A student can solve a numerical puzzle accurately and still struggle to judge an argument. Another can discuss evidence fluently but overlook a timetable condition. Cambridge AS Thinking Skills brings these demands into the same qualification, so preparation needs to make both kinds of thinking visible.
For Cambridge Thinking Skills 9694 in 2026–2027, prepare both AS papers deliberately: represent and check unfamiliar problems for Problem Solving, and examine evidence and reasoning for Critical Thinking. Diagnose the first unsupported decision in each strand, practise it on a small task, then return to a fresh complete question.
This guide uses original fictional schedules, evidence summaries and arguments. They are teaching exercises, not Cambridge examination questions or model submissions. The official sources were checked on 22 October 2026. The broader purpose comes from The Gold Standard of Education: learning transfers into life: understanding should remain available when the question no longer looks familiar.
Find your next learning step
Identify the two AS papers, diagnose the kind of reasoning needed, then practise original problems and a separate evidence-and-argument case.
ROUTE 1 · CHAPTERS 1–3
Map the two-paper task
Keep qualification, reasoning strands and diagnosis distinct.
ROUTE 2 · CHAPTERS 4–6
Work through problem solving
Represent a schedule, check constraints and justify a solution.
ROUTE 3 · CHAPTERS 7–9
Work through critical thinking
Evaluate evidence, trace an argument and construct a bounded response.
ROUTE 4 · CHAPTERS 10–12
Repair and test independence
Choose targeted practice, transfer and an independent check.
ROUTE 5 · CHAPTERS 13–15
Plan and verify preparation
Use practical roles, useful FAQs and official sources.
Full chapter index · Independent learning check · Education parent guide
Full chapter index
Map the two-paper task · Chapters 1–3
Work through problem solving · Chapters 4–6
Work through critical thinking · Chapters 7–9
Repair and test independence · Chapters 10–12
Plan and verify preparation · Chapters 13–15
CHAPTER 1 OF 15 · Map the two-paper task
1. Identify both AS papers and the examination cycle
The 2026–2027 syllabus identifies Paper 1 as Problem Solving, lasting 1 hour 30 minutes, and Paper 2 as Critical Thinking, lasting 1 hour 45 minutes. Each carries 50 marks. The AS route includes both; preparing only the strand a student enjoys leaves a substantial part of the task unattended.
The A Level has additional papers. This article focuses on AS preparation rather than treating the AS and full A Level as identical. Confirm the actual entry with the school. The qualification is also distinct from a university admissions test with a similar emphasis on reasoning.
The qualification overview lists separate 2026–2027 and 2028–2030 syllabuses. Use the examination year, rather than the publication date of a textbook, to choose the governing document. A useful older exercise may still practise a skill without reproducing the current paper exactly.
Make a two-column preparation record. In one column, collect evidence of solving constrained problems. In the other, collect evidence of evaluating and constructing arguments. An overall practice total can hide an uneven profile; the columns make the next teaching decision easier to see.
Do not fill the record only with topic names. Record what the learner can do without help: identify a missing condition, justify a complete search, distinguish a claim from its support, or explain why an objection matters.
The starting question is therefore practical: which kind of reasoning is currently unstable, and what independent work shows it? Once that is clear, practice can become purposeful instead of a general instruction to do more thinking questions.
CHAPTER 2 OF 15 · Map the two-paper task
2. Distinguish a constrained solution from an evidence judgement
A constrained problem defines conditions that an acceptable answer must satisfy. A proposed timetable either meets a stated finish time or does not. An evidence judgement asks how strongly particular information supports a conclusion, often while uncertainty or competing explanations remain.
Both require care, but the checking methods differ. In a schedule, one overlooked condition can make the entire proposal infeasible. In an argument, one relevant example may support a claim without establishing it conclusively. Treating every judgement as a calculation, or every calculation as a matter of opinion, creates confusion.
Consider two fictional questions about an evening learning event. The first asks whether three sessions fit a room booking under specified conditions. The second asks whether attendance comments show that the event should always be held online. The same broad setting contains two different intellectual jobs.
For the first, extract the durations, ordering requirements and time limits. For the second, inspect who commented, what they actually said, what conclusion is proposed and which alternatives remain possible. A beautifully drawn timetable cannot settle the second question.
| Reasoning task | What the answer needs | What can go wrong |
|---|---|---|
| Find a feasible arrangement | Every stated condition satisfied | One attractive arrangement checked only partly |
| Show a search is complete | A reason no relevant case was omitted | Several examples mistaken for all possibilities |
| Evaluate a claim | Support matched to its scope | A narrow observation turned into a universal conclusion |
| Construct an argument | Reasons connected to a defensible position | An opinion followed by unrelated facts |
This is an original preparation map, not an official marking grid. It helps the student choose a checking method before committing effort.
The two strands can inform one another. Careful condition reading helps argument interpretation, while awareness of assumptions helps modelling. The connection is useful when it preserves the distinct job of each task rather than reducing both to one memorised formula.
The IB CP guide to framing a career-related ethical dilemma explores a further distinction: evidence can clarify a decision while competing values still require justification. It offers a changed learning context for noticing why not every important question can be settled like a schedule with fixed constraints.
CHAPTER 3 OF 15 · Map the two-paper task
3. Diagnose the first wrong decision before increasing difficulty
Give a short unfamiliar task and ask the learner to think aloud just enough to reveal their decisions. Stop at the first unsupported step. That step may concern reading, representation, calculation or inference. The final answer alone often conceals where the difficulty began.
In a problem-solving task, a student might copy the correct numbers but attach a deadline to the wrong event. More arithmetic practice will not repair that reading error. Another may understand every condition but search cases haphazardly, leaving an organisation problem. A third may have a sound method and make one calculation error.
In critical thinking, a learner may identify the topic but mistake a reason for the main conclusion. Another may recognise the conclusion while accepting evidence from an unrepresentative group. A third may notice a weakness but fail to explain its effect on the argument.
Use a second small task to test the diagnosis. If the student again attaches a condition to the wrong event, practise that relationship directly. If the second task is secure, the first error may have been a local misreading rather than a stable weakness.
The BTT guide to choosing a diagnostic representation owns the broader Mathematics teaching question. Here its useful connection is changing the form of a task to expose an interpretation, while keeping the Cambridge course requirements separate.
Ask what level of help was needed. A correct answer after the teacher names every relevant condition is different evidence from a correct independent solution. Record the assistance honestly so the next practice task can reduce it.
Avoid turning an error into an identity. “I have not yet organised all possible orders” gives the learner a task. “I cannot do logic” gives them a discouraging label. A useful diagnosis should be specific enough to repair and temporary enough to be tested again.
CHAPTER 4 OF 15 · Work through problem solving
4. Represent a fictional scheduling problem clearly
Use this original preparation case. A room is available from 09:00 to 10:30. Session A lasts 30 minutes, B lasts 20 minutes and C lasts 25 minutes. Allow a five-minute changeover between sessions. Sessions cannot overlap. A must occur before B, C must finish by 10:00, and B cannot start before 09:40.
First separate the quantities from the relationships. The durations total 75 minutes. Two changeovers add 10 minutes, giving 85 minutes of occupied sequence time. The room is available for 90 minutes. That establishes a necessary capacity check, but it does not yet establish a feasible order.
A learner who stops at 85 being less than 90 has ignored the ordering and timing conditions. The arithmetic is correct while the conclusion is incomplete. This is a useful diagnostic distinction: a correct subtotal can become the first unsupported step when it is treated as the whole answer.
Write a condition ledger before arranging sessions. A is before B. C ends no later than 10:00. B begins no earlier than 09:40. Every interval remains within the booking, and each transition needs its changeover.
Then choose a representation that keeps the relevant relationships visible. A timeline can show start and finish times. A table can record candidate orders and the condition each fails. A list alone may be sufficient if it remains clear and complete.
The Sengkang guide to learning by drawing explains a representation as something to inspect and revise. In this case, the timeline must represent the actual conditions. A neat drawing with an omitted changeover is still a faulty model.
Before calculating further, ask the learner to explain what each written time refers to. Does 10:00 constrain C’s start or finish? Does before mean the sessions can overlap? The problem states no overlap. Small words determine the mathematical relationship.
CHAPTER 5 OF 15 · Work through problem solving
5. Check candidate schedules against every condition
Try order A, C, B, beginning at 09:00. A runs until 09:30. After a changeover, C runs from 09:35 to 10:00. A second changeover allows B to run from 10:05 to 10:25. The arrangement fits the room booking and satisfies the stated ordering and timing restrictions.
Now test A, B, C. A can finish at 09:30, but B cannot begin before 09:40. B then ends at 10:00, and the next changeover places C after 10:05. C cannot finish by its 10:00 deadline. The order fails even though the total session duration looked manageable.
Try C, A, B. C runs from 09:00 to 09:25, A from 09:30 to 10:00, and B from 10:05 to 10:25. This order also satisfies the stated conditions. The problem has not established a unique feasible order.
| Candidate order | What the timing shows | Decision under the stated conditions |
|---|---|---|
| A, C, B | C ends at 10:00 and B at 10:25 | Feasible |
| A, B, C | C begins after its finish deadline | Infeasible |
| C, A, B | C ends early and B starts after 09:40 | Feasible |
The table contains worked consequences of invented conditions, not actual event data. It illustrates why each candidate needs a complete check.
The next question depends on the task wording. If asked for one feasible schedule, a fully checked example may answer it. If asked for every feasible order, a search-completeness argument is needed. If asked for the earliest finish or least waiting, the learner must compare a defined objective as well.
Do not answer a harder or different question simply because it seems more impressive. Identify what the requested result means, then provide the evidence necessary for that result.
There are six possible orders of three distinct sessions before applying conditions. Because A must precede B, only A–B–C, A–C–B and C–A–B remain. The three orders with B before A can be excluded immediately. The previous chapter checked the remaining orders.
That argument is stronger than saying the student tried many arrangements. It explains the search space and why no ordering case was omitted. The representation is doing intellectual work by making completeness inspectable.
Be precise about what has been exhausted. The argument covers session orders under the stated conditions. It does not list every possible start time, because some schedules could include additional waiting within the available window. If the question asks about exact timings rather than order alone, further reasoning may be required.
Change one condition to test the model. Suppose C must now finish by 09:50. The earlier A–C–B timing no longer works because C finishes at 10:00. C–A–B still meets the revised condition. The learner should update the relevant check rather than repeat the original conclusion.
Now ask what information could matter if the problem became more realistic. A real event might involve different rooms or variable changeover time. Those features should not be invented inside the original exercise. They belong to an explicitly changed problem.
This distinction protects both mathematical accuracy and modelling judgement. Solve the stated model faithfully, then explain how a different model would need different information. Do not silently repair an inconvenient condition by importing an assumption from ordinary life.
Finish with a short justification that another reader can follow. A correct schedule without visible checks may hide whether the student understood the constraints or found the arrangement accidentally. Clear working gives the learner a way to inspect their own reasoning before depending on external feedback.
CHAPTER 7 OF 15 · Work through critical thinking
7. Evaluate a fictional evidence set for Paper 2 preparation
Turn to a separate original case. An organiser receives 18 positive comments about moving a learning event online. All comments come from people who chose to answer an online feedback form. The organiser concludes that every future event should be online because all participants prefer it.
Begin with the actual evidence. Eighteen respondents expressed positive views. The case does not state how many people attended, how many received the form, how non-respondents felt or whether respondents were asked to compare all relevant alternatives.
The conclusion extends beyond that evidence. It concerns all participants, every future event and an exclusive delivery format. Each expansion needs support. The learner should identify the particular gap rather than dismiss the entire feedback collection as useless.
Positive comments may provide valuable information about those respondents’ experiences. They might identify features worth preserving, such as easier access from home. They do not automatically establish that the same arrangement is best for everyone or for every kind of event.
Ask what additional evidence would be relevant. A response count alone is not enough; the selection process and the question asked matter. Evidence from participants unable or unwilling to use the online form may change the interpretation. Information about practical learning activities may affect whether format preferences transfer across events.
The task is not to invent a survey result supporting the opposite view. It is to explain how the available evidence should limit the conclusion. A more cautious recommendation might retain online options while investigating which activities and participants benefit under which conditions.
This is a critical-thinking exercise, not a reported study. Keep that label visible. The learner can reason carefully about an evidence gap without pretending to possess the missing evidence or knowing in advance what a better investigation would find.
The Cambridge History guide to comparing sources as evidence develops a related question in a different discipline: what can this particular record establish about the claim being tested? Historical comparison adds its own contextual knowledge and examination demands; the shared habit is keeping evidential reach visible.
CHAPTER 8 OF 15 · Work through critical thinking
8. Reconstruct an argument before judging it
An argument connects reasons to a conclusion. A passage may also contain background, examples, explanations or rhetorical emphasis. If the student labels every sentence as a reason, they may miss the structure they need to evaluate.
In the fictional event argument, the proposed conclusion is that every future event should be online. One stated reason is the positive feedback. An unstated assumption is that the respondents represent all participants in a way relevant to the future decision.
Another possible assumption concerns the relationship between preference and suitability. Even if many participants prefer online delivery, the writer may still need to consider the purpose of particular activities. Preference is relevant, but it is not necessarily the only criterion for the recommendation.
Use a plain-language reconstruction. What is the writer asking me to accept? What is offered in support? Which connection must hold for that support to reach the conclusion? The reconstruction should represent the argument fairly before criticism begins.
Avoid guessing motives. The organiser may be enthusiastic, mistaken, selective or simply writing briefly. Unless the evidence supports a claim about intention, focus on the reasoning’s structure and support. An unsupported accusation about the writer creates a new weakness in the student’s answer.
Then evaluate the effect of the identified assumption. If the respondents are unrepresentative, the universal preference claim becomes weaker. If online format works differently across activities, the recommendation for every event needs qualification. Explain that consequence explicitly.
The LNAT guide to a defensible conclusion concerns a different assessment. Its useful connection is keeping a conclusion within the support the argument provides. Cambridge Thinking Skills preparation still needs its own current tasks, command words and marking guidance.
CHAPTER 9 OF 15 · Work through critical thinking
9. Construct a response that uses evidence proportionately
When constructing an argument, the student needs to do more than list weaknesses in someone else’s reasoning. They should choose a position and show how the available evidence supports it, while acknowledging relevant uncertainty.
For the fictional event case, a student might argue for retaining an online option while gathering broader evidence before choosing an exclusive format. The positive comments support keeping something that respondents valued. The selection limitation weakens the case for making that option compulsory for everyone.
That recommendation is conditional, but it is still a decision. It identifies what should happen now and what remains unresolved. A vague statement that both sides have points would leave the organiser without a reasoned next step.
Use the evidence directly. Explain which observation supports which part of the recommendation. If the student introduces a claim that online events always cost less or always improve learning, they have added an unsupported factual assertion. A familiar claim is not automatically available evidence.
Consider a serious objection. Keeping multiple formats may create practical demands. The response should acknowledge that possibility and explain how it affects the proposal under the information given. It should not invent a budget or staffing arrangement to make the objection disappear.
The ESS guide to evaluating an environmental strategy offers another evidence-and-recommendation case. Its relevant connection is separating recorded information, inference and value judgements. The qualifications and assessment instruments remain distinct.
After writing, ask the student to underline the recommendation and identify the sentence that provides its strongest support. If that connection is unclear, improve the reasoning before adding more elaborate language. Economy is useful when it removes repetition while preserving the relationship the reader needs to inspect.
CHAPTER 10 OF 15 · Repair and test independence
10. Repair the strand that actually needs attention
A student who performs well on constrained numerical problems may need targeted work on evidence and argument. A student who writes persuasively may need targeted work on systematic case checking. Equal practice time is not automatically the best response to an unequal learning profile.
For Problem Solving, classify the first error. Was relevant information missed? Was a condition attached to the wrong quantity? Was a search incomplete? Was the method sound but the calculation inaccurate? Choose a short practice task that isolates that difficulty.
For Critical Thinking, distinguish interpretation from evaluation. A learner who misidentifies the conclusion needs work on argument structure before being asked for a sophisticated criticism. A learner who sees the conclusion but overgeneralises from a sample needs an evidence-scope task.
Use a fading sequence. First discuss a complete worked case. Next remove some of the prompts. Then change the setting and ask for an independent response. Finally return after a delay. A method used only while the example is open has not yet become dependable independent preparation.
Keep corrections small enough to understand. Rewriting an entire answer for the student can produce a polished page without revealing whether the learner can make the central decision. Ask for a corrected step and a reason, then test that reason in a new task.
Do not let one successful exercise erase breadth. A learner who has repaired scheduling still needs the range of problem-solving work in the syllabus. A learner who understands sample limitations still needs to analyse other forms of reasoning and evidence.
Review the active error record periodically. Retire a priority when it survives independent retesting, and identify the next consequential gap. The goal is a changing pattern of decisions, not an ever-growing collection of mistakes that leaves the student feeling permanently behind.
CHAPTER 11 OF 15 · Repair and test independence
11. Change the context while preserving the reasoning test
Replace the sessions with a fictional delivery schedule. Three tasks require different durations, one must precede another and a particular task has an early deadline. The surface story changes, while the need to represent time and check all conditions remains.
Do not give the student the old table immediately. Ask which representation would make the new problem easier to inspect. If they choose a timeline, they should explain what it shows. If they choose a case table, they should explain how it prevents omissions.
Then change a deeper feature. Suppose two tasks can run simultaneously with separate resources. The earlier assumption of a single non-overlapping sequence no longer applies. Reusing the old method without adjustment would be a transfer failure, even if the numbers looked familiar.
For the critical-thinking strand, replace online event comments with a fictional collection of reviews of a new study app. Ask whether the reviews support a claim about every learner. The same sampling questions may help, but the specific population and outcome have changed.
Now alter the evidence. Suppose the reviews come from a defined random sample with a high response rate. The learner should not repeat self-selection as a criticism when it is no longer part of the case. They need to inspect the actual evidence rather than retrieve a favourite objection.
Transfer therefore includes recognising when a familiar criticism or method does not apply. That judgement is more demanding than noticing a superficial resemblance between two tasks.
After each changed case, ask what stayed stable and what required revision. The answer should name a relationship: checking every constraint, justifying a complete search, matching a claim to its sample, or separating preference from the criterion for a decision. Those relationships can travel while the specific answer changes.
CHAPTER 12 OF 15 · Repair and test independence
12. Try an independent two-strand check
First try this original scheduling task. A room is available for 70 minutes. X lasts 20 minutes, Y lasts 15 and Z lasts 20. Two five-minute changeovers are needed. X must precede Y, and Z must finish within the first 30 minutes. Sessions cannot overlap and the first session starts when the booking begins.
Before reading further, identify a feasible order and explain why the simple total-time calculation is insufficient. Then decide whether you have checked every order or only found one.
The occupied sequence totals 65 minutes. Z–X–Y is feasible: Z ends at minute 20, X runs from minute 25 to 45, and Y from minute 50 to 65. An order beginning with X cannot place Z after X and still finish Z within 30 minutes, because a changeover and Z’s duration would take it to at least minute 45.
The reasoning identifies why Z must come first under these conditions. With X before Y, the remaining order is fixed. That is a completeness argument about orders, not merely a successful example.
Now try a separate evidence task. Six students who attended an optional revision session say it was helpful. A tutor concludes that all students should attend the same session every week. What does the evidence support, and where does the conclusion extend beyond it?
A useful answer recognises positive reported experiences from six attendees while questioning the expansion to all students and every week. The participants selected themselves into attendance, and the case gives no evidence about others’ needs or repeated sessions. Those limits do not make the six comments worthless.
Finally, explain why the two answers were checked differently. The schedule required satisfaction of explicit conditions. The recommendation required a proportionate judgement about evidence and scope. Being able to state that difference is part of preparing both AS strands independently.
CHAPTER 13 OF 15 · Plan and verify preparation
13. Give students, parents and teachers practical roles
Students can begin a week by choosing one target in each strand. For example, check a complete set of orders in Problem Solving and explain an assumption’s effect in Critical Thinking. Keep the tasks short enough to reveal decisions, then include complete current-paper practice when the underlying work is more secure.
After practice, record the first unsupported step rather than only the final mark. Write the correction in plain language and identify a changed task that could test it. A useful record makes the next session easier to choose.
Parents can ask the student to explain a decision without taking over the solution. Which condition rules this order out? What does this evidence actually establish? Which part of the conclusion remains uncertain? These questions support clarity without supplying an answer.
Teachers and tutors can compare the two strand profiles and avoid assuming that fluent speech means sound reasoning or that fast calculation means complete problem solving. Ask for visible checks and proportionate explanations.
Use timed work for a purpose. It can reveal pacing and integration problems, but it is a blunt tool for repairing a basic misunderstanding. If the learner cannot distinguish a start-time condition from a finish-time condition, another full timed paper may simply rehearse the same error under pressure.
Preserve rest and variation. Preparation should not become an endless run of near-identical puzzles whose familiarity is mistaken for progress. A fresh representation, unfamiliar setting or different evidence structure can provide a more informative check.
Review independence after a delay. Can the learner identify the task type, choose a method and justify an answer without the earlier worked example? That gives the family and teacher a more useful picture than the number of pages completed during a heavily supported session.
CHAPTER 14 OF 15 · Plan and verify preparation
14. Questions about preparing both AS papers
Can I prepare only the strand I find easier?
That leaves part of the AS task unattended. Use the confirmed paper entry and current syllabus to plan coverage, then adjust practice according to the student’s actual profile. Strength in one strand does not establish readiness in the other.
Is a correct answer enough evidence of a sound method?
Not always. A correct answer can result from an incomplete search or a lucky guess. Ask the student to show the relevant checks. Clear working also helps reveal which part of a method needs repair when the result is wrong.
Must every problem use a diagram?
No. Choose a representation that exposes the relevant relationships. A small table or careful list may be better for some tasks. The diagram is useful when it helps reasoning, not because every answer needs visual decoration.
Should I reject evidence whenever it has a limitation?
No. Explain what the evidence can still support and how the limitation affects a particular conclusion. A narrow observation can remain useful when its scope is respected.
Is Critical Thinking just expressing a balanced opinion?
No. The student needs to inspect reasons and evidence and reach a justified position. Listing opposing views without explaining their support or consequences leaves the central judgement unfinished.
Can I use a 2028 resource for a 2027 examination?
Some underlying exercises may be useful, but the current governing syllabuses are separate. Check the applicable year before using a resource as a paper simulation or relying on its terminology and assessment details.
Does this qualification guarantee university admission?
This guide makes no such claim. Entry requirements and decisions belong to the receiving institution and applicant category. The article concerns learning and preparation for the named Cambridge qualification.
CHAPTER 15 OF 15 · Plan and verify preparation
15. Official sources and the next independent task
The principal sources checked on 22 October 2026 are Cambridge’s Thinking Skills 9694 overview and 2026–2027 syllabus, version 2. Use the complete current syllabus, official assessment materials and school entry information for formal preparation. This article does not reproduce the whole course or an official marking instrument.
All session durations, room bookings, comments, review counts and arguments here are original fictional teaching material. Their numerical conclusions follow from the stated assumptions. They are not measured learning outcomes, real participant research or Cambridge specimen questions.
The suggested diagnosis and practice sequence is a teaching approach. It does not guarantee a mark or replace subject coverage. A learner still needs varied tasks across both strands and accurate attention to the command and evidence in each actual question.
For the next session, choose one short constrained problem and one short evidence argument. Before answering, identify what would count as a successful response in each. After answering, inspect the first unsupported decision and choose a changed task that tests it.
Keep the distinction clear when reviewing progress. More confident calculation should be supported by complete checks. More fluent discussion should be supported by accurate interpretation and proportionate inference. Confidence becomes useful when the reasoning underneath it is visible.
Return to the Education parent guide for the wider aim. Preparing both AS papers can develop more than examination familiarity when the student learns to recognise the job, select a method, check its limits and carry the reasoning into an unfamiliar situation.
Contents · Previous chapter · Continue with the Education parent