Every July the same thing happens. The big test results come in, the top few students from the second batch move up to the first, a few move the other way, and the year carries on. Nobody records it as a decision, because it does not feel like one. It is just what you do with results.
Then in December somebody looks at the batch averages and something is off. The batch you strengthened has barely improved. The batch you weakened looks like it fell off a cliff in July. Neither of those is a teaching story, and reading them as one is an expensive mistake — usually paid for by whoever teaches the second batch.
A mark has two owners
The whole thing fits in one sentence: a mark belongs to the student who earned it and to the test it was sat in. Those two things move at different speeds.
A student is a person, and people change batch. A test is an event — it happened on a Sunday in August, to a particular batch, and it will not happen again. So a mark gets filed twice: once against the student, once against the batch that sat the paper.
When you move a student in July, only the first of those follows them.
| The question you are asking | What it reads | What a batch move does to it |
|---|---|---|
| How is this student doing? | Every mark the student has, in any batch | Nothing. The record stays whole. |
| How is this batch doing? | Every mark sat in that batch | Their past marks stay with the batch they left. |
Nothing is lost and nothing is duplicated. But those two views answer different questions, and after a reshuffle they stop telling the same story.
The arithmetic nobody does
Take two batches at the end of the first term.
- Batch A — 20 students, average 72
- Batch B — 30 students, average 54
In July you move the top 6 from B up to A, and 4 from A down to B. The six going up averaged 71 in B. The four coming down averaged 55 in A. A completely ordinary reshuffle.
Now assume something that will never be true, to isolate the effect: nobody's performance changes at all. Every student scores in the second term exactly what they scored in the first. Here is what happens to the two averages anyway.
| First term | Second term | Change | |
|---|---|---|---|
| Batch A | 72 (20 students) | 74.8 (22 students) | +2.8 |
| Batch B | 54 (30 students) | 50.5 (28 students) | −3.5 |
Batch A had 20 × 72 = 1440. The four who left took 220, so the 16 who stayed average 76.25. Add the six arriving at 71 and A averages (1220 + 426) ÷ 22 = 74.8.
Not one student got better or worse. A gained nearly three marks, B lost three and a half.
So the batch you promoted people into looks like it is improving, and the batch you promoted them out of looks like it is failing. Both numbers moved because the people behind them changed, and for no other reason.
It gets worse in the direction you would not expect. The 24 students who stayed in Batch B all year are now averaging 49.75 instead of the 54 their batch showed in September — not because they slipped, but because they were always the lower two-thirds of that 54. The reshuffle did not lower their marks. It removed the students who had been hiding them.
That is the number a teacher gets asked about in December, and there is nothing they can say that sounds like anything other than an excuse.
Why the old average is not rewritten, and should not be
The obvious fix is to make the past follow the student: recalculate Batch B's September average using only the students who are in B now. It is the wrong fix, for two reasons.
It would be false. Those six students were in Batch B when they sat that paper. 54 is the true average of the batch that sat the test. Recomputing it around today's membership produces a number describing a group that never sat anything together.
It would move a number you have already given away. You showed Batch B's parents an average of 54 in September. If a reshuffle in July quietly rewrote that to 49.75, the report you handed a parent would disagree with the same report opened in December — and you would have no way of knowing which version anyone had seen. A figure that changes behind you is not a record.
So the snapshot is right. The mistake is not in the data. It is in comparing two snapshots that describe different groups of people and calling the difference progress.
The three comparisons that stay honest
Once you accept that a batch label is not a stable group, the fair comparisons are obvious:
- Compare the same students, not the same batch name. The 24 who stayed in B went from 49.75 to whatever they got next. That is a real answer about real teaching.
- Compare a batch only across a stretch with no movement. September to December is comparable. June to August, with a reshuffle in between, is not.
- Compare each student against their own trajectory. This is the only view a reshuffle cannot distort, because it never depended on who else was in the room.
The third one is why a per-student record matters more at a coaching centre than at a school. Schools move students once a year. Coaching centres move them after every major test, which means the batch average is being reset as a measuring instrument several times a term.
The operation to be careful with: renaming
There is a related move that looks harmless and is not.
Say you want last year's "Class XI Physics" to become this year's "Class XII Physics". The instinct is to rename the batch. But a mark was filed against the batch name that existed when the test was sat, so the rename does not take the history with it. The new name starts with no marks at all, and the old name is left holding a year of results with nobody in it.
Four things to do before the next reshuffle
- Write down each batch's average and its headcount before anybody moves. An average without a date and a headcount beside it cannot be compared to anything later.
- Note who moved, and in which direction. This is the one piece of information that explains every strange number for the rest of the year, and it is the one nobody keeps.
- Do not read the first test after a reshuffle as a teaching signal. Both batches are different groups than they were. Read the second.
- For the students who moved, look at their own trajectory, not their new batch's average. A strong student moved into a stronger batch will drop in rank while improving in marks, and rank is the thing their parents will ask about.
What software changes, and what it does not
Both halves, because the second is usually left out.
What it removes:
- Re-entering a student who moves. Changing a student's batch moves the student, not their marks — their record stays whole and nothing is retyped.
- Drift in published numbers. A batch average you showed a parent in September reads the same in December, because it is tied to the test rather than to today's membership.
- The spreadsheet archaeology. A per-student trajectory is already there, so the honest comparison is available without rebuilding last term by hand.
What it does not touch:
- It cannot tell you whether the move was right. That is a teaching judgment and it stays yours.
- It will not rebuild a batch's history around its current members, and after the argument above, you should not want it to.
- It cannot make a batch average mean more than it does. If you reshuffle every term, that number is a weak instrument no matter what tool is producing it. The student-level view is the one that keeps working.
The arithmetic worth doing once
Take your own two batches, your own averages, and the number of students you moved last July. Work out how much of this year's change in each batch average is explained by the move alone, before any teaching is credited or blamed.
For most coaching centres the answer is: most of it. And once that number is in front of you, the awkward follow-up answers itself — if the batch average mostly measures who you put in the batch, it was never the number to run the place on.
The one that survives a reshuffle is the student's own record, which is the argument for keeping it in one place to begin with. More on that in one student record across paper and online tests, and on the general problem of unlike comparisons in comparing two batches that did not sit the same test.
Related: what absent does to a class average, and if you want this bookkeeping to stop being yours, that is what an assessment platform for coaching institutes is for.