BTT Mathematics / Primary Mathematics Learning Hub / Approximation & Uncertainty
Primary Mathematics: Conditional Decision Making, Two-Way Tables and Trade-Offs | Worked Learning Guide
A decision can change when the condition changes. The overall percentage may favour one option while a particular subgroup, threshold or constraint points somewhere else.
Many everyday decisions use data conditionally. “What fraction of all pupils walk?” is different from “Among Primary 6 pupils, what fraction walk?” “Which product has the higher overall rating?” is different from “Among low-price options, which has the better reliability score?” The denominator changes with the condition.
This guide develops simple conditional reasoning through two-way tables, subgroup percentages, joint counts, marginal totals, thresholds, trade-offs and transparent decision rules. It is an upper-Primary enrichment and transition bridge. Formal conditional-probability notation is not required here.
Two-way tables · Conditional percentages · Compare conditions · Decision rules · Trade-offs · Limits · 24 questions · Worked solutions
1. A two-way table records two variables at once
| Walk | Other transport | Total | |
|---|---|---|---|
| Primary 5 | 12 | 18 | 30 |
| Primary 6 | 18 | 12 | 30 |
| Total | 30 | 30 | 60 |
Variables:
- year group: P5/P6,
- transport: walk/other.
The interior cells are joint counts: they satisfy both a row and column condition.
2. Marginal totals describe one variable without the other condition
Overall walkers=30.
Overall pupils=60.
Overall walking percentage=30/60=50%.
This ignores year group.
3. Conditional percentages change the denominator
Among P6 pupils,18 of30 walk.
P6 walking percentage=18/30=60%.
Among P5 pupils,12/30=40%.
The overall50% is neither subgroup’s percentage.
Worked example: reverse condition
Among walkers, how many are P6?
18 of30 walkers are P6 →60%.
This happens to match the previous percentage because the row/column totals are equal in this constructed example. In general, reversing the condition changes the denominator and need not give the same percentage.
4. “Among A” tells you the denominator
Question: “Among pupils who take the bus, what percentage are P5?”
Denominator = all bus pupils, not all P5 pupils and not all pupils.
Underline the conditioned group before calculating.
5. Joint percentage uses the grand total
From the table, P6 walkers=18 out of60 pupils.
Joint percentage=30%.
Conditional P6 walking percentage=18/30=60%.
The same cell count18 supports different percentages because the denominator answers a different question.
6. Never compare percentages with hidden denominators
“60% walk” is incomplete unless the reader knows 60% of which group.
Write “60% of P6 pupils in this sample walk.”
Precision prevents accidental denominator switching.
7. Overall patterns can hide subgroup patterns
Suppose School A has 80% success overall and School B70%. If the groups contain different mixtures of easier and harder tasks, subgroup comparisons may tell a different story.
At this level, the key lesson is simple: overall percentages combine groups. If the group composition differs, inspect the subgroup data before making a broad comparison.
8. Compare like subgroups
Option A performs well in dry weather; Option B performs well in rain.
If the decision is for a rainy-day route, the relevant conditional data are rainy-day outcomes, not only overall averages.
The condition should match the decision context.
9. A condition can be a threshold
Three products:
| Product | Price | Reliability score |
|---|---|---|
| A | $18 | 92 |
| B | $15 | 85 |
| C | $12 | 75 |
Decision rule: budget≤$16 and reliability≥80.
B qualifies. C fails reliability; A fails budget.
Conditional decision making often means filtering by constraints before comparing remaining options.
10. State the decision rule before choosing
Without a rule, “best option” may be ambiguous.
Possible rules:
- lowest cost under a minimum quality threshold,
- highest score under a fixed budget,
- shortest time while meeting safety constraints,
- largest capacity that fits a space.
Different rules can produce different winners.
11. Thresholds create feasible and infeasible options
If delivery must arrive within30 minutes, a 25-minute option is feasible and a35-minute option is not, even if the latter is cheaper.
Apply hard constraints before soft preferences.
12. Multiple conditions can be combined
Choose a container with:
- capacity≥2 L,
- height≤30 cm,
- cost≤$12.
A table allows each option to be checked against all three conditions.
Only after feasibility is established should remaining options be compared by preference.
13. Trade-offs occur when one option is not best on every measure
Option A: cheaper but slower.
Option B: faster but more expensive.
There is no universal best until the decision maker states how cost and time are valued.
Mathematics can make the trade-off explicit without pretending to supply the preference.
14. Use difference or ratio to quantify trade-offs
A costs$20 and takes30 min.
B costs$26 and takes20 min.
Choosing B costs$6 more and saves10 min.
This converts a vague trade-off into two quantities.
15. Unit-rate comparisons can help
Pack A: 6 items for$18 →$3 each.
Pack B: 10 items for$28 →$2.80 each.
B has lower unit price, but a buyer needing only6 items may still consider total spending and waste.
Unit price is one decision measure, not the entire decision.
16. Weighted scoring is a model, not a fact
Suppose a classroom decision assigns:
- cost weight40%,
- speed weight60%.
A weighted score reflects those chosen priorities.
Change the weights and the ranking may change.
This is an enrichment-level example of how model assumptions affect conclusions.
17. Sensitivity analysis asks whether the decision is stable
If Option A wins only when cost weight is41% rather than40%, the decision is sensitive.
If A wins across a wide range of reasonable weights, the decision is more robust.
At Primary transition level, this can be explored by trying a few alternative rules rather than formal optimisation.
18. Two-way tables support conditional evidence
| Passed | Not passed | Total | |
|---|---|---|---|
| Method A | 36 | 24 | 60 |
| Method B | 42 | 28 | 70 |
Pass percentage A=36/60=60%.
Pass percentage B=42/70=60%.
B has more passes by count because its group is larger, but equal pass proportion.
19. Conditional percentages should not be mistaken for causal effects
If one subgroup has higher success, the table alone may not show why.
Other differences between groups may matter.
Conditional comparison is evidence description, not automatic proof of cause.
20. Missing categories can distort decisions
A table comparing cost and speed but omitting reliability may favour an option that fails frequently.
A decision table is only as complete as the criteria included.
Ask which important variables are missing.
21. Unknown data should remain unknown
If reliability score for Product D is missing, do not replace it with zero unless zero is actually the meaning of missing data.
Mark unknown and decide whether the missing information prevents a fair comparison.
22. Decisions depend on model assumptions
Changing:
- the subgroup considered,
- the threshold,
- the weights,
- the time period,
- the data source
can change the result.
A transparent decision explains the rule as well as the chosen option.
23. Common conditional-decision errors
- Using the grand total when the question says “among” a subgroup.
- Reversing the condition without changing denominator.
- Comparing raw counts from unequal group sizes.
- Choosing “best” before defining criteria.
- Ignoring hard constraints.
- Treating weighted scores as objective truth rather than a model.
- Claiming cause from subgroup percentages.
- Filling missing data with invented values.
24. A conditional-decision protocol
- Condition: Which subgroup or constraint is active?
- Denominator: What is the relevant reference group?
- Filter: Which options satisfy hard constraints?
- Compare: Which quantities distinguish the remaining options?
- Trade-off: What is gained and lost?
- Rule: What priorities or weights are being used?
- Sensitivity: Would a reasonable rule change alter the choice?
- Limit: What data or variables are missing?
25. Practice: 24 original questions
Questions 1–8: Two-way tables
Use this table:
| Walk | Other | Total | |
|---|---|---|---|
| P5 | 12 | 18 | 30 |
| P6 | 18 | 12 | 30 |
| Total | 30 | 30 | 60 |
- Find overall walking percentage.
- Among P5, find walking percentage.
- Among P6, find walking percentage.
- What percentage of all pupils are P6 walkers?
- Among walkers, what percentage are P6?
- Explain why Question3 and Question4 use different denominators.
- Write one sentence correctly interpreting the60% in Question3.
- Create one reversed-condition question from the table.
Questions 9–16: Constraints and trade-offs
- Products A($18,92), B($15,85), C($12,75). Choose if budget≤$16 and reliability≥80.
- Which condition eliminates A?
- Which condition eliminates C?
- Option A costs$20 and takes30 min; B costs$26 and20 min. Quantify the trade-off.
- Pack A6 for$18; Pack B10 for$28. Find unit prices.
- Which has lower unit price?
- Give one reason Pack A might still be chosen.
- Create a three-condition feasibility rule for choosing an object.
Questions 17–24: Evidence and decision limits
- Method A36/60 pass; Method B42/70 pass. Compare proportions.
- Which method has more passes by count?
- Why does more passes not imply higher pass rate?
- Why does a higher conditional success percentage not automatically prove cause?
- A decision table omits reliability. What risk does this create?
- One product has missing cost data. Should missing be treated as zero? Explain.
- Explain how changing a hard threshold can change the chosen option.
- Create a two-way table or multi-criteria decision problem and state one model assumption.
26. Worked solutions
1. 30/60=50%. 2. 12/30=40%. 3. 18/30=60%. 4. 18/60=30%.
5. 18/30=60%. 6. Question3 conditions on all P6 pupils; Question4 uses all pupils. 7. “60% of P6 pupils in this sample walk.” 8. Example: “Among walkers, what percentage are P5?” Answer40%.
9. B. 10. Budget. 11. Reliability. 12. B costs$6 more and saves10 minutes.
13. $3 and$2.80 per item. 14. Pack B. 15. Lower total spending or less unused quantity. 16. Answers vary.
17. Both60%. 18. Method B:42 versus36. 19. Group sizes differ; rates use each method’s own total. 20. Other differences between groups may influence outcomes.
21. An option may look attractive while failing an important omitted criterion. 22. No; unknown is not zero. 23. It changes which options are feasible before comparison. 24. Answers vary; assumptions should be explicit.
27. Decision laboratory: choose under changing conditions
| Route | Time | Cost | Reliability |
|---|---|---|---|
| A | 25 min | $4 | 80% |
| B | 30 min | $2 | 95% |
| C | 20 min | $6 | 70% |
Decision 1: cost≤$4 and reliability≥90% → B only.
Decision 2: time≤25 min and cost≤$5 → A only.
Decision 3: fastest regardless of cost/reliability → C.
No route is universally “best”. The condition defines the decision problem.
28. Parent and tutor guide
Ask learners to circle the denominator words in conditional questions: “among P6”, “of walkers”, “within the low-cost group”. This simple habit prevents many percentage errors.
For decisions, require the rule before the answer. “I choose B because…” is stronger than “B is best.”
29. Mastery receipt
- I read joint counts and marginal totals in two-way tables.
- I choose the correct denominator for conditional percentages.
- I distinguish overall, joint and subgroup percentages.
- I filter options by hard constraints.
- I quantify simple trade-offs.
- I recognise that decision rules and weights are model assumptions.
- I avoid causal claims from conditional percentages alone.
- I state what data or criteria are missing from a decision.
Sources and scope
OECD PISA 2022 Mathematics Framework identifies conditional decision making as a focus within uncertainty and data, including interpreting relationships between variables and understanding how modelling choices affect conclusions.
For Singapore Primary scope, use the MOE Primary Mathematics Syllabus P1–P6, updated October 2025. This page is an enrichment/transition bridge.
Continue the Approximation & Uncertainty collection
- Geometric Approximation, Irregular Shapes and Estimating Area & Volume
- Measurement Precision, Error, Tolerance and Accuracy
- Simulation, Experimental Probability and Random Trials
- BTT Primary Mathematics Learning Hub
The Quiet Return
Data do not make decisions by themselves. Conditions choose the denominator, constraints define the feasible options, and values determine which trade-offs matter.

