The A-Math Feedback Loop
Doing Additional Mathematics questions is useful only if the work changes what happens next.
A worksheet, school test or full paper should produce evidence. That evidence should improve the diagnosis. The diagnosis should determine the intervention. The intervention should then be retested.
Practice without feedback is activity. Practice with feedback becomes a control system.
A useful A-Math loop is:
Attempt → Evidence → Diagnose → Intervene → Retest.
1. Attempt
The loop begins with independent output.
The attempt may be one question, a mixed set, a timed section or a full paper. What matters is that the student’s own mathematical state becomes visible.
- What can the student start alone?
- Where does working slow?
- Which methods are retrieved?
- Where does the route change?
- Which errors appear?
- What happens under time pressure?
A perfectly supported exercise gives less diagnostic information because the student’s independent state is partly hidden.
2. Evidence
The wrong answer is only the beginning of the evidence.
Look at the path:
- the first hesitation;
- the first unjustified step;
- the point where a condition disappears;
- the line where algebra diverges;
- the moment a poor route is chosen;
- the amount of time consumed;
- whether the student notices the problem independently.
This raises the resolution of feedback.
3. Diagnose
Now classify the earliest meaningful failure.
- Concept: the mathematical model is missing or wrong.
- Retrieval: the knowledge exists but cannot be accessed.
- Recognition: the student does not see when the method applies.
- Method: the chosen route is invalid or unnecessarily expensive.
- Execution: the route is correct but the working breaks.
- Examination control: timing, fatigue, checking or recovery changes the outcome.
The diagnosis should be specific enough to select the next action.
If the diagnosis does not change the intervention, it is probably too vague.
4. Intervene
The repair should match the failure rather than default to “do more questions”.
- Concept failure: rebuild meaning and conditions.
- Retrieval failure: use closed-note reconstruction and delayed return.
- Recognition failure: use mixed questions with reduced topic cues.
- Method failure: compare routes and identify the decision point.
- Execution failure: train the specific algebraic or checking control.
- Examination-control failure: use timed sections, move-on rules or full-paper analysis.
The smallest intervention that repairs the cause is usually better than adding unrelated volume.
5. Retest
The loop is incomplete until the repair is tested independently.
Retest in increasing difficulty:
- same capability without the worked solution;
- fresh question of similar structure;
- delayed return;
- changed representation or mixed context;
- timed or full-paper condition when appropriate.
If the same failure returns, do not simply repeat the intervention. Reopen the diagnosis.
Why Corrections Often Fail
Many students correct by copying the model answer.
This can make the page look complete while leaving the student’s capability unchanged.
A useful correction should answer:
- Where did my original route first fail?
- Why did it fail?
- What signal should I notice next time?
- What should I do differently?
- Can I now reproduce the repair without seeing the answer?
One Attempt Can Produce Several Signals
A difficult problem can expose concept, retrieval, execution and timing issues at the same time.
Do not try to repair all of them simultaneously.
Find the first important cause and repair upstream. Some later symptoms may disappear automatically.
Feedback Should Change Resolution with the Student
A beginner may need feedback such as “this concept is not yet built”.
A stronger student may need much finer feedback: “the mathematics is correct, but route selection is adding two minutes and increasing algebraic risk”.
As performance rises, the feedback system should become more precise rather than simply producing harder questions.
The Tutor’s Role in the Loop
A tutor initially increases the resolution of the loop.
The tutor can see the student’s working, classify the failure, select an intervention and decide what should be retested.
But the long-term goal is to transfer more of this control to the student:
- notice the error;
- identify the failure point;
- choose a repair;
- retest;
- decide whether the problem is closed.
Good feedback eventually teaches self-feedback.
The Weekly Feedback Loop
The same runtime can operate across a week:
- Attempt schoolwork, mixed work or timed work.
- Collect the meaningful failures.
- Rank them by recurrence and downstream impact.
- Repair the highest-value one.
- Retest it later.
- Use the result to decide the next week’s priority.
This makes the timetable responsive to evidence instead of fixed in advance.
A Feedback Loop for Strong Students
For strong students, the loop often focuses on small leaks:
- recognition speed;
- route efficiency;
- late-paper accuracy;
- recovery after a difficult question;
- checking value;
- score variance across papers.
The subject may already be largely learned. Feedback is now refining reliability.
A Feedback Loop for Recovering Students
For recovering students, keep the loop narrow.
One clear attempt, one useful diagnosis, one high-leverage repair and one retest can be more productive than a large stack of undifferentiated worksheets.
Reduce uncertainty before increasing workload.
The A-Math Feedback Loop
Attempt → Observe → Diagnose → Intervene → Retest → Compare → Update.
This is the engine underneath good A-Math study. Every attempt should either strengthen a capability or produce enough evidence to choose a better next action.
For a detailed error-reading method, see How to Read an A-Math Mistake. For the repair sequence after diagnosis, use How to Repair A-Math Mistakes.

