Plenty of IB Math AA HL students answer a bad practice session the same way: by booking another round of questions. The trouble is that a shaky grasp of why a method applies, algebra slips, a clumsy calculator routine, and trouble with long, timed questions each need a different fix. Raw volume treats them all the same.
A randomized field experiment using PISA data, published in Educational Psychology Review, shows the limits of that default. Students who received more mathematics items in an initial test made only small gains in later math performance, and their general problem-solving ability did not improve. Extra exposure alone is a weak plan. So the better question for an HL student is which gap is actually costing marks, and what kind of practice fixes that gap rather than a different one.
Four gap types and how to recognize yours
“I got it wrong” covers at least two very different problems, and the fix for one barely helps the other. Conceptual and procedural gaps feel identical from the inside, so look for the earliest point where things failed. If you needed the markscheme to tell you which method to start with, the gap is conceptual. If you chose the right method but lost marks to sign slips, missed cases, or messy algebra, it’s procedural.
How you react to a worked solution is a second tell. If one more worked example suddenly makes the topic cohere, that points to a conceptual gap. If the solution looks obvious when you read it but you still drop marks on similar questions, that points to a procedural one. Conceptual gaps usually respond best to worked examples and example–problem pairs. Procedural gaps usually respond best to reworking your own errors.
Here are all four side by side. With a conceptual gap, you can follow a worked solution but couldn’t have built the approach yourself. Procedural gaps look different: you know the method, yet errors multiply as the questions get longer. A GDC gap is the one where you know what the question wants, but the calculator slows you down or leads you astray because the function or key sequence you reached for was clumsy or wrong. And in a multi-step-and-pacing gap, each part works on its own, but you lose the thread when one part feeds the next or you run out of time before the final parts.
Most students carry a mix, and it varies by topic. One topic might be mainly a conceptual problem while another is mainly a procedural one, so the plan has to be set at that level of detail. Naming the gap is only half the job. The harder half is knowing what each one asks you to do differently.
Matching each gap type to the right activity
Each gap calls for its own kind of practice, and worked examples are the better tool for conceptual ones. Worked examples and example–problem pairs help students build schemas for multi-step problems and reduce cognitive load compared with unguided practice. That’s the finding of “Developing Problem-Solving Expertise for Word Problems,” a 2022 review in Frontiers in Psychology, framed by cognitive load theory. Put it to work in two steps. First, study a worked solution closely and ask why each step follows. Then tackle a closely matched problem on your own. That pairing is the example–problem pair the review describes, and the worked solution gives you the structure before you’re asked to produce it.
Procedural gaps call for a less comfortable habit: working with your own mistakes. Students who deliberately generated and then corrected their own procedural errors later transferred those procedures more effectively to novel, more complex problems than peers who practiced without errors or only studied others’ incorrect worked examples. That comes from “Deliberately making and correcting errors in mathematical problem-solving practice improves procedural transfer to more complex problems,” a 2024 study in the Journal of Educational Psychology. Its learners made their errors on purpose, but the same logic should carry over to the ones you already make without trying. So catch your mistakes, explain them, and rework those steps until the method runs clean.
GDC gaps call for the narrowest fix, and it isn’t more content questions. Practice the exact calculator function or key sequence that slowed or misled you within the topic where the gap appears. The aim is tool fluency, so the calculator stops costing time in the exam.
Multi-step-and-pacing gaps are different because they only appear when the pieces meet. Worked examples, error repair, and calculator drills each fix one piece. None of them can show whether those pieces hold together across a full exam.
What past papers are actually for
IB Math AA HL past papers test something ordinary drills don’t: all the demands at once. You have to draw on content without prompts, stay accurate through long multi-part questions, and manage time across a full exam. A worked example can repair a conceptual gap, but only a past paper shows whether that understanding survives when the topic isn’t signposted, the parts connect in unexpected ways, and the clock is running.
That gives past papers two jobs. They test whether your methods transfer to exam conditions, and they reveal which gaps are still costing marks.
They’re the right next tool when your main problem is integrating topics and managing timing across multi-step questions. They’re a low-efficiency first response when you can’t choose an approach without cues, which is a conceptual gap. They’re equally poor when you keep losing marks to slips despite choosing the right method, which is a procedural gap.
The paper itself tells you which situation you’re in. If an attempt produces repeated “I wouldn’t know where to start” moments in a topic, go back to worked examples or example–problem pairs for that topic. If it shows clean approaches undone by slips, move to structured self-error analysis before another timed paper. Sitting a full paper again in either case is the wrong move, given that extra test items produced only small gains in the PISA-based experiment.
So the process runs in a loop, not a straight line. A paper you finish and file away has used up hours without telling you much. Write down what each one exposed.
A simple decision framework, topic by topic
“I got it wrong” is too vague to guide the next revision step. What helps is a decision per topic after every session, and a one-page readout makes that quick. Here’s how to run it.
- Straight after the session, spend two minutes noting the topic, the activity type (worked examples, drills, GDC technique, or timed set), and three to five mistakes or slowdowns. Label each one: C (conceptual) if you couldn’t choose an approach, P (procedural) if you slipped mid-execution, T (GDC) if the calculator or tool slowed or misled you, M (multi-step and pacing) if you got lost across steps or ran out of time.
- Tally the labels. Which one showed up most today? If it’s mixed, pick the one that cost you the most marks or time.
- Match the next session on that topic to the dominant label. For C, study a worked example, then do one example–problem pair. For P, rework your exact error until you can do it cleanly twice. For T, isolate the calculator function or key sequence that caused the trouble and drill it briefly before returning to full questions. For M, do a timed multi-part question or a past-paper set focused on pacing and linking steps.
- Escalate when a label repeats. If C, P, or T dominates two sessions in a row on the same topic, don’t push through with more full questions. Switch to the activity matched to that label before you attempt another timed set or past-paper run. If M dominates twice in a row, change how the timed work is structured: drop from a full past-paper set to a single timed multi-part question, and concentrate on pacing and linking the steps before trying another full set.
- Once a week, scan your logs and circle the one or two topics with the most repeated labels. Give those topics next week’s limited time first, instead of spreading it evenly.
The escalation rule exists because repeating a full timed set on a weak topic adds more exposure without addressing the gap. After a few weeks of logs, you’ll also know which topics keep coming back. That’s where your last stretch of revision belongs.
Aim your revision at lost marks
Revision time runs out before the syllabus does, so it matters what each session is aimed at. In my view, spending your limited hours on the activity that matches the gap is a better use of them than adding more questions.
The next time a session goes badly, resist the urge to book another round of the same questions. Label the mistakes, find the label that cost you the most, and let that decide what you do next. Raw volume treats every gap the same. A log of labeled mistakes doesn’t.
