Did you know? “75% of workshop attendees submitted early” does not mean “75% of early submitters attended the workshop”. The same joint count can produce different conditional probabilities because the denominator changes. This is one of the quickest ways for a JC2 H2 Mathematics probability solution to go wrong even when every number in the table has been read correctly.
For Bukit Timah families supporting A-Level H2 Mathematics 9758, this guide tackles one focused issue: Why is P(A | B) generally different from P(B | A)? We use a fictional two-way table, an independence test, worked probabilities and short questions students can answer without memorising an entire textbook.
Quick answer: the part after the bar becomes your new universe
P(A | B) means the probability that A happens given that B has happened. Within this question, B is the restricted sample space. When P(B) > 0, P(A | B) = P(A ∩ B) / P(B). Reversing the condition changes the denominator: P(B | A) = P(A ∩ B) / P(A) when P(A) > 0.
The numerator is the same intersection in both expressions. The denominator changes because we have changed which group we are looking inside. That is the entire trick, and it is why reordering the letters changes the question.
A two-way table you can read in thirty seconds
Consider this made-up example: a school activity has 100 participants. Let A mean “submitted a task early” and B mean “attended an optional workshop”. The table records counts. These are synthetic numbers for learning, not observed school data.
| Fictional group | A: early submission | Not A | Total |
|---|---|---|---|
| B: attended workshop | 30 | 10 | 40 |
| Not B: did not attend | 20 | 40 | 60 |
| Total | 50 | 50 | 100 |
Start by checking all totals. Thirty people are in both A and B. There are 40 workshop attendees in total, 50 early submitters in total, and 100 participants overall. Each condition tells us which denominator is valid.
First question: among attendees, who submitted early?
The wording “among workshop attendees” tells us B is already true, so the denominator is 40. Of those 40, 30 also submitted early. Therefore P(A | B) = 30/40 = 0.75. Said plainly, 75% of the workshop-attending group submitted the task early in this fictional exercise.
It is useful to draw a rectangle around the entire B row. Once we know B has occurred, the 60 people who did not attend are no longer in the sample space for this particular question. We must not divide by 100 because that would answer an unconditional probability.
Second question: among early submitters, who attended?
Now the condition has reversed. “Among early submitters” means A is already true, so the denominator is 50. Of those 50, 30 attended the workshop. Thus P(B | A) = 30/50 = 0.60. The first answer was 0.75 and the second is 0.60. Both use the same numerator 30, and both are correct — for different questions.
This is not a mathematical contradiction. We changed the population being considered. The overlap of two sets can occupy 75% of one set but only 60% of the other.
An intersection is not a conditional probability
The probability that a randomly selected participant both submitted early and attended is P(A ∩ B) = 30/100 = 0.30. That divides by everybody, because no condition has restricted the sample space. We therefore have three distinct results: P(A ∩ B) = 0.30; P(A | B) = 0.75; P(B | A) = 0.60.
“Both A and B” means intersection. “A given B” means a conditional probability. “A or B” means a union. The words are not interchangeable; they direct us to different operations.
Use inclusion–exclusion for A or B
From the table, P(A) = 50/100 = 0.50 and P(B) = 40/100 = 0.40. The probability of A or B, including their overlap, is P(A ∪ B) = P(A) + P(B) − P(A ∩ B) = 0.50 + 0.40 − 0.30 = 0.60. Subtract the intersection once because adding the two separate group counts includes those 30 people twice.
A separate audit gives the same answer from the table: 30 people do both, 20 submit early without attending, and 10 attend without submitting early. That is 60 out of 100, or 0.60.
What independence does — and does not — mean
Events A and B are independent when knowing one occurred does not change the probability of the other. When the relevant probabilities are defined, this gives P(A | B) = P(A), equivalent to P(A ∩ B) = P(A)P(B). In the fictional table, P(A | B) = 0.75 but P(A) = 0.50, so A and B are not independent. The product test agrees: 0.50 × 0.40 = 0.20, different from the observed intersection probability 0.30.
A warning for interpretation: dependence or association is not proof that attending the workshop caused early submission. Participants might have differed in motivation or other circumstances before the activity. The table supports probability calculations, not a causal claim.
Another conditional probability from the same table
Among the 60 people who did not attend, 20 submitted early. Therefore P(A | not B) = 20/60 = 1/3. Compare that with P(A | B) = 3/4. A conditional probability is always tied to its named group; simply replacing B by not B changes which row matters.
Now reverse the question again: P(not B | A) = 20/50 = 0.40. This is not P(A | not B). For every conditional calculation, pause long enough to identify the group after the vertical bar.
One-minute diagnosis for a JC2 H2 learner
Show the table. Ask the student to circle the denominator for P(A | B), P(B | A), P(A | not B) and P(not B | A), in that order. The answers are 40, 50, 60 and 50. Only after those four denominators are correct should the student write the corresponding numerator. If their denominators are wrong, more calculator practice will not repair the underlying logic.
Five errors a tutor should look for
- Reversing the conditioning event: “A given B” is not “B given A”. Rewrite the question as “inside group B, how many are also A?”
- Dividing by the grand total each time: divide by the condition’s total, unless no event is given.
- Confusing the intersection and union: “both” uses A ∩ B, “either” uses A ∪ B.
- Assuming two events are independent: confirm P(A ∩ B) = P(A)P(B) before applying the simplified rule.
- Claiming causation from a table: statistical association and causal explanation are different questions.
Eight rapid checks with answers
- 1. In the fictional table, P(A) = 50/100 = 0.50.
- 2. P(B) = 40/100 = 0.40.
- 3. P(A ∩ B) = 30/100 = 0.30.
- 4. P(A | B) = 30/40 = 0.75.
- 5. P(B | A) = 30/50 = 0.60.
- 6. P(A | not B) = 20/60 = 1/3.
- 7. P(A ∪ B) = 0.60.
- 8. Are A and B independent? No: 0.30 ≠ 0.50 × 0.40.
A short tutoring routine that survives new numbers
A capable student can redraw the two-way table without the original numbers, label the conditioning group, and explain why the denominator changes. Next, switch the words from “attended given submitted” to “submitted given attended”. If the student can do that without prompting, they understand the relationship rather than memorising a formula.
The 2026 H2 Mathematics 9758 syllabus explicitly includes simple conditional probabilities, independence, tables, tree diagrams and the conditional probability identity. For more general probability contexts, see the Bukit Timah H2 Probability and Statistics guide.
Frequently asked questions
Is P(A | B) the same as P(B | A)?
Not in general. Both involve P(A ∩ B), but they divide by different conditioning-event probabilities.
What if the condition has probability zero?
The elementary ratio formula P(A ∩ B)/P(B) is undefined when P(B) = 0. Do not divide by zero or invent a percentage.
Can I use counts directly instead of converting everything to decimals?
Yes. With equally likely selections from the table, favourable counts divided by the conditioned group’s count are often the clearest calculation.
Does P(A | B) = P(A) always imply independence?
When P(B) > 0 under the usual event-probability framework, that equality is a test for independence, equivalently P(A ∩ B) = P(A)P(B).
Does the table prove workshops cause early submission?
No. It is a fictional illustration of conditional probability. Even a real observational table would not establish causation by itself.
What to read next
- Binomial ‘At Least One’ Without Off-by-One Errors
- Why P(A|B) Is Not P(B|A) — you are here
- Integration by Parts: Choosing u and dv
- Real-Coefficient Polynomials and Conjugate Roots
For the broader study route, use the Bukit Timah H2 Mathematics tuition subject guide and Bukit Timah JC2 tuition hub. A separate eduKateSG small-group mathematics tutoring example from Clementi illustrates the tuition format in another neighbourhood.
Syllabus and editorial note
See the official SEAB 2026 Singapore–Cambridge H2 Mathematics syllabus 9758, and consult the syllabus for your own examination year. Every question, count and practice dataset here is an original illustrative teaching example, not actual learner data, a past-year exam question or a promised examination outcome. This article explains a specific error; it is not a substitute for the complete official syllabus.