Biology by Bradford
To every student who has ever told me "I knew it, I just went blank": you didn't know it. You recognised it. Those are not the same skill, and only one of them is on the exam.
I've watched a student underline an entire chapter on facilitated diffusion in three different colours, close the folder, and tell me with total sincerity that she had it. Twenty minutes later she sat a timed section on the same subtopic and scored four out of nine. She hadn't skipped the revision. She'd done two hours of it. She had just spent two hours doing something that felt exactly like learning without ever once being learning, and nobody had told her the difference, because from the inside the two feel identical.
That gap, between feeling like you know something and being able to produce it on paper while a mark scheme decides, is the whole exam. Rereading closes the gap in your head. It does nothing for the gap on the page, and the page is the only one that counts on results day.
Rereading is comfortable. That's the problem.
Fluency with a page is not the same as retrieval of its contents, and your brain is genuinely bad at telling the two apart. The words look familiar, the diagram triggers recognition, and recognition feels like knowledge from the inside. It isn't. Dunlosky and colleagues' 2013 review of revision techniques ranked rereading and highlighting among the least effective strategies students actually reach for, and ranked practice testing near the top of the same list. Not a marginal gap either. A large one. The uncomfortable finding wasn't that rereading doesn't work at all, it's that the techniques students like best and the techniques that actually work turn out, most of the time, to be different techniques.
The testing effect is why. Dragging information out of memory under pressure, rather than just recognising it sitting on a page, is the part that actually strengthens the pathway. Every time you force yourself to answer a question cold, you're doing the thing that builds long-term storage. Every time you reread a paragraph you already recognise, you're doing the thing that feels most like effort for the least actual return.
Retrieval alone still isn't enough
Here's where I'd normally stop and say "so just test yourself instead," and for years that's exactly what I told students. It's better advice than rereading. It still isn't enough on its own, because self-testing without a real mark scheme lets you mark your own homework in the most flattering light available. You recall roughly the right idea, you nod at yourself, you move on. The exam does not nod at roughly the right idea.
Take facilitated diffusion B2.1. Describe it and the mark scheme wants: molecules move through a channel or carrier protein, down a concentration gradient, without ATP. Explain it and the scheme wants the mechanism underneath that description, the gradient represents stored potential energy, so movement down it is spontaneous, and the protein only lowers the energy barrier, it doesn't do the pushing. A student marking herself from memory will happily award full marks for describing when the question said explain. A student marking herself against the actual scheme finds out, in real time, that the command term was the whole question.
This is also where you meet the scheme's own quiet mercy, the part nobody tells you about until you've held the actual document in your hand. Own figure rule: use the wrong number from an earlier part, and if the working that follows is logically sound, the later marks still stand. Error carried forward, ecf in every examiner's shorthand, means one slip doesn't have to cost you the rest of the question. AP runs the same trick in different words: one point per bullet, accept any of the following, the scoring guideline being generous in ways the exam paper itself never signals. Rereading your notes will never teach you any of this exists. Sitting down with a coloured pen and the real scheme teaches you inside of five minutes, usually the first time you nearly deduct a mark from yourself that the examiner wouldn't have.
What self-marking actually diagnoses
Every mark lost on a properly self-marked paper falls into one of three buckets, and they need three completely different fixes. This, not the score itself, is the actual point of the exercise.
You didn't know it. A genuine content gap. This is the one bucket rereading can help with, and even then a targeted five minutes on the exact idea beats another pass over the whole chapter.
You knew it but couldn't produce it. The information was in there somewhere. Under time pressure, with a blank box in front of you, it didn't come out. This is a retrieval problem, not a knowledge problem, and the only fix is more retrieval practice under the same pressure, not more rereading.
You said it, but not in scheme language. You wrote something true and got nothing for it, because the marking point wanted a specific causal link, a named structure, a comparison, and you gave a vibe. This bucket is invisible unless you mark against the real document, and it's usually where most of the marks sitting in the middle of the grade range actually live.
Rereading only ever finds the first bucket. The other two stay invisible until you mark your own paper properly.
How I actually get students to do this
Generate the paper properly, then treat the clock as real. I point students at the exam maker, get them to pick their level and whichever topics are due for review, and set a genuine time limit, phone in another room, no half measures. A paper sat with one eye on the notes measures nothing except how well someone can copy.
Mark it in a different colour, against the real scheme, the same day. Not the next day, once the memory of what you meant to write has quietly edited itself into what you needed to write. Every mark scheme point is either there or it isn't. If in doubt, don't award it. The instinct the first time is always to give yourself the mark for basically knowing what you meant. The IB does not do vibes marking, and neither should you.
Sort every lost mark into one of the three buckets, then act accordingly. Content gaps go back to the notes, narrowly, for that one idea only. Retrieval gaps go into the next practice paper, not into another highlighting session. Language gaps go onto a short list of command terms you personally keep fumbling, which is usually shorter, and more embarrassing, than students expect.
The exam maker's seed feature earns its keep here too. Every paper it generates can be regenerated identically from its seed, or rerolled into a fresh version with nothing repeated, so a second attempt at a weak topic is never just the first attempt memorised. I've written before about why the bank exists at all and how it's built from the syllabus statements rather than lifted from real papers, which matters here for exactly the reason it matters for a classroom set: the shape of the exam is what gets rehearsed, not the answer to one specific question a student happened to see twice.
I did the same thing to myself this year, not a self-generated paper, the actual thing, and it was every bit as humbling as it sounds for someone who has taught this course for fifteen years. If sitting down with a genuine mark scheme finds gaps in me, it will find gaps in a seventeen year old who's spent two hours rereading a highlighted textbook and calling it revision. It found gaps in every one of my classes this year. It will find gaps in yours.
Rereading tells you the material is familiar. Marking your own paper against the real scheme tells you whether you can actually produce it, and those turn out to be two entirely different skills. Only one of them is what the exam is testing.
Write the paper. Time it properly. Mark it against the real scheme. Trust the document over your gut. Then go and fix exactly what it found.