marks not entered yet |
Two of the five have a score. The other three each have a different reason for not having one, and those reasons are the entire problem. A spreadsheet column has no way to hold "absent" or "still being graded" - it holds a number or it holds nothing - so at some point somebody types a 0, or leaves the cell blank and lets the formula decide.
What each shortcut costs
Same student, same two real marks of 16 and 14. Only the treatment of the three non-scores changes.
| How the three non-scores are treated | Arithmetic | Term % |
|---|---|---|
| All three entered as zeros | 30 ÷ 175 | 17% |
| Only the absence entered as a zero | 30 ÷ 120 | 25% |
| Only the unentered practical entered as a zero | 30 ÷ 70 | 43% |
| Only the ungraded test entered as a zero | 30 ÷ 65 | 46% |
| All three excluded and flagged instead | 30 ÷ 40 | 75% |
One student, one term, five possible answers.
The spread is 58 percentage points. Nothing about Ananya's work varies across those rows. Every one of those numbers would look perfectly plausible printed on a report, and four of them describe a student who does not exist.
The largest single distortion comes from the absence, because the half-yearly is the biggest paper in the term. Entering one zero for one missed 80-mark exam takes the term from 75% to 25%. That is not a rounding problem you can catch by reviewing the report - it is a fifty-point error created by one keystroke that looked like data entry.
Three statuses that a number cannot hold
Absent is not zero
A zero says the student sat the paper and earned nothing. An absence says the paper contains no information about this student at all. Those are opposite claims, and only one of them is true.
The correct handling is to exclude the assessment from the denominator as well as the numerator. Ananya's 75% is her performance on what she actually sat, which is the only defensible thing to say about her. Whether she should be given a re-test, a zero on purpose, or an exemption is a separate decision, and one a human should make deliberately rather than inherit from a spreadsheet default.
The reason to record the absence rather than leave the row empty is that you cannot recover the distinction later. An empty cell six weeks on is indistinguishable from a mark nobody got round to typing.
Not yet graded is not zero
Ananya's Mathematics test has four multiple-choice questions marked automatically and one written answer waiting for a teacher. Fifteen of its 25 marks are already known.
The temptation is to count the 15 and move on. That is worse than it looks: it reports the student as having scored 15 out of 25 on a paper she may well have scored 19 on, and it does so silently, because 15/25 is a perfectly reasonable-looking number. A partly graded paper should be excluded whole - not partly counted - until the grading is finished, and it should be visible as pending so somebody finishes it.
Not yet entered is not zero either
The practical was sat by everyone and marked on paper. The marks are on a sheet in a staff room. This is the most common state in any institution that runs paper assessments, and it is the one most likely to be silently converted into a zero at report time.
It deserves its own status rather than being lumped in with absences, because it describes a job that is not finished - and the job belongs to a member of staff, not to the student.
The most useful thing a report can do is refuse
Here is Ananya's term broken down by subject:
| Subject | Assessments | Counted | Marks | Percentage |
|---|---|---|---|---|
| Science | 3 | 2 | 30 / 40 | 75% |
| Mathematics | 2 | 0 | - | no percentage yet |
Neither Mathematics assessment is finalised - one is ungraded and the student was absent for the other - so there is nothing honest to say about her Mathematics this term, and the right output is to say exactly that.
This is the behaviour that separates a record from a spreadsheet. A spreadsheet will always give you a number. Averaging an empty range returns something, dividing by a smaller denominator returns something, and none of it warns you that the number rests on no evidence. "No percentage yet" is a more accurate statement about Ananya's Mathematics than any figure between 0 and 100.
Why 75% on two of five is the right answer
It is fair to object that 75% based on two assessments is not much of a term report either. That is true, and it is the point. The number comes with its own working:
- 2 assessments finalised
- 1 awaiting grading
- 1 absence
- 1 with marks not yet entered
Read that way, it stops being a grade and becomes a list of three specific jobs: finish grading one written answer, type in one set of practical marks, decide what happens about one missed half-yearly. Do all three and the record becomes complete - and it can be completed, because nothing was overwritten with a zero.
Grading that single written answer at 4 out of 5 moves Ananya to 49 out of 65. Still 75%, but now resting on three assessments instead of two, and Mathematics finally has a number. That is what progress looks like: the percentage barely moves while the confidence behind it goes up.
Compare that with the 17% version. It is complete, it is precise, it is available immediately, and it is wrong - and because every gap has been filled with a zero, nothing in it indicates that any work remains.
Two things worth knowing
A regrade should flow through on its own
If a teacher fixes a marking error on an online test, the term percentage should change without anyone recalculating anything. That works when the record derives a score from the graded answers each time it is read, rather than storing a total that was correct on the day. Anywhere a term report is assembled by copying totals into another sheet, a correction has to be made twice, and the second one is the one that gets forgotten.
The same student can legitimately show two different percentages
A teacher opening this report normally sees the assessments they created. A head of department or administrator sees everything the institution ran. For a student taught by several teachers those are different sets, so they are different percentages - both correct.
This is only a problem if the report does not say which one it is showing. Any report combining paper and online results should state its scope on the page, because "why does my number not match yours" is otherwise an argument nobody can settle.
What one record actually requires
Very little of this is sophisticated. It is four things:
- Paper assessments are first-class. Name, subject, date, total marks, per-student marks - held the same way an online test is, not in an attached file.
- Missing has a reason attached. Absent, awaiting grading and not yet entered are three different facts and must not collapse into an empty cell.
- Only finalised assessments reach the average, and everything excluded is counted and shown, so the gaps are visible rather than silently absorbed.
- The percentage is derived, never typed. Marks obtained over marks available across everything that counts - which is the same method covered in calculating a term percentage when tests have different total marks.
What you get is a term report that is also a to-do list, and a number nobody has to defend, because the assessments behind it are listed next to it.
The gaps are the information
The instinct when a report has holes in it is to fill them so the document looks finished. Filling them with zeros does not complete the record - it destroys the part that was telling you something. Three empty rows on Ananya's report are not a failure of the report; they are the report doing its job.
Teemly keeps paper assessments and online tests in the same student record, with absent, awaiting-grading and not-yet-entered as distinct states that are excluded from averages and counted separately, so a term percentage always says what it is based on. If you are also setting up how individual papers are scored, our worked example of what negative marking does to a batch average covers that end. You can see how the combined record works on our assessment software for schools page.