A student scores 42% on a full paper.
The response is:
Do another full paper.
The next score is 43%.
Then 41%.
The problem is not effort.
The revision instrument is operating at the wrong phase.
A revision phase transition is the evidence-based move from one form of practice to another as the learner becomes ready for less support, more selection, more integration and more realistic performance conditions.
This is the third pillar beneath How Revision Works. The master owns the complete revision loop. This page owns stage movement: relearning → supported practice → independent practice → mixed practice → timed sections → full papers, plus the equally important reverse movement when evidence shows the learner is not ready.
Quick Read
Revision should not use one practice format from start to finish. When knowledge is weak, worked examples, explicit explanation and focused question families reduce unnecessary load and reveal the structure. Once execution is reliable, support should fade. Mixed practice then tests whether the learner can select among competing methods. Timed sections add execution pressure after selection is reasonably stable. Full papers are most informative when enough underlying knowledge and question-family competence exist for the paper to test integration, switching, time allocation and endurance rather than merely expose huge unlearned regions. Phase transitions should be gated by performance evidence—not calendar dates, boredom or a belief that “full papers are more serious.”
diagnose → relearn → supported example → partial support → independent same-family questions → changed variants → mixed/interleaved selection → timed section → full paper → classify failures → remain / advance / drop back
Revision Has Different Jobs at Different Phases
Early phase:
rebuild what is missing.
Middle phase:
stabilise independent retrieval and discrimination.
Late phase:
integrate across topics and execute under examination constraints.
If the practice form does not match the current job, revision becomes inefficient.
Phase 1: Relearning and Representation Repair
Use this phase when the learner:
- cannot explain the core idea;
- cannot reconstruct the method at all;
- holds a misconception;
- lacks a prerequisite;
- does not understand the worked solution they are copying.
The correct intervention may include:
- teacher explanation;
- textbook/source return;
- worked example;
- diagram or representation rebuild;
- small prerequisite repair;
- one or two high-information examples.
At this stage, “test more” can be less useful than “teach the broken representation correctly.”
A Full Paper Is Usually Too Coarse for Phase 1
If half the syllabus is genuinely unlearned, a two-hour full paper spends much of its time repeatedly confirming the same fact:
large regions are not ready.
Short targeted diagnostics and teaching create more repair per minute.
Phase 2: Worked Examples and Supported Completion
Once the learner understands the idea, the next problem is producing a correct route.
The IES Organizing Instruction and Study guide recommends interleaving worked-example solutions with problem-solving practice. This supports a gradual exchange of responsibility between model and learner.
A useful sequence:
worked example → explain each step → partially completed example → independent example of the same family.
Do Not Remove Support Just Because the Learner Watched the Example
Watching a teacher solve a problem can create recognition.
It does not guarantee production.
Require the learner to:
- explain why each step exists;
- predict the next step before it is shown;
- complete missing steps;
- solve one near variant without the model visible.
Gate to Phase 3: Independent Same-Family Performance
Move forward when the learner can repeatedly:
- retrieve the method without looking;
- execute with acceptable accuracy;
- explain the governing relationship;
- handle small variations;
- correct ordinary mistakes from feedback.
One correct question is weak evidence.
Repeated independent performance is stronger.
Phase 3: Independent Question-Family Practice
Now support is mostly removed, but the learner still knows what family is being practised.
Example:
Quadratic equations practice.
This phase builds execution and exposes within-family variation.
It should include:
- easy-to-hard variants;
- different representations;
- common traps;
- counterexamples;
- questions where the usual shortcut fails.
Blocked Practice Is Useful Here—Temporarily
Blocked practice reduces the selection problem so the learner can stabilise one procedure.
That is useful when the procedure itself is new.
It becomes limiting when the learner stays blocked so long that the chapter heading continues to choose the method for them.
Gate to Phase 4: Can the Learner Select the Method?
Before mixed practice, each candidate method should be sufficiently familiar that comparison is meaningful.
Then test:
- two or three related problem types;
- no chapter headings;
- no method labels;
- learner states the deciding feature before solving.
If performance collapses because individual methods are still unknown, return to Phase 2/3.
If methods are known but confused, Phase 4 is exactly where work belongs.
Phase 4: Mixed and Interleaved Practice
Mixed practice changes the task from:
execute the named method
to:
identify which method or concept fits this situation.
This is where Retrieval Interference becomes visible and useful.
Accuracy May Drop at the Phase Transition
This does not automatically mean the learner became worse.
The task became richer.
Previously the worksheet supplied category information.
Now the learner must infer it.
The Bjork Lab discusses interleaving as a desirable difficulty that can reduce immediate fluency while improving later discrimination in appropriate settings.
Mixed Practice Needs Error Classification
Wrong answer can mean:
- wrong method selected;
- right method, execution failed;
- question misread;
- prerequisite failed;
- representation not translated;
- time pressure emerging.
Do not respond to every mixed error by returning to full chapter relearning.
Gate to Phase 5: Selection Stable Enough for Timing
Timed work should not be used to teach knowledge that is still fundamentally missing.
Move toward timed sections when:
- most relevant content is retrievable;
- method selection is reasonably accurate;
- common variants are manageable;
- remaining errors are increasingly execution-related;
- untimed accuracy is near the required floor.
Timing should reveal execution under constraint, not merely amplify missing knowledge.
Phase 5: Timed Sections
A timed section introduces:
- pace;
- question switching;
- mark allocation;
- decision speed;
- checking choices;
- recovery after a difficult item.
It is smaller than a full paper, so failure remains easier to diagnose.
Use Timed Sections to Build Execution Without Full-Paper Fatigue
Twenty-minute section.
Mark it.
Classify timing errors.
Repair.
Retest later.
This is often more information-dense than repeatedly sitting full papers before execution routines are stable.
Gate to Phase 6: Is the Paper Ready to Become the Unit?
Full papers become valuable when the learner can benefit from testing:
- integration across topics;
- switching;
- endurance;
- time allocation;
- strategic skipping/returning;
- checking;
- emotional recovery after uncertainty.
If the learner is still failing because entire chapters are inaccessible, the paper remains too blunt an instrument for most of the revision time.
Phase 6: Full-Paper Practice
A full paper is a system test.
It asks whether many previously separate capabilities can coexist under one constrained performance:
- knowledge retrieval;
- question interpretation;
- method selection;
- switching;
- writing/calculation accuracy;
- pace;
- checking;
- endurance.
How Examination Craft Works owns the broader craft of real examination performance. Revision Phase Transitions owns when revision becomes ready to use that environment as its main test.
A Full Paper Must Produce a Repair Map
Score alone:
68%
Not enough.
Paper analysis should identify:
- unavailable knowledge;
- selection errors;
- transfer failures;
- time loss;
- careless execution;
- checking failures;
- question-order issues;
- confidence mistakes.
The next revision phase may remain full-paper—or may drop back to targeted repair for one subsystem.
Phase Progression Is Not One-Way
A student reaches full papers.
Three papers reveal the same electrolysis misconception.
Do not keep hoping another full paper will repair it.
Drop the electrolysis subsystem back to Phase 1/2 while the rest of revision remains in Phase 6.
revision phase is local to the capability, not one global label for the entire student.
Different Topics Can Occupy Different Phases Simultaneously
- Algebra: full-paper maintenance.
- Trigonometry: mixed practice.
- Vectors: targeted repair.
- Statistics: timed-section calibration.
This is normal.
The revision system should track capability states, not force every topic into the same practice format.
Calendar Dates Should Not Force Phase Changes
“September means full papers.”
Maybe.
If underlying capability is not ready, a calendar transition produces repeated measurement without enough repair.
The exam horizon matters, but phase should still be chosen from the best available intervention under the remaining time.
Starting Full Papers Too Late Has a Cost Too
A learner practises topic-by-topic until the final week.
They never train:
- switching;
- whole-paper timing;
- endurance;
- paper navigation;
- checking under pressure.
Content competence can be strong while examination execution remains untested.
The phase transition must happen early enough to leave time for paper-level errors to be repaired and retested.
Phase Transition Depends on Error Distribution, Not Only Average Score
Two students score 70%.
Student A:
errors scattered, mostly execution.
Student B:
two entire chapters missing, everything else near perfect.
Same score.
Different next phase.
Use Gates, Not Rigid Thresholds
There is no universal rule such as:
80% means full papers.
A better gate asks several things:
- Is core knowledge available?
- Are major misconceptions repaired?
- Can methods be selected in mixed sets?
- Can changed representations be handled?
- Is untimed execution reasonably stable?
- Will timing practice now reveal a new class of useful information?
The gate is qualitative but evidence-based.
Revision Triage Determines Which Subsystem Drops Back
A full paper reveals five failures.
Revision Triage decides which repair creates the most leverage.
Do not return every weak area to Phase 1 simultaneously.
Revision Scheduling Determines When the Repaired Subsystem Re-enters
Repair electrolysis today.
Immediate correct practice is not enough.
Revision Scheduling sets a later mixed retest before the topic is trusted again inside full papers.
Stopping Rules Prevent Phase 3 From Becoming Permanent
Students often keep doing familiar chapter questions because they are rewarding and measurable.
The fourth sibling, Revision Stopping Rules, defines when enough evidence exists to move a secure topic out of focused repetition and into mixed maintenance.
A Phase Transition Should Create a New Failure Opportunity
Why move from blocked questions to mixed?
To expose selection failure.
Why move from mixed to timed?
To expose pace and decision-cost failure.
Why move from timed sections to full papers?
To expose switching, endurance and whole-paper strategy.
advance when the next phase can reveal something the current phase can no longer test efficiently.
A Better Revision-Phase Model
diagnostic evidence → relearn representation → worked example + explanation → faded support → independent family practice → mixed selection → timed section → full paper → stage-specific error analysis → local regression/repair or advance → retest
A 30-Lens Revision Phase Audit
- Target: what capability is being revised?
- Current phase: relearn, support, independent, mixed, timed or full paper?
- Knowledge: is the core representation available?
- Misconception: is a wrong model still active?
- Prerequisite: is an earlier weakness blocking progress?
- Worked example: is modelled structure still useful?
- Explanation: can learner explain the steps?
- Fading: can support be reduced?
- Independent execution: can same-family problems be solved?
- Variant: can small surface changes be handled?
- Selection: can method be chosen without a label?
- Interference: do neighbouring methods compete?
- Mixed accuracy: does the learner survive interleaving?
- Transfer: can representation/context change?
- Untimed floor: is accuracy stable before timing?
- Timed section: what new errors appear under pace?
- Latency: which decisions are too slow?
- Checking: can errors be caught under time?
- Switching: can topics change without collapse?
- Endurance: can performance persist through a paper?
- Paper readiness: will a full paper test integration rather than only missing chapters?
- Error distribution: scattered or concentrated?
- Local phase: do topics occupy different stages?
- Regression: what evidence demands dropping back?
- Triage: which subsystem deserves repair first?
- Schedule: when will the repair be retested?
- Stopping: when can focused practice end?
- Exam horizon: is enough time left for paper-level learning?
- Calendar pressure: is date overriding evidence?
- World return: did the phase transition expose and improve the next real performance bottleneck?
Laboratory 1: Classify the Phase
Choose five weak topics and identify the correct current phase for each. Do not use the same revision format for all five.
Laboratory 2: Build a Gate
Write evidence gates for moving one mathematics topic from supported examples to independent practice, then from independent to mixed practice.
Laboratory 3: Paper Failure Decomposition
Take one full paper and classify every lost mark as knowledge, selection, transfer, execution, timing or checking. Decide which failures belong in paper phase and which require a local regression to targeted repair.
For Primary Readers
When learning to ride a bicycle, someone may hold the bike first. Later they let go. Then you ride around cones. Only after that does a long road ride become a useful test. Revision support should change the same way.
For Secondary Readers
Explain why full papers are not always the best early revision tool and design evidence gates from targeted repair to mixed practice to timed sections.
For Advanced Readers
Model revision as a staged control policy in which each phase removes support or adds task dimensions only after prerequisite capabilities cross an evidence floor. The purpose of transition is not difficulty escalation for its own sake but access to a new diagnostic regime: selection, transfer, latency, switching, endurance and whole-performance coordination.
Common Misconceptions
- “Full papers are always the most serious revision.” They are useful system tests only when underlying capabilities are ready enough for integration to be the real question.
- “Once you reach full papers, never return to topic practice.” Paper evidence can send one subsystem back to targeted repair.
- “A lower mixed-practice score means learning got worse.” The task may now include method selection that blocked practice supplied automatically.
- “Timing should start early so students get used to pressure.” Timing cannot substitute for a missing representation or unknown method.
- “All topics should be in the same revision phase.” Capability states can differ across the syllabus.
Research Corridor
- US Institute of Education Sciences — Organizing Instruction and Study — worked examples with problem solving, spacing, quizzing and representation connections.
- Bjork Learning and Forgetting Lab — interleaving, varying practice and desirable difficulty boundaries.
- eduKateSG — Retrieval Transfer.
- eduKateSG — How Examination Craft Works.
Frequently Asked Questions
When should students start full papers?
When enough core knowledge, mixed method selection and untimed execution are stable that a full paper can meaningfully test integration, timing, switching and endurance rather than simply rediscovering large unlearned areas.
Should students stop topic practice after starting full papers?
No. Full papers diagnose the system; recurring local failures should still be repaired with targeted work and later reintroduced into mixed/paper conditions.
How do I know when to move from blocked to mixed practice?
Move when the individual methods are sufficiently understood and independently executable that the next valuable diagnostic question is whether the learner can discriminate and select among them.
Final Thought: Revision Should Change Its Question as the Learner Changes
Early revision asks:
Can you do it?
Later revision asks:
Can you choose it?
Finally:
Can you still do it when everything else is happening at once?
A revision phase transition is justified when the learner has extracted most of the useful information from the current practice form and is ready for the next form to expose a deeper class of failure.
REVISION · FOUR PILLAR LEGS
Return to How Revision Works, or continue through Revision Scheduling, Revision Triage and Revision Stopping Rules. Return to the How X Works Hub.