Reviewing a Marked Past Paper Properly

The total is the least useful number on the page

A returned paper gets looked at once, at the top right corner, and then put away. That is the entire review most students do, and it is why doing twenty past papers can produce almost no improvement. The mark tells you where you are. It contains no information at all about what to do next.

Everything useful is in the lost marks, and specifically in why each one went. Two students can both drop 18 marks on the same paper and need completely different terms of work — one has never met the topics, the other knows all of them and is losing marks to rushing and misreading.

So the review begins by covering the total and starting at question one.

The log: four columns, one row per lost mark

The log is deliberately narrow. It is not a set of corrections and it is not a rewriting of the paper — those come later, and only for some of the rows. Its job is to make the pattern visible, which a pile of corrected questions never does.

One row per question where marks were lost, not one per paper. A student who fills this in honestly for two papers will usually see the pattern before they finish the second.

• Question — number and topic, in three words: 'Q14, ratio to fractions'
• Marks lost — how many, out of how many available
• Cause — one of the five categories below, and only one
• Action — the specific next thing, written as a verb: 'redo Q14', 'reteach compound interest', 'practise ten of these against a clock'

Five causes, five different actions

The classification is the part that does the work, and it is also the part students resist, because three of the five categories are variations on 'I could have got that'. Writing it down anyway is the point.

The rule is that each lost mark gets exactly one cause, chosen by asking what would have had to be different for the mark to be earned. Only the first category needs reteaching. The other four need practice of a particular kind, and treating them as content gaps wastes weeks going back over material that was never the problem.

• Never learned it — the method is genuinely unknown. Action: reteach, then a set of questions on that topic alone.
• Knew it, misread the question — the maths was fine, the question asked something else. Action: no reteaching; practise annotating the question before starting.
• Method right, arithmetic wrong — the route was correct and a number went astray. Action: no reteaching; check whether the calculator was used where it was allowed.
• Ran out of time — the question was reached but not finished, or never reached. Action: pacing work; this row is only valid if the paper was properly timed.
• Working not shown — the answer was right or nearly right and the method marks went unawarded. Action: none on the topic; this is about how the script is written.

The only proof the review worked

Corrections done immediately, with the mark scheme open beside you, prove nothing. The method is in front of you, so of course you can follow it. The test is whether it is retrievable when it is not.

So the last step is scheduled rather than immediate: two weeks later, redo the questions from rows one to three of the log, blind, with no notes and no mark scheme, under the same per-mark time as the exam. Questions that come out are finished with. Questions that do not go back into the log with the same classification, which is now a much more serious signal than it was the first time.

This is also the point at which a paper stops being a test and becomes a teaching resource. Most students never revisit a paper at all; the ones who improve fastest use each one twice.

How we run this in class

In our GCSE and IGCSE groups the log is kept by the student, not the tutor, because the classification is the learning. A tutor who fills it in produces a tidy document and teaches nothing. What we do is check the classifications, since students systematically misfile content gaps as careless errors — it is a more comfortable diagnosis.

We also read the script rather than the total. A paper where 20 marks went to unfinished questions at the end and a paper where 20 marks were scattered across the middle describe two different students, and they get two different plans.