If you ask a teacher for the one revision activity worth doing, most will say past papers. They are right, and the evidence is not close. Practising retrieval under conditions resembling the real thing beats re-reading, highlighting and note-copying by a wide margin, which is what the research on revision techniques keeps finding.
And yet a very large number of students do dozens of past papers and improve barely at all.
The activity is not the problem. The way it is usually carried out is, and the failure has a specific shape.
The version that does not work
It looks like this. You sit down with a past paper and the mark scheme. You start question 1. You get stuck on question 3, so you glance at the scheme, see how it is done, think “yes, of course”, and carry on. There is no clock running. You take a break in the middle. At the end you total up your marks, get something around a grade 6, feel roughly fine about it, and move on to the next paper.
Every step in that is comfortable, and almost every step is a mistake.
Looking at the scheme when stuck turns retrieval practice into reading. The struggle before the answer arrives is the part that builds the memory. Skip it and you have read a worked solution, which is one of the least effective things you can do with an hour.
“Yes, of course” is not knowing. Recognising a method when you see it feels identical to being able to produce it, and it is a completely different skill. This is the fluency illusion, and past papers marked this way are a machine for generating it.
No clock means you never find out that your knowledge is fine and your problem is that you take eleven minutes on a question worth four marks.
Totalling up and moving on is the biggest one, and it deserves its own section.
Doing the paper is not the revision
Here is the thing to change if you change nothing else.
The past paper is the diagnostic. It is not the treatment. Sitting the paper tells you where the marks went. The revision is what you do about it afterwards, and if you finish the paper, write down a score, and start the next paper, you have collected data and then thrown it away.
We looked at 155 GCSE maths scripts marked question by question, and the pattern in where the marks actually went makes the point sharply: 67.9% of every mark lost came from questions that scored nothing at all, and the 3, 4 and 5 mark questions carried 78.1% of the total loss.
That is not a profile you fix by doing more papers. Doing another paper puts you in front of another set of questions you will abandon in the same way. It is fixed by going back to the specific questions that scored zero and working out whether the cause was not knowing the topic, not recognising what was being asked, or running out of time.
Those three problems have completely different solutions, and the score on the front of the paper cannot tell them apart.
A version that works
Not complicated, and it takes about twice as long per paper. You will do fewer papers and learn considerably more from each.
1. Sit it properly, once. Timed, phone elsewhere, no scheme, in one sitting. If you cannot face a whole paper, do half a paper properly rather than a whole one badly. When you get stuck, leave a gap and move on, exactly as you would in the exam.
2. Before you mark it, do a second pass in a different colour. Untimed now, still no scheme. Go back to everything you left blank and everything you were unsure about, and have another go. This costs you nothing and it separates two things that look identical on a marked paper: what you did not know, and what you did not get to. That distinction is most of the value of the whole exercise.
3. Mark it against the scheme, one mark at a time. Not by impression. Line by line, awarding each mark separately. If you do not know what M1 and A1 mean, start here, because self marking without that vocabulary is guesswork.
4. Write a list of the lost marks, with one word each. Not a score. A list. Next to each, write the cause: unknown, misread, rushed, not written down, ran out of time. Five words is enough.
5. Re-do the zeros a week later. From blank, without the scheme. This is the actual revision, and it is the step everyone skips. A question you scored zero on and never returned to is a question you will still score zero on in June.
The category that is embarrassing to admit
When students categorise their lost marks honestly, a large share turn out not to be knowledge gaps at all. They are:
- the method done on a calculator and never written down;
- the answer left as an expression rather than evaluated;
- the unit missing;
- the question that asked about H and was answered about G;
- the last part of a three part question, unattempted, because time ran out on question 2.
None of those are fixed by more content revision, and all of them are fixed inside a fortnight by changing a habit. Which is the good news: the cheapest marks available to most students are not in what they know.
How many papers, and which ones
Fewer than you think, done properly. A student who sits six papers under real conditions and re-does every zero will beat a student who has done twenty with the scheme open. Reliably.
Recent first, but not exclusively. Specifications change, and a paper from before the current specification may contain content you have not been taught, which is discouraging for no reason. Check the specification year. Beyond that, older papers are fine.
Do not save them all for the end. There is a common belief that past papers should be hoarded for the final weeks so you do not “waste” them. This is backwards. A paper sat in February tells you what to spend the spring on. The same paper sat in May tells you what you should have spent the spring on.
Doing the same paper twice is not cheating. Retrieval works on the second attempt too, and if you have genuinely forgotten a question a month later, that is exactly the information you needed.
The one thing that makes all of this harder alone
Every step above depends on marking your own work accurately, and that turns out to be the hardest part of the whole exercise, for reasons that have nothing to do with honesty. The scheme shows you the right answer at the same moment it asks you to judge your own, and after that you cannot reconstruct what you did and did not know. It is worth reading why that happens and what to do about it before you rely on a self-marked score.
The practical version: the number you write on the front is soft. The list of lost marks is the part to trust.
For teachers
The step that gets skipped in class is the same one that gets skipped at home: the return to the zeros. A mock analysed to question level, with the questions everybody abandoned identified, is worth more reteaching time than any general review of the paper.
Marky marks a class set at question level and reports it that way, so the questions with the highest zero rate across the group are visible the same evening rather than after an hour of tallying. See how it works.