Did you know that flipping a fair coin ten times does not guarantee five heads? In one short run you might see seven heads; in another, only three. The theoretical probability of heads stays one-half, while the experimental proportion depends on what actually happened. Knowing the difference turns a table of results into useful mathematical evidence.
A Secondary 2 learner can calculate 18 ÷ 50 confidently and still announce that the next 50 trials will produce exactly 18 successes. The calculation was fine; the interpretation moved too far. We can correct that with one ordinary question: ‘Are you reporting what happened, or predicting what must happen?’ The answer changes how carefully the learner should describe the result.
For Secondary 2 G2 Mathematics learners and parents around Bukit Timah, this topic guide explains the essential skill, shows how to solve it and helps identify what to practise next. The purpose is not to work through as many questions as possible; it is to make mathematical decisions explicit and test whether they still work when the next question looks unfamiliar.
The quick answer
| Question | Useful starting point |
|---|---|
| What is this lesson about? | Experimental Probability and Relative Frequency within Secondary 2 G2 Mathematics. |
| What should a student understand first? | Name the concept and any condition needed before choosing a method. |
| How do we repair an error? | Locate the first incorrect decision and practise a changed example. |
| How do we check progress? | Look for correct independent work, a justified method and sensible interpretation. |
| How should parents approach G2? | Use the student’s current subject level and actual school assessment coverage. |
Did you know? The same wrong answer can hide very different learning needs. One student may have a solid understanding but copy a number incorrectly. Another may execute the method flawlessly once the tutor has already chosen it, yet struggle to recognise the right approach in a new problem. Effective teaching pays attention to that distinction.
What G2 Mathematics means for Secondary 2 students
In Singapore’s Full Subject-Based Banding system, G2 is a subject level rather than a single label for a child’s capability in every subject. The work builds relationships between numerical reasoning, algebra, geometry, data and mathematical interpretation. Chapter schedules vary by school. A tutorial plan should begin with the learner’s actual materials, marked papers and teacher feedback, not a generic online calendar.
A public 2026 Secondary 2 G2 content outline from Yishun Secondary School includes the main subject area relevant to this guide. It is an example of one school’s sequence, not proof that every class will assess a specific extension. The 2027 SEC Mathematics syllabus can be useful for wider direction, but should not replace the student’s present school lesson plan.
Curriculum background: 2026 Secondary 2 G2 school content outline and SEAB’s published G2 Mathematics syllabus. These are educational references, not a claim about an individual’s enrolment or class timetable.
The key idea behind experimental probability and relative frequency
Experimental probability is the relative frequency of an event in observed trials. The formula is number of observed occurrences of the event divided by total number of trials. It can be compared with a theoretical probability under a specified fair model, but an observed proportion is not a promise about future outcomes. Students must learn to keep sample size, event definition and uncertainty visible.
An effective first conversation has three parts: describe the question in ordinary language, say which mathematical rule represents it and explain what a believable answer would look like. If the student cannot yet choose a rule, the tutor can simplify the situation, then gradually return to the original question. The aim is for the learner to own the reasoning, not just recognise the teacher’s handwriting.
The concepts that make this topic easier
1. What counts as a trial
A trial is one performance of a random experiment under defined conditions, such as one spin of a spinner or one draw with replacement. The total trial count forms the denominator.
Teaching move: Count the attempts, including those where the desired event did not occur. Each trial should have a clearly defined outcome.
Warning to notice: A student who divides by successes rather than all attempts is calculating the wrong ratio.
Before adding difficulty, give one short related question without the answer visible. Ask what has changed, what remains true and why the same principle still applies—or does not apply. This test of transfer is stronger evidence than finishing ten examples that all have the same layout.
2. An event must be named precisely
The event ‘land on blue’ is not the same as ‘land on a primary colour’ or ‘do not land on red’. The event definition determines which observations count as successes.
Teaching move: Underline the exact event words and mark the matching recorded outcomes.
Warning to notice: If the event changes halfway through tallying, the experimental probability loses meaning.
Before adding difficulty, give one short related question without the answer visible. Ask what has changed, what remains true and why the same principle still applies—or does not apply. This test of transfer is stronger evidence than finishing ten examples that all have the same layout.
3. Relative frequency is a proportion
If red appears 18 times in 50 spins, its relative frequency is 18/50=0.36, equivalent to 36%.
Teaching move: Keep numerator and denominator labelled as red outcomes and all spins before reducing.
Warning to notice: The probability is 0.36, not 18 or 36; percentages must be explicitly written as 36%.
Before adding difficulty, give one short related question without the answer visible. Ask what has changed, what remains true and why the same principle still applies—or does not apply. This test of transfer is stronger evidence than finishing ten examples that all have the same layout.
4. Theoretical and experimental probability are different
For an ideal fair coin, P(head)=1/2 by the model. If eight of twelve observed flips are heads, experimental P(head)=8/12=2/3. Both statements can be correct.
Teaching move: Ask which calculation uses the model and which uses actual recorded counts.
Warning to notice: A small sample does not have to match theoretical proportions exactly.
Before adding difficulty, give one short related question without the answer visible. Ask what has changed, what remains true and why the same principle still applies—or does not apply. This test of transfer is stronger evidence than finishing ten examples that all have the same layout.
5. Larger samples often give more stable proportions
As independent trials of a stable chance process accumulate, relative frequency tends to settle closer to the underlying probability, although any finite run can vary.
Teaching move: Compare simulated or hypothetical data across several trial counts rather than asserting that every larger sample must be closer.
Warning to notice: No particular next toss is ‘due’ to make up for earlier results.
Before adding difficulty, give one short related question without the answer visible. Ask what has changed, what remains true and why the same principle still applies—or does not apply. This test of transfer is stronger evidence than finishing ten examples that all have the same layout.
6. Predictions from experiments are estimates
If a trial-based relative frequency is 0.35, one might estimate 0.35×200=70 successes in 200 future comparable trials. This is an expected count under the estimate, not a guaranteed total.
Teaching move: Use cautious words such as estimate or expect approximately.
Warning to notice: It is misleading to tell a parent or learner that probability predicts an exact sequence.
Before adding difficulty, give one short related question without the answer visible. Ask what has changed, what remains true and why the same principle still applies—or does not apply. This test of transfer is stronger evidence than finishing ten examples that all have the same layout.
7. Complement counts are part of the same data
When an event occurs 18 times in 50 trials, its complement occurs 32 times, giving experimental probability 32/50=0.64.
Teaching move: Check that counts sum to 50 and the two relative frequencies sum to 1.
Warning to notice: Subtract the success count from total trials, not from the number 1 directly.
Before adding difficulty, give one short related question without the answer visible. Ask what has changed, what remains true and why the same principle still applies—or does not apply. This test of transfer is stronger evidence than finishing ten examples that all have the same layout.
8. Experiments require comparable conditions
A probability estimate based on one spinner is not automatically transferable to a differently weighted spinner. Changing replacement rules or equipment changes the chance model.
Teaching move: Record fairness assumptions and whether draws are with replacement if that affects later trials.
Warning to notice: Data from incompatible experiments cannot simply be merged without caution.
Before adding difficulty, give one short related question without the answer visible. Ask what has changed, what remains true and why the same principle still applies—or does not apply. This test of transfer is stronger evidence than finishing ten examples that all have the same layout.
9. A table and a graph tell complementary stories
A table gives exact counts. A relative-frequency line plot across increasing trials can show fluctuations and overall stabilisation, but its shape does not predict the next outcome.
Teaching move: Ask students to label both axes and explain what each plotted point summarises.
Warning to notice: A steadily rising count of successes is not the same as a steadily rising relative frequency.
Before adding difficulty, give one short related question without the answer visible. Ask what has changed, what remains true and why the same principle still applies—or does not apply. This test of transfer is stronger evidence than finishing ten examples that all have the same layout.
Worked examples with an explanation and a check
Read the question and attempt your first step independently. The worked explanation is there to show why the decision is valid. After reading it, cover the answer and try a modified version. These are teaching examples, not leaked or predicted school examination questions.
Worked example 1: Calculate a relative frequency
Question: A spinner lands on red 18 times in 50 spins. Find experimental P(red).
Method: Divide favourable observed count by total trials: 18/50=9/25=0.36.
Check: 0.36=36%, and it lies between zero and one.
Common wrong turn: Using 50/18 would give a value greater than one.
Why this matters: the check connects the arithmetic to the original meaning. A learner who can explain that connection can often catch a mistaken denominator, an invalid algebraic step, a mismatched unit or an exaggerated interpretation before handing in the paper.
Worked example 2: Compare with a fair coin model
Question: A fair coin has 8 heads in 12 flips. Find theoretical and experimental P(head).
Method: Theoretical P(head)=1/2. Experimental P(head)=8/12=2/3.
Check: Their difference is 2/3−1/2=1/6; short samples can vary.
Common wrong turn: Do not declare the coin automatically unfair based only on twelve flips.
Why this matters: the check connects the arithmetic to the original meaning. A learner who can explain that connection can often catch a mistaken denominator, an invalid algebraic step, a mismatched unit or an exaggerated interpretation before handing in the paper.
Worked example 3: Calculate the complement
Question: A machine trial records 27 successful outcomes from 45 attempts. Find experimental P(not successful).
Method: Unsuccessful attempts=45−27=18. Relative frequency=18/45=2/5=0.4.
Check: Successful frequency 27/45=0.6; total is 1.
Common wrong turn: Subtract counts before dividing, rather than performing 1−27.
Why this matters: the check connects the arithmetic to the original meaning. A learner who can explain that connection can often catch a mistaken denominator, an invalid algebraic step, a mismatched unit or an exaggerated interpretation before handing in the paper.
Worked example 4: Estimate a future count
Question: In 80 comparable trials, an event occurred 28 times. Estimate its count in 200 future trials.
Method: Experimental relative frequency=28/80=0.35. Estimated count=0.35×200=70.
Check: 70 is a plausible average-like estimate under the same conditions, not a guaranteed outcome.
Common wrong turn: Do not claim exactly 70 must happen.
Why this matters: the check connects the arithmetic to the original meaning. A learner who can explain that connection can often catch a mistaken denominator, an invalid algebraic step, a mismatched unit or an exaggerated interpretation before handing in the paper.
Worked example 5: Compare two separate batches
Question: Batch A has 12 successes out of 40; Batch B has 36 successes out of 100. Compare frequencies.
Method: A=12/40=0.30, B=36/100=0.36. The observed proportions differ by 0.06, or 6 percentage points.
Check: Compare relative frequencies, not raw success counts, because sample sizes differ.
Common wrong turn: Saying B is three times as successful because 36 is three times 12 ignores the trials.
Why this matters: the check connects the arithmetic to the original meaning. A learner who can explain that connection can often catch a mistaken denominator, an invalid algebraic step, a mismatched unit or an exaggerated interpretation before handing in the paper.
Worked example 6: Estimate from a recorded table
Question: A game has outcomes red 15, blue 25 and green 10 across 50 plays. Find P(blue) and P(not blue) experimentally.
Method: P(blue)=25/50=0.5. P(not blue)=(15+10)/50=25/50=0.5.
Check: The event and its complement sum to 1; counts sum to 50.
Common wrong turn: Three colour labels do not make each probability automatically one-third.
Why this matters: the check connects the arithmetic to the original meaning. A learner who can explain that connection can often catch a mistaken denominator, an invalid algebraic step, a mismatched unit or an exaggerated interpretation before handing in the paper.
Worked example 7: Keep ratios and percentages consistent
Question: An experiment has 21 desired outcomes in 60 attempts. Express relative frequency as fraction, decimal and percentage.
Method: 21/60=7/20=0.35=35%.
Check: All representations describe the same observed proportion.
Common wrong turn: A decimal 0.35 is not 0.35%; that would mean 0.0035.
Why this matters: the check connects the arithmetic to the original meaning. A learner who can explain that connection can often catch a mistaken denominator, an invalid algebraic step, a mismatched unit or an exaggerated interpretation before handing in the paper.
Worked example 8: Reconstruct an observed count
Question: An event has experimental frequency 0.4 across 75 trials. How many event occurrences were recorded?
Method: Event count=0.4×75=30.
Check: 30/75=0.4 exactly.
Common wrong turn: An observed count must be a whole number; when a rounded frequency is given, exact reconstruction may not always be possible.
Why this matters: the check connects the arithmetic to the original meaning. A learner who can explain that connection can often catch a mistaken denominator, an invalid algebraic step, a mismatched unit or an exaggerated interpretation before handing in the paper.
Worked example 9: Explain an unexpected short run
Question: A fair die lands on six three times in its first ten rolls. Does that change theoretical P(six) for the next roll?
Method: No. Under independent fair-roll assumptions, theoretical P(six) remains 1/6. Experimental fraction so far is 3/10.
Check: An individual independent roll does not owe the experiment a correction.
Common wrong turn: Thinking a six is now less likely because it appeared often is the gambler’s fallacy.
Why this matters: the check connects the arithmetic to the original meaning. A learner who can explain that connection can often catch a mistaken denominator, an invalid algebraic step, a mismatched unit or an exaggerated interpretation before handing in the paper.
Seven recurring errors and how to correct them
| Mistake | Clue | Tutor’s correction | Independent proof |
|---|---|---|---|
| Denominator is success count | 18/18 reported for 18 successes in 50 | Count every trial | 18/50 used |
| Observed count treated as probability | Answer written ’18’ | Divide by total attempts | Value within 0 and 1 |
| Percentage place value lost | 0.36 written as 0.36% | Multiply decimal by 100 for % | 36% correctly labelled |
| Theoretical replaced by observed | Eight heads of twelve changes ideal fair-coin P | Separate model from record | Both values reported |
| Estimated count promised as exact | 70 expected becomes 70 guaranteed | State prediction is approximate | Interpretation cautious |
| Sample sizes ignored | Raw success counts compared | Convert each to relative frequency | Fair comparison |
| Incompatible trial conditions combined | Different mechanisms pooled blindly | Inspect trial definitions | Only comparable data used |
We do not want an error log that simply says ‘careless mistake’ sixteen times. Name what changed the Mathematics: an invalid assumption, a misread denominator, a wrong relationship, a copied sign or a missing unit. Then practise that cause directly. A page full of red crosses may be one broken rule repeated downstream.
Sixteen quick practice questions with answers
Attempt the questions with notes closed. If you need to check an answer, make sure you also read the reason. Return to difficult ones after a day or two and alter one number. The point is spaced retrieval and independent method choice, not speed alone.
Practice 1: 18 successes in 50 trials: probability?
Answer: 0.36. Why: 18/50.
Practice 2: 21 successes in 60 trials: probability?
Answer: 0.35. Why: 21/60.
Practice 3: What percent is 0.35?
Answer: 35%. Why: Multiply by 100.
Practice 4: What fraction is 0.4 in simplest form?
Answer: 2/5. Why: 4/10 reduces.
Practice 5: 8 heads in 12 flips: experimental probability?
Answer: 2/3. Why: 8/12.
Practice 6: Fair coin theoretical heads probability?
Answer: 1/2. Why: Two equally likely faces.
Practice 7: Event 27 of 45 trials: event probability?
Answer: 0.6. Why: 27/45.
Practice 8: Not-event probability in that record?
Answer: 0.4. Why: 18/45.
Practice 9: 28 successes in 80: observed frequency?
Answer: 0.35. Why: 28/80.
Practice 10: At 0.35 frequency, approximate successes in 200?
Answer: 70. Why: 0.35×200.
Practice 11: 12/40 relative frequency?
Answer: 0.30. Why: Divide 12 by 40.
Practice 12: 36/100 relative frequency?
Answer: 0.36. Why: Divide 36 by 100.
Practice 13: Their difference in percentage points?
Answer: 6. Why: 36% minus 30%.
Practice 14: Total spins with red 15 blue 25 green 10?
Answer: 50. Why: Sum counts.
Practice 15: P(blue) experimentally in those spins?
Answer: 0.5. Why: 25/50.
Practice 16: Does one long run guarantee the next result?
Answer: No. Why: A finite sample cannot promise an individual outcome.
How a 90-minute, three-pupil Mathematics tutorial could work
The published eduKateSG small-group tutorial reference describes a three-pupil approach. In a small group, individual reasoning can be heard while classmates compare methods. But a small class is not a guarantee of grades; it works when each pupil has the chance to solve, explain and receive feedback. The following 90-minute structure is illustrative, not a claim of particular class vacancies.
| Stage | Time | What the student should demonstrate |
|---|---|---|
| Independent diagnostic | 10 minutes | Show a recent first error and explain what they tried. |
| Build one core concept | 15 minutes | State why the rule or representation is appropriate. |
| Guided comparison | 20 minutes | Compare two different-looking questions and their methods. |
| Independent work | 25 minutes | Solve unfamiliar variations with fewer prompts. |
| Mixed retrieval | 10 minutes | Bring back an earlier chapter’s connected idea. |
| Reflection and next task | 10 minutes | Record the corrected first step and the next review date. |
A thoughtful tutor will not assume that one confident student answering aloud means the other two understand. Each learner can write an initial line and then share the reasoning. That is how a small-group format becomes genuinely individual rather than merely a smaller lecture.
A practical four-week learning plan
Week 1: Diagnose and clarify
Build a reliable table of events and total attempts, then express frequencies as fractions, decimals and percentages.
Use one marked school paper and one fresh question. Record which part of the reasoning is missing and whether the learner knows the relevant vocabulary. Target the smallest foundational gap rather than assigning an entire textbook chapter.
Week 2: Build a reliable method
Contrast a fair theoretical model with short experimental samples and explain why results need not match.
Give short sessions on different days. Move from one tutor-guided success to two unprompted variations. Ask for a clear explanation of the first mathematical step, and make checking part of the routine.
Week 3: Mix and compare
Estimate future counts cautiously and compare different sample sizes using relative frequency.
Use questions without topic labels. A student should learn how to select a method rather than rely on the chapter heading. Introduce one contrast case where the familiar shortcut would be inappropriate.
Week 4: Check independent transfer
Complete mixed probability statements that require checking assumptions, complements and interpretations.
Give a new mini-assessment with changed values or presentation. Compare the first unaided steps, method choice and checks with Week 1. Use this evidence to select the next target.
What parents can ask at home
Parents do not have to supply solutions to help. A short, calm conversation can reveal what the student understands. Ask them to describe the event, figure, quantity or equation in their own words, then explain their first step. Let the learner think before offering a hint.
- What exactly does the question want?
- Which values or conditions were given?
- Why is that calculation valid for this situation?
- What would happen if one number or assumption changed?
- Can we use the original problem to check the result?
- Which step would you revisit first if the answer were wrong?
Ten or fifteen focused minutes may be enough for a daily practice conversation. Follow the school’s workload and give the learner time to rest. A small number of questions fully understood and revisited is usually a better foundation than a long exercise set that nobody has time to review.
How to tell whether tuition is working
| Skill | Starting point | Progressing | Independent |
|---|---|---|---|
| Understand question | Needs explanation of wording | Understands after one prompt | Explains the goal independently |
| Choose method | Copies teacher’s model | Selects from two choices | Justifies an appropriate method alone |
| Calculate | Repeated structural errors | Correct with occasional checking | Reliable on unseen examples |
| Explain | Recites a rule | Explains part of the reasoning | Can defend each essential step |
| Check | Stops at final answer | Checks with reminder | Checks original conditions voluntarily |
| Transfer | Stalls when question changes | Handles changed numbers | Handles different context and presentation |
Do not reduce progress to how many pages were finished. A child who can independently correct a common misconception, identify a missing condition and explain a result has gained a useful transferable skill. A school mark is important feedback, but it combines several topics and is not a complete measure of growth in this lesson.
What about Secondary 3 and later subject-level pathways?
Building a strong G2 foundation can help learners approach Secondary 3 with more confidence. Decisions about the level at which a student takes Mathematics should be discussed with the school and based on the student’s broader readiness and applicable criteria. Tutoring should support understanding, not promise automatic movement to G3 or a particular SEC result.
Frequently asked questions
1. What is experimental probability?
A proportion of observed trials in which an event occurred, calculated as the observed event count divided by total trials.
A practical next step is to ask the student to apply this explanation to one different question from their actual school work. If the first decision still needs prompting, the skill deserves another short lesson.
2. Why can a fair coin give seven heads out of ten?
Random results vary from run to run. Fairness specifies the probability for each independent flip, not a fixed count in every short sequence.
A practical next step is to ask the student to apply this explanation to one different question from their actual school work. If the first decision still needs prompting, the skill deserves another short lesson.
3. Is experimental probability always equal to theoretical probability?
No. They can differ in finite samples even when the theoretical model is correct.
A practical next step is to ask the student to apply this explanation to one different question from their actual school work. If the first decision still needs prompting, the skill deserves another short lesson.
4. What makes an experimental comparison fair?
Count all trials, use the same event definition and compare relative frequencies under reasonably comparable conditions.
A practical next step is to ask the student to apply this explanation to one different question from their actual school work. If the first decision still needs prompting, the skill deserves another short lesson.
5. Can I predict an exact future number of successes?
Probability can help estimate expected counts, but not guarantee the exact result of future random trials.
A practical next step is to ask the student to apply this explanation to one different question from their actual school work. If the first decision still needs prompting, the skill deserves another short lesson.
6. Why do we use relative frequency instead of raw counts?
It makes batches with different total numbers of attempts easier to compare.
A practical next step is to ask the student to apply this explanation to one different question from their actual school work. If the first decision still needs prompting, the skill deserves another short lesson.
7. Does a run of sixes make a six less likely next time?
Not for independent rolls of a fair die. The next roll still has the same theoretical chance.
A practical next step is to ask the student to apply this explanation to one different question from their actual school work. If the first decision still needs prompting, the skill deserves another short lesson.
8. How can parents check understanding?
Ask the student to name the event, total trial count, observed successes and one limitation of the resulting estimate.
A practical next step is to ask the student to apply this explanation to one different question from their actual school work. If the first decision still needs prompting, the skill deserves another short lesson.
The takeaway for Bukit Timah families
Did you know? Many Mathematics breakthroughs are quieter than a jump in a mark. Sometimes the breakthrough is that the student asks the right question before making the first calculation. Teach that habit, explain the rule, test a fresh example and return to it later. The confidence becomes more durable because it rests on something the learner can do.
Continue reading Sample spaces and single events and Probability and statistics overview. For broader planning, see the Bukit Timah Secondary 2 Mathematics topic guide and G2 or G3 subject-level support guide.
The original three-pupil Clementi Mathematics tutorial reference remains unchanged and explains the teaching model. Families can contact eduKate Singapore for actual class arrangements and suitability; lesson vacancies are not assumed.
Read the other three Secondary 2 G2 Mathematics guides
These complementary lessons connect mathematical reasoning, interpretation, real-world modelling and spatial visualisation. Pick the area that matches your current classwork and try one independent changed question after reading.