The Tutor Handbook Vol No.0020 | The Mixed Set — How a Tutor Tests Method Selection When No Topic Label Announces the Answer

The Tutor Handbook · Volume 0020 · Series ID THB-0020

Put ten algebra questions on a worksheet titled Simultaneous Equations and the page has already made one important decision for the learner.

It has told the student what kind of problem is coming.

Put the same student in front of a mixed set containing ratio, rate, percentage, simultaneous equations, graphs and geometry, with no topic labels, and a new layer of competence becomes visible.

Before a learner can execute the right method, the learner has to recognise which method belongs.

That is the job of the mixed set.

This is Volume 0020 of The Tutor Handbook, eduKate Sengkang’s long-form practical series on the decisions, observations and handovers inside tutoring.

Volume 0019, The Changed Question, asked whether a learner could recognise the same underlying capability when the surface changed. It ended with a stronger next test:

same item → changed item → mixed item → performance condition.

The Mixed Set owns that third step.

What This Volume Owns—and What It Does Not

The wider theory of interleaving already has canonical owners elsewhere in the eduKate ecosystem. How Interleaving Works | Learning to Choose Before Learning to Execute explains the general learning mechanism. MindOS Learning Manual: Interleaving State owns the learner-state route. Subject-specific pages such as the Secondary 4 Additional Mathematics Learning Guide | Mixed-Topic Route Selection, Transfer and Integration and the Primary 6 Science Mixed Practice Workshop own their disciplinary implementations.

This handbook does not duplicate those pages.

Its job is tutor-operational:

How does a tutor design a mixed set that reveals method selection, preserve enough control to interpret mistakes, and decide what the learner needs next?

The tutor classification remains anchored in the Tutor Classification Model by eduKateSG, the Tutorial Class Control Tower and the Three Modes of Tuition. The Mixed Set is not a new tutor class. It is a practice-design and evidence-reading operation used differently by Class 0 through Class 6 tutors.

It also sits deliberately inside eduKate Sengkang’s larger architecture. The Education Runtime connects learner state, diagnosis, repair, practice, transfer and checking. How Learning Works owns the learning mechanisms. How Studying Works owns the learner’s study interface. How Teaching Works owns the intervention logic. How to Improve Studying owns the improvement loop. The Mixed Set connects them at one precise moment: when practice stops announcing the answer family in advance.

Quick Read

Blocked practice trains execution when the method is already known. Mixed practice adds method recognition and selection.
A learner can be accurate on every separate topic worksheet and still fail when several possible methods compete.
The strongest mixed sets are not random piles. They are designed around useful distinctions between plausible alternatives.
Mix only after the component methods are sufficiently available. Mixing several methods that are all unstable creates noise rather than useful discrimination.
Track selection separately from execution. “Wrong answer” is too coarse.
Ask learners to identify the relationship or method family before solving when diagnostic visibility matters.
Near-neighbour contrasts are often more informative than mixing unrelated topics.
Mixed practice should gradually remove labels, method prompts, chapter order and tutor hints.
Immediate mixed success is stronger evidence than blocked success, but delayed mixed success is stronger still.
Timed mixed sets add a new variable. Do not confuse time-pressure failure with knowledge failure.
In a 3-pax class, independent method choice should happen before peer discussion so one student does not accidentally solve the selection problem for everyone else.
The long-term destination is a learner who can read the task, identify the relevant structure, reject plausible wrong routes, choose a method, execute it and check the result without a worksheet heading doing the thinking first.

1. A Topic Label Is a Form of Help

Students rarely think of a worksheet title as scaffolding.

But it is.

Percentage Increase and Decrease.

Subject–Verb Agreement.

Heat Transfer.

Differentiation.

Each heading narrows the search space. The learner may still need to execute correctly, but the page has already answered part of the question: Which family of knowledge is relevant?

The mixed set removes that assistance.

2. Blocked Practice and Mixed Practice Have Different Jobs

Blocked practice is not the enemy.

If a learner is meeting a new algebraic operation, the tutor may deliberately keep the problem family stable so attention can remain on the mechanics. A sequence of similar questions can help the learner understand the procedure, reduce unnecessary switching cost and build initial fluency.

The problem begins when blocked practice is mistaken for complete readiness.

Blocked practice asks, “Can you execute this method?” Mixed practice adds, “Can you tell when this method belongs?”

3. Selection Is a Real Academic Skill

Examinations rarely behave like a chapter exercise.

A Mathematics paper moves between algebra, geometry, statistics and applications. English Paper 2 moves across grammar, vocabulary, synthesis, cloze and comprehension. Science questions can require observation, comparison, inference, explanation, prediction, variable control or evaluation.

The student therefore needs more than stored procedures.

The learner needs a routing system.

4. Separate Classification, Selection and Execution

A tutor can read a mixed-set attempt through three different questions.

Classification: What kind of problem does the learner think this is?
Selection: Which method, concept or reasoning route does the learner choose?
Execution: Can the learner carry that route accurately?

A student may classify correctly, select appropriately and then make an arithmetic error. Another may execute beautifully once the tutor names the method but fail to select it independently. A third may classify the item into the wrong family before any procedure begins.

Those are different learners.

5. The Final Answer Hides the Most Interesting Decision

Suppose two students both get Question 4 wrong.

Student A selects the correct percentage-change method but subtracts incorrectly.

Student B treats the question as a ratio problem because the wording reminds them of a recent worksheet.

Same final score.

Different repair.

The mixed set becomes useful only when the tutor reads the route before the answer.

6. A Mixed Set Is a Diagnostic Instrument

A strong mixed set is not merely a revision worksheet with many topics.

It can be designed as an experiment.

The tutor may suspect that a learner knows both ratio and percentage procedures but confuses the structural cues that distinguish them. A six-question set can alternate the two families, keep arithmetic load comparable, remove topic labels and require the learner to state the relationship before calculating.

Now the set tests the hypothesis.

7. Do Not Mix What Has Not Yet Been Built

Interleaving several unstable methods can make a learner look globally weak when the real problem is much simpler: the component skills are not ready.

Before using a mixed set, ask:

Can the learner execute each component method when cued?
Can the learner explain the core distinction between the methods?
Is retrieval possible without copying a model?
Are prerequisite skills stable enough that the mix will test selection rather than basic survival?

If the answer is mostly no, return to explanation, blocked practice or targeted repair.

8. The Readiness Threshold Is Not Perfect Mastery

Waiting for absolute mastery before mixing can also be a mistake.

Selection itself needs practice. If every learning session remains perfectly sorted by chapter until the final week before an examination, the learner receives too little experience deciding among alternatives.

A practical threshold is:

The learner can usually execute the component methods when they are named, and the tutor is now ready to test whether the learner can choose among them.

9. Mix Near Neighbours Before Mixing the Whole Syllabus

Randomness is not the same as good mixing.

A ratio question beside a poetry-analysis question and a plant-transport question would certainly be mixed, but the contrast may teach very little if the learner never had trouble distinguishing those worlds.

Near neighbours are often more powerful:

ratio vs rate vs percentage;
factorisation vs solving an equation;
product rule vs chain rule;
cause vs evidence vs description in comprehension;
diffusion vs osmosis vs active transport;
observation vs inference;
summary detail vs supporting example;
mean vs median when outliers matter.

The learner has to notice the discriminating feature.

10. Contrast Is the Engine

When two plausible methods appear close together, the tutor can ask a stronger question than “Which one is correct?”

What is true here that makes Method A fit—and what is missing that makes Method B wrong?

This turns method selection into discrimination.

The learner stops memorising “when I see this word, do this” and begins building boundaries between related structures.

11. The Wrong Methods Should Be Plausible

A mixed set is weak if every item advertises its own answer family.

Good discrimination requires alternatives that a real learner might genuinely confuse.

For a Secondary Mathematics student, simultaneous equations and substitution into a single equation may both involve letters and unknowns, but the structural conditions differ. For English comprehension, a cause question and an evidence question may both require quotation from the passage, but the relationship requested is different. For Science, two diagrams may look similar while one asks for a mechanism and another for a comparison.

Plausible alternatives reveal judgement.

12. Score Selection Separately From Execution

A tutor can track two simple scores:

Score	Question
Selection score	Did the learner choose an appropriate method, concept or reasoning route?
Execution score	Once chosen, did the learner carry it out accurately?

A student with 9/10 selection and 5/10 execution needs a different route from a student with 5/10 selection and 9/10 execution on correctly selected items.

One needs execution repair.

The other needs routing work.

13. Add a Cue Score When the Distinction Matters

Method selection can be correct for a weak reason.

The student says “percentage” because the word discount appeared. That cue happens to work today.

Ask what structural feature actually matters.

A cue score can record whether the learner identified the relevant relationship rather than a superficial keyword.

14. Confidence Can Reveal Fragile Selection

Before solving, ask the learner to give a quick confidence judgement: low, medium or high.

This should not become a bureaucratic ritual.

It becomes useful when calibration is part of the problem.

A student may select the correct method with very low confidence, suggesting unstable recognition. Another may choose the wrong method with high confidence, suggesting a misleading rule that needs explicit contrast.

The canonical owner for the broader mechanism is How Learning Calibration Works.

15. Hesitation Is Evidence, but Not a Verdict

Mixed practice often creates a pause.

That pause can be productive.

The learner is no longer riding the momentum of one repeated method. They are scanning alternatives, comparing features and retrieving a route.

Do not automatically treat slower work as weaker learning.

Ask what the hesitation contains.

16. Ask for the Relationship Before the Method Name

Students can memorise method labels without understanding why they apply.

A stronger prompt is:

What relationship do you see, and what does that relationship allow you to do?

For Mathematics, the relationship might be proportional, additive, quadratic, geometric or statistical. For English, it might be cause, contrast, reference, evidence, purpose or sequence. For Science, it might be a causal mechanism, variable relationship, system interaction or evidence pattern.

Method names come after structure.

17. Teach Rejection, Not Only Selection

Strong method selection includes knowing why an attractive alternative does not fit.

Ask:

Why is ratio tempting here?
What condition rules ratio out?
Why does this look like a cause question?
What wording shows that evidence, not cause, is required?
Why might a student choose diffusion?
Which membrane condition makes osmosis the better model?

Rejection strengthens boundaries.

18. Classify Before Solving When You Need Clean Evidence

Sometimes the tutor should temporarily separate choice from calculation.

Give ten short problems and ask the learner to write only:

the problem family;
the likely method;
one structural cue;
one method that does not fit.

Do not solve yet.

This creates a pure routing exercise. Calculation cannot hide weak selection, and weak calculation cannot contaminate a correct choice.

19. Start With a Two-Choice Mixed Set

For a fragile learner, the smallest useful mixed set may contain only two neighbouring methods.

Ratio or percentage.

Cause or evidence.

Diffusion or osmosis.

Expansion or factorisation.

This creates one clean discrimination boundary.

20. Expand to Three Choices When the Boundary Holds

Once the learner can distinguish A from B, add C.

Ratio vs percentage vs rate.

Cause vs evidence vs purpose.

Diffusion vs osmosis vs active transport.

The search space becomes larger, but still interpretable.

21. Build a Cumulative Mix, Not a Random Heap

A cumulative mixed set should reflect what the learner has actually built.

One useful progression is:

A → B → A/B → C → A/B/C → D → A/B/C/D.

This keeps earlier knowledge alive while expanding the choice problem gradually.

The learner is not asked to face the entire syllabus at once merely because “exams are mixed”.

22. Changed Questions and Mixed Sets Test Different Things

The changed question of Volume 0019 preserved one target family while altering the surface.

The mixed set adds competition between families.

Practice design	Main question
Blocked set	Can the learner execute one known method?
Changed question	Can the learner recognise the same structure under a new surface?
Mixed set	Can the learner choose the relevant structure among plausible alternatives?

These are related but not interchangeable forms of evidence.

23. Delay Makes the Mixed Set Stronger

An immediate mixed set can still benefit from recent activation.

If the tutor just taught ratio, rate and percentage for sixty minutes, the student knows which families are “in play”.

Return several days later and mix the same families with older material.

Now the learner must retrieve both the knowledge and the selection boundary after the lesson context has faded.

How Studying Works and the wider spacing/retrieval library own the study architecture. The tutor’s narrower question here is whether the route still self-selects after time.

24. Time Pressure Is a New Variable

A timed mixed set is not simply a faster mixed set.

The clock changes behaviour.

Students may abandon comparison, choose the first familiar method, skip representation, stop checking or become reluctant to switch routes after a poor start.

That means time pressure should usually be added after the tutor already understands the learner’s untimed selection behaviour.

The canonical performance owner is How Examination Performance Works.

25. The Full Paper Is Not the First Mixed Set

A full examination paper mixes content, representations, difficulty, marks, time, fatigue, response formats and emotional pressure.

That is useful when the learner is ready for whole-system performance.

It is a poor first instrument for diagnosing one selection boundary.

Build upward:

two-method contrast → small mixed set → cumulative mixed set → delayed mixed set → timed mixed set → section → full paper.

26. Mathematics: The Learner Must Choose the Relationship Before the Operation

Mathematics is an obvious home for mixed-set training because the same symbols can support different routes.

A learner may know how to solve equations, factorise quadratics, manipulate ratios and calculate percentages separately.

The examination asks a deeper question:

Which relationship is present here, and which mathematical tool exposes it most cleanly?

Mixed practice trains that routing step.

27. Additional Mathematics: Familiar Symbols Create False Friends

A strong A-Math student may execute product rule, chain rule, trigonometric identities and logarithmic equations accurately on separate worksheets.

Put them together and the question becomes one of classification under symbol similarity.

The learner has to see what role the expression is playing rather than react to a familiar symbol.

The existing Secondary 4 A-Math mixed-topic guide owns the detailed subject route. The handbook keeps the tutor’s cross-subject evidence logic.

28. English Comprehension: Question Type Is a Relationship, Not a Keyword

English mixed sets can combine reference, cause, evidence, inference, vocabulary-in-context, purpose and comparison questions.

If the learner uses keywords alone, selection becomes brittle.

Ask what relationship the question requests.

“Why” often signals cause, but not every causal question uses the word “why”. “What shows” often asks for evidence, but evidence can be requested in many forms.

The mixed set exposes whether question classification has become semantic rather than cosmetic.

29. English Grammar: Mixed Sentences Remove the Rule Heading

A learner can score highly on a worksheet labelled Pronouns because attention is already pointed at pronoun relationships.

Mix agreement, tense, reference, connectors, modifiers and punctuation.

Now the learner must notice what kind of relationship is unstable before applying a rule.

This resembles real editing, where the page does not announce the error family beside each sentence.

30. Writing: Mixed Decisions Occur Inside One Piece of Work

Writing does not present one clean question at a time.

The writer must decide whether the current problem is idea selection, paragraph structure, evidence, sentence control, vocabulary precision, cohesion, register or revision priority.

A tutor can create a mixed editing set from short excerpts with different hidden faults. The learner identifies the main job before rewriting.

That trains diagnosis inside production.

31. Science: The Topic Is Not the Scientific Job

A Science question about plants can ask for observation, variable identification, comparison, mechanism, inference, prediction or experimental design.

Topic recognition is not enough.

The learner also needs to identify the scientific job.

That is why good mixed Science practice varies both content and reasoning demand while keeping enough structure visible for the tutor to locate the first failure.

32. Vocabulary: Recognition and Production Belong in Different Mixes

Vocabulary mixed practice can combine meaning, synonym boundaries, collocation, register, morphology and sentence use.

But the tutor should know which job is being selected.

Choosing the correct meaning among options is not the same as deciding whether a word fits a new sentence. A mixed vocabulary set is strongest when the response demand matches the capability the tutor wants to observe.

33. Primary Learners Need an Explicit Selection Language

Younger learners may experience a mixed set as “random questions”.

Make the hidden job visible.

Before you solve, tell me what kind of relationship you think this is and what clue helped you decide.

This gives children a vocabulary for selection without requiring adult jargon.

Over time, the spoken classification can become internal.

34. Secondary Learners Need Labels Removed Earlier

By Secondary school, learners should increasingly practise without chapter headings announcing the route.

This does not mean every worksheet should be mixed.

It means the learning programme should deliberately move between focused acquisition and unlabeled selection practice.

The learner should know the difference between “today I am stabilising a method” and “today I am practising how to choose among methods”.

35. JC and Advanced Learners Need Method Competition

At higher levels, more than one method may be valid.

The mixed-set question becomes:

Which route is valid?
Which is efficient?
Which is robust under the given conditions?
Which exposes the structure most clearly?
What assumption makes one route preferable?

Selection matures from “which method works?” to “which method should I choose here, and why?”

36. Class 0 · Homework Helper: Stop the Homework Folder From Doing All the Routing

A Class 0 tutor can use a tiny mixed set to help a learner organise independent homework.

Instead of saying, “Now do your fractions,” the tutor can place several familiar tasks together and ask the child to decide what kind of work each requires.

The goal is not advanced interleaving.

It is beginning to move task identification from the adult to the learner.

37. Class 1 · Explainer: Check Whether the Explanation Becomes a Selection Rule

An Explainer may teach the difference between ratio and percentage beautifully.

Do not finish with another labelled example.

Place two new questions side by side and ask which explanation applies to each.

A concept explanation becomes more useful when it can guide a future choice.

38. Class 2 · Drill Builder: Drill the Choice, Not Only the Procedure

A Drill Builder can accidentally produce fast dependence on worksheet order.

The repair is to alternate focused drilling with selection drilling.

Ten execution questions may be followed by ten rapid classifications in which the learner names the method and cue but does not calculate.

Fluency now includes routing.

39. Class 3 · Diagnostic Tutor: Use Mixing to Separate Competing Hypotheses

A Diagnostic Tutor may ask whether poor examination performance comes from weak knowledge or poor selection.

Test each method separately.

Then mix them.

If blocked execution is strong but mixed selection collapses, the evidence shifts toward discrimination and routing rather than basic content absence.

That is a much more useful diagnosis than “needs more practice”.

40. Class 4 · Route Designer: Decide When the Programme Should Become Mixed

The Route Designer controls the transition from acquisition to integration.

Too early and the learner is overwhelmed.

Too late and the learner becomes excellent only inside labelled chapters.

A strong route might look like:

teach → blocked stabilisation → changed question → contrast pair → small mix → cumulative mix → delay → timed mix → examination performance.

41. Class 5 · Performance Coach: Train Selection Under Real Constraints

The Performance Coach eventually needs the learner to choose under time, marks, fatigue and uncertainty.

But the coach should preserve diagnostic visibility.

When a timed item fails, ask:

Was the method never recognised?
Was the right method selected too slowly?
Did the student switch methods unnecessarily?
Did time pressure remove checking?
Did the learner know the route but abandon it because of perceived cost?

Performance coaching should not flatten every error into “careless”.

42. Class 6 · Learning Architect: Make Selection Part of the Learner’s Own Study System

A Learning Architect does not merely deliver mixed sets.

The learner is taught to create them.

Ask the student to choose four methods that are easily confused, write one fresh example of each, remove the labels, shuffle the order, attempt them after a delay and then analyse every method choice.

Now the student is learning to engineer their own selection practice.

43. Repair Mode: Mix Only Around the Repaired Boundary

In Repair mode, the tutor may have just fixed one fragile distinction.

Do not immediately throw the learner into a twenty-topic paper.

Mix the repaired skill with the nearest confusable alternative.

If the learner can now choose correctly, widen the set gradually.

44. Alignment Mode: Let School Work Supply Authentic Mixing

School homework, weighted assessments and revision papers naturally combine topics and question types.

In Alignment mode, the tutor can use this material as an authentic mixed environment.

The important move is not to pre-sort every question for the learner.

Let the student classify first. Then compare the school signal with tutor-designed diagnostic sets.

45. Frontier Mode: Mix Models That Can All Work

Strong learners need more than selecting the only correct method.

They can be asked to choose among several valid representations or solution routes.

Which route is shortest?

Which is clearest?

Which generalises?

Which is least fragile under algebraic complexity?

The mixed set becomes a model-selection environment.

46. A 3-Pax Class Makes Method Choice Visible

Three students can receive the same unlabeled problem and privately write:

their chosen method;
the cue they used;
their confidence;
their first move.

Only then do they discuss.

The tutor can now see three different internal routes before peer influence merges them.

This is one reason a deliberately small group can be powerful: comparison remains available without sacrificing individual diagnostic visibility.

47. Peer Explanation Should Follow Independent Selection

If one confident learner announces “chain rule” immediately, the other two students no longer have to solve the selection problem.

Use this sequence:

silent read → private method choice → first move → reveal → compare cues → solve → discuss.

Collaboration now adds reasoning rather than erasing evidence.

48. Three Students Can Receive Three Different Mixes

A mixed set does not have to be identical for the entire group.

Alicia may need ratio–percentage discrimination. Beatrice may need cause–evidence discrimination. Ciara may need algebraic route selection.

The shared classroom operation is still method selection.

The content can differ while the learning job remains common.

49. Learning, Studying, Teaching, Training and Improvement Are Not the Same Layer

The Mixed Set becomes clearer when we keep five layers separate.

Layer	Question
Learning	What durable capability changes inside the learner?
Studying	What actions does the learner perform to create or strengthen that change?
Teaching	What intervention helps the learner build the missing capability?
Training	What repeated practice conditions are deliberately arranged so future performance adapts?
Improvement	What evidence changes the next plan?

A mixed set can operate across all five, but it should not be confused with any one of them.

50. Learning: The Target Is Better Independent Selection

The learner has changed when they can later recognise the relevant problem structure, retrieve plausible methods, reject inappropriate ones and choose a route with less external help.

Completing a mixed worksheet is activity.

Becoming a better selector is learning.

This distinction belongs to the wider How Learning Works architecture.

51. Studying: The Learner Must Stop Over-Sorting Their Own Revision

Students naturally organise revision into comfortable chapters.

That is useful for acquisition.

But a study system that never recombines old topics may produce knowledge that is available only when the chapter is already known.

A mature study plan deliberately includes periods where the student closes the chapter map and lets the question itself reveal what knowledge should be retrieved.

See How Studying Works and How to Improve Studying.

52. Teaching: The Tutor Must Know When to Remove the Label

Good teaching supplies structure when structure is needed.

Good teaching also removes structure when the learner is ready to carry it.

The topic label, worked example, hint, method list and worksheet grouping are all candidates for fading.

That handover is part of How Teaching Works and How Independent Learning Works.

53. Training: Practice Must Resemble the Decision Problem That Matters Later

If the future examination requires the learner to identify the method independently, a training programme made entirely of labelled blocked sheets is missing part of the performance demand.

Training therefore becomes more representative over time.

Not immediately.

Not chaotically.

But deliberately.

The learner practises the decisions that later performance will actually require.

54. Improvement: The Set Must Change the Next Decision

A mixed set that produces a score and then disappears into a file has wasted much of its diagnostic value.

After the set, ask:

Which method families were confused?
Which cues were misread?
Which methods were selected correctly but executed poorly?
Which errors appeared only under mixing?
Which distinctions are now stable enough to expand?
Which weak method should leave the mix for focused repair?

Improvement is the loop that converts evidence into a changed plan.

55. Alicia: Ratio and Percentage Look Too Similar

Alicia scores above 85% on separate ratio and percentage worksheets.

In a six-question mixed set, she chooses percentage whenever she sees “increase” and ratio whenever two groups are named.

The tutor removes calculation for ten minutes.

Alicia classifies each item by relationship: part-to-whole, multiplicative comparison, rate per unit or percentage change.

Her procedures were not the first weak link.

Her routing rules were.

56. Beatrice: The Passage Changes Faster Than the Question Type

Beatrice handles cause questions well when the worksheet is labelled by question type.

In a mixed comprehension set, she answers an evidence question as though it were asking for cause.

The tutor asks her to underline the relationship requested before looking back at the passage.

Her retrieval of passage details is fine.

The selection error occurs earlier, in question interpretation.

57. Ciara: The Correct Method Appears Only After the Tutor Names the Family

Ciara sees a Mathematics problem and freezes.

The tutor says, “Think simultaneous equations.”

Ciara finishes accurately.

The tutor does not say, “Good, you know it.”

The evidence is more precise:

Execution is available after classification support. Independent method recognition is not yet reliable.

The next practice should target classification, not another explanation of elimination.

58. Denise: The Mix Exposes Overuse of the Most Recent Method

Denise has spent a week on differentiation.

In the next mixed set, she tries to differentiate two problems that should be handled with algebra and coordinate geometry.

This is not random carelessness.

The recently strengthened method has become too available relative to its selection boundary.

The repair is discrimination: what conditions must be present before differentiation is justified?

59. Emily: Mixed Practice Looks Worse Before It Looks Better

Emily drops from 90% on blocked sets to 68% on her first serious mixed set.

She thinks she has become worse.

The tutor explains that the task changed.

Yesterday, the worksheet selected the method.

Today, Emily had to do that work herself.

The lower score is not automatically good, but it is not automatically regression either. The tutor now tracks whether selection improves over repeated, delayed mixed sets.

60. Faith: Science Definitions Are Available but the Wrong One Wins

Faith can state diffusion, osmosis and active transport accurately when asked one by one.

In an unfamiliar diagram, she chooses diffusion because particles are moving.

The tutor asks which condition is discriminating: concentration gradient alone, water movement through a partially permeable membrane, or energy-requiring movement against a gradient.

The mixed set reveals that the definitions exist but the classification boundary is weak.

61. Failure Mode: Mixing Too Early

The learner cannot reliably execute any of the component methods.

The tutor mixes all of them anyway.

Every question becomes a compound failure.

Repair: isolate the weakest components, rebuild them, then re-enter a smaller mix.

62. Failure Mode: Random Mixing

The tutor shuffles unrelated questions and calls the result interleaving.

The learner switches topics, but there is little useful discrimination because the alternatives are too obviously different.

Repair: mix problem families whose boundaries are genuinely worth learning.

63. Failure Mode: The Method Is Still Hidden in the Instructions

The worksheet looks mixed, but each question begins with “Use the quadratic formula to…” or “Differentiate…”

Execution is being tested.

Selection is not.

That may be appropriate—but call it what it is.

64. Failure Mode: The Tutor Announces the Method Orally

The page is unlabeled.

The tutor says, “This one is like the ratio question we did earlier.”

The selection problem has just been solved externally.

If the learner is stuck, use a prompt ladder and record the smallest support that restores progress.

65. Failure Mode: Recording Only the Final Mark

A 70% mixed-set score can hide:

excellent selection with careless execution;
weak selection rescued by strong calculation;
one method family causing most errors;
correct methods chosen for superficial cues;
slow but accurate classification;
fast guessing that happens to work.

Record the route, not only the total.

66. Failure Mode: Treating the Lower Practice Score as Proof the Method Is Bad

Mixed practice often feels harder because it asks the learner to do more.

Research in Mathematics has repeatedly found cases where blocked practice produces stronger immediate performance while interleaved practice produces stronger later test performance.

That does not mean every low mixed score is desirable.

It means immediate fluency is not the only criterion.

67. Failure Mode: Never Returning to Focused Repair

A mixed set exposes factorisation as genuinely unstable.

The tutor keeps factorisation inside the mix for six more weeks.

That is not productive difficulty.

Remove the method temporarily, repair it in focused practice, then return it to competition.

68. Failure Mode: Mixing Too Many New Decisions at Once

The tutor adds new topics, unfamiliar wording, new notation, tighter time, new response format and reduced calculator access in one set.

The learner fails.

The evidence becomes difficult to interpret.

Increase representativeness progressively, not theatrically.

69. Failure Mode: Making Every Mixed Set Huge

A three-question set can be enough to expose one boundary.

A six-question set can test three method families.

Twenty questions are not automatically better.

Use enough items to answer the diagnostic question, then stop or move the route.

70. The Mixed-Set Design Board

Field	Tutor question
Target families	Which methods, concepts or reasoning jobs are competing?
Readiness	Can each family be executed when cued?
Discriminating feature	What difference should guide selection?
Surface control	Am I accidentally adding unrelated difficulty?
Label condition	Which headings, hints or method names are removed?
Selection response	Will the learner state the route before solving?
Execution response	Will the learner complete the full solution?
Delay	Immediate, next lesson or later return?
Time	Untimed or timed?
Success claim	What would strong performance actually justify?
Failure claim	What would the first breakdown make more plausible?
Next move	Expand the mix, delay, time, repair or advance?

71. A 90-Minute Tutor Session Using Mixed Practice

0–10 minutes: Return check. Retrieve two or three older methods without notes.
10–25 minutes: Focused repair. Rebuild one method that is not yet ready for competition.
25–40 minutes: Contrast pair. Compare two near-neighbour problem families and name the discriminating feature.
40–60 minutes: Small mixed set. Learner states method and cue before full execution.
60–70 minutes: Error classification. Separate selection, execution and checking errors.
70–82 minutes: Fresh unlabeled set. Reduce tutor prompts and change surface features.
82–88 minutes: Learner explanation. Student explains which cues mattered and which were misleading.
88–90 minutes: Handover. Decide what will return after delay and what the learner will practise independently.

The exact timings can change. The architecture is the point: focused work and mixed work are allowed to have different jobs inside the same lesson.

72. A Five-Minute Mixed-Set Routine

Choose three already-taught problem families.
Give one unlabeled item from each.
Ask the learner to name the relationship or method before solving.
Ask for one rejected alternative and why it does not fit.
Record whether the failure was selection or execution.

Five minutes can reveal more than another twenty minutes of routine blocked work when the question is whether the learner can choose.

73. A One-Minute Mixed-Set Routine

Do not solve these three yet. Tell me which method belongs to each one, what clue matters, and which nearby method would be wrong.

That single minute can make the invisible routing layer visible.

74. Twelve Questions Every Tutor Should Ask Before Giving a Mixed Set

Which component methods are already available?
Which two or three families are most worth discriminating?
What feature should trigger the correct choice?
What misleading surface feature might attract the wrong method?
Am I testing selection, execution or both?
Should the learner state the route before solving?
Which labels or prompts am I removing?
Is the reading or arithmetic load comparable across items?
How much delay should precede the mix?
Should this be untimed or timed?
What result would send a method back to focused repair?
What result would justify expanding the mix?

75. Twelve Questions Learners Can Ask Themselves

What is this question really asking me to find, explain or decide?
What relationship is present?
Which methods are plausible?
What condition makes one method fit?
What condition rules another method out?
Am I reacting to a keyword or to the actual structure?
What representation would make the structure clearer?
Have I chosen the method before beginning the calculation?
If I am stuck, is the problem selection or execution?
Could another valid method work?
What should I check after I finish?
What did this question teach me about how I choose?

76. Twelve Questions Parents Can Ask Without Turning Home Into Another Tuition Centre

Can my child do the method when the topic is named?
Can my child still choose it when the topic is not named?
Does revision contain any mixed practice?
Is the mix purposeful or simply random?
Are similar methods compared directly?
Does the tutor distinguish selection errors from calculation errors?
Are topic labels and hints gradually being reduced?
Does my child know why one method fits and another does not?
Is mixed practice introduced only after enough foundation is built?
Does the learner return to weak methods for focused repair when necessary?
Does mixed practice eventually become delayed and timed?
Is my child becoming better at choosing independently rather than merely completing more worksheets?

77. What the Research Supports

Research on interleaved practice gives useful support for the central claim of this volume, especially in Mathematics and category-learning tasks, while also warning against universal slogans.

Taylor and Rohrer reported that interleaving different kinds of Mathematics problems could reduce practice-session performance while improving performance on a later test, with error patterns suggesting better matching of problems to procedures. Rohrer, Dedrick, Hartwig and Cheung later ran a preregistered cluster-randomized classroom trial across 54 seventh-grade Mathematics classes. Students receiving greater interleaving emphasis outperformed the blocked-practice group on an unannounced test one month later.

The What Works Clearinghouse reviewed an earlier randomized study of interleaved Mathematics practice and rated it as meeting its standards without reservations, while correctly noting that one study does not represent an entire evidence base.

Reviews of interleaving also suggest an important mechanism: alternating similar categories or problem families can draw attention to discriminating differences. That is especially relevant to tutoring because the learner often does not need another twenty examples of Method A. The learner needs to know why this problem is A rather than B.

78. What the Research Does Not Justify

The evidence does not justify saying that all mixed practice is always better than all blocked practice.

It does not justify mixing methods before they are understood.

It does not prove that a random worksheet is educationally sophisticated merely because topics alternate.

It does not tell every tutor the same ideal ratio of blocked to mixed questions.

And it does not erase subject knowledge. A tutor still needs to know what distinctions matter inside Mathematics, English, Science or any other discipline.

The evidence supports a more careful operating principle:

When future performance requires learners to discriminate among plausible alternatives, practice should eventually require them to discriminate among plausible alternatives too.

79. Evidence and Further Reading

80. The Mixed-Set Operating Cycle

Confirm that component methods can be executed when cued → choose the smallest useful set of competing families → identify the discriminating feature → remove unnecessary topic labels and method prompts → let the learner classify before solving → record the selected route and cue → observe execution separately → compare plausible alternatives → repair weak boundaries or weak procedures differently → return after delay → expand the mix gradually → add time only when untimed selection is understood → use school and examination work as external return evidence → hand more routing responsibility back to the learner.

This is how a mixed set becomes more than variety.

It becomes a test of judgement.

81. Final Compression

The learner can solve ratio questions.

The learner can solve percentage questions.

The learner can solve rate questions.

Good.

Now remove the headings.

Put the methods beside one another.

Ask the learner to decide what relationship exists before touching the calculation.

Watch the first move.

Watch the rejected alternatives.

Watch whether the student reacts to a keyword or reads the structure.

Separate choosing from doing.

Separate hesitation from ignorance.

Separate a wrong route from a poor calculation.

Do not mix too early.

Do not mix randomly.

Do not keep every method permanently isolated.

Do not let the tutor’s voice keep naming the answer family.

Build the method.

Stabilise the method.

Change the surface.

Then place the method among competitors and see whether the learner can still find it.

The mixed set asks whether knowledge can select itself when the worksheet stops announcing where it belongs.

When the learner can do that, practice has moved beyond repetition.

When the learner can explain why one route belongs and another does not, classification has become more structural.

When that choice survives delay, unfamiliar wording and eventually the clock, the learner is approaching real performance readiness.

That is the discipline of the mixed set.

That is Tutor Handbook Volume 0020.

Next in the series: The Tutor Handbook Vol No.0021 | The Timed Set — How a Tutor Adds the Clock Without Confusing Speed With Learning.

Small Group Tutorials