2027 G3 SEC Mathematics K310 | Learner’s Guide Vol 0182. Students learn to add probabilities for an “or” question and multiply them for an “and” question. Then an examination problem changes one condition and the familiar shortcut fails. The difficulty is not the multiplication or addition; it is deciding whether two events can overlap and whether the first event changes the chance of the second.
This chapter is the next step after Volume 0178 on probability tree diagrams. We now decide what each branch means before adding or multiplying. The worked models are original. They use accessible examples, precise sample spaces, honest conditions and deliberate checks so a student can transfer the rules instead of memorising slogans.
1. What the 2027 syllabus actually says
The 2027 G3 syllabus directory gives Mathematics subject code K310. Its official syllabus includes the probability of single and simple combined events, possibility diagrams and tree diagrams where appropriate, and the addition and multiplication of probabilities involving mutually exclusive and independent events. This chapter concentrates on that specific probability decision.
The Mathematics assessment has two papers, each 2 hours 15 minutes, each worth 90 marks and contributing 50 per cent. An approved calculator may be used in both. Paper 2 ends with a real-world mathematical context question. The calculation might appear in different contexts, so students must read what is defined as the sample space and what assumptions are given rather than assume that every coin, die or card is automatically fair.
The syllabus says that non-exact answers normally follow three significant figures unless the question states a different requirement; angles in degrees have their own one-decimal-place convention. Many examples in this chapter produce exact fractions. Preserve those until a decimal answer is explicitly requested. The method should be shown with a clear explanation of why the chosen rule applies.
2. Begin with one fair die
Roll a fair six-sided die once. There are six equally likely outcomes: 1, 2, 3, 4, 5 and 6. Let A be “the result is 1” and B be “the result is 6”. P(A) = 1/6 and P(B) = 1/6. Both events concern the same single roll, so the die cannot show 1 and 6 at the same time.
These two events are mutually exclusive. The probability of A or B is P(A) + P(B) = 1/6 + 1/6 = 1/3. You can check the result by listing favourable outcomes: {1, 6} contains two outcomes out of six, so 2/6 = 1/3.
Notice how little arithmetic is involved. The important fact is that the events do not overlap. If A were “an even number” and B were “a number larger than 3”, a single roll of 4 or 6 would satisfy both. Adding the individual probabilities without adjusting for overlap would count those outcomes twice.
3. Mutually exclusive describes what cannot happen together
Two events are mutually exclusive within a specified trial if they cannot both occur on that trial. “This single die shows 1” and “this same die shows 6” cannot both be true. “The selected ticket is red” and “the selected ticket is blue” may also be mutually exclusive if every ticket has exactly one colour and the categories do not overlap.
Do not infer exclusivity merely because a sentence uses the word or. “A student is in the music club or the sports club” may include students in both. Without information about membership overlap, the events are not known to be mutually exclusive. The categories and trial must be defined before you choose a rule.
The first diagnostic question is therefore concrete: can one outcome belong to both events? If the answer is yes, do not use the simple addition rule without accounting for the shared outcomes. If the answer is no, the events are mutually exclusive and the straightforward sum is suitable.
4. Independent describes whether one result changes another
Now toss a fair coin and roll a fair die. Let C be “the coin shows heads” and D be “the die shows 6”. If the devices are fair and the trials do not influence each other, learning that the coin showed heads does not change the probability of the die showing 6. The events are independent.
Under these conditions, P(C and D) = P(C) × P(D) = 1/2 × 1/6 = 1/12. The sample space has twelve equally likely ordered coin–die outcomes, and exactly one is heads with six. This sample-space check confirms the multiplication.
Independence is not another word for “events occurring on different days” or “outcomes far apart”. A trial can influence a later trial even if they are separated in time. Conversely, two variables may be independent under a well-specified model even when observed in the same session. Independence is a relationship between chances, not a statement about physical distance.
5. Why exclusive is not the same as independent
Consider the two mutually exclusive positive-probability events on one die: A is “1” and B is “6”. If you learn that A occurred, the chance of B is zero. Yet without that information, P(B) was 1/6. The first outcome changes the probability of the second; those events are not independent.
Under the usual mathematical definitions, two mutually exclusive events with non-zero probabilities cannot be independent. Their intersection has probability zero, while independence would require P(A and B) = P(A)P(B), which would be positive. The two requirements conflict. A zero-probability event is a special boundary case that should not obscure the practical distinction.
The easiest classroom memory aid is not a slogan about words. Draw two overlapping circles when outcomes can satisfy both events; draw separated circles when the events are mutually exclusive. Then ask a different question: if I know one event happened, does it change the chance of the other? These are two different tests.
6. Simple addition for a genuine either-or result
A fair die is rolled. The event “1 or 2” has probability 1/6 + 1/6 = 1/3. The event “odd or even” has probability 3/6 + 3/6 = 1. This second result is sensible because every face is either odd or even. Neither pair overlaps on a single roll.
For a bag containing four red counters and six blue counters, one counter drawn uniformly at random must be either red or blue. The probability of red or blue is 4/10 + 6/10 = 1. But if the bag also contains green counters, the listed probabilities would not sum to one. The sample space should always be checked against the bag contents.
If a question says “either red or blue” but some objects are striped red and blue, the categories might overlap. Read the definitions. A correctly executed addition on an incorrect event definition is still the wrong model.
7. Overlapping events need care
Suppose one fair die is rolled. Let A mean “even”, giving {2, 4, 6}, and B mean “greater than 3”, giving {4, 5, 6}. A has probability 3/6; B has probability 3/6. Their common outcomes are 4 and 6, with probability 2/6.
Adding 3/6 and 3/6 gives 1, but the event “even or greater than 3” is {2, 4, 5, 6}, which has probability 4/6 = 2/3. The shared outcomes were counted twice in the naive sum. The general identity is P(A or B) = P(A) + P(B) − P(A and B). This follows simply from counting each shared outcome once.
The 2027 K310 syllabus specifically names addition for mutually exclusive events. This broader overlap identity is useful as a transparent extension and a check on when the simple rule fails; do not treat every more advanced notation in a textbook as a guaranteed standalone examination requirement. Listing outcomes is often sufficient for simple combined events.
8. Multiplication requires a correct branch probability
Imagine a bag with three red counters and two blue counters. A first counter is drawn uniformly, then replaced and mixed before the second draw. The probability of red first is 3/5. Because of replacement and mixing, the probability of red second remains 3/5 under this model. P(red then red) = 3/5 × 3/5 = 9/25.
Without replacement, the second branch changes. If red was drawn first, two red and two blue counters remain. The probability of a second red, given a first red, is 2/4. The probability of two reds is 3/5 × 2/4 = 3/10. This is not multiplication of two independent 3/5 probabilities.
The same word and appears in both questions. What changes is the second factor. A tree diagram is a convenient way to show that dependence, but the reasoning can also be written in a sentence: after a red counter is taken and not replaced, both the number of red counters and the total number change.
9. The difference between with and without replacement
Replacement returns the selected object to the collection before the next selection, often making the draws independent if the selection process is otherwise unchanged and random. Without replacement changes the collection, so later probabilities commonly depend on what happened earlier. The essential issue is whether the conditions have changed, not merely whether an object physically moved.
One exception worth understanding: separate selections could depend on a hidden shared condition even if each object is replaced. For example, if the bag’s composition were itself unknown and the first observation supplied information about that composition, the inferred probability of later outcomes could change. Elementary SEC questions usually state enough assumptions for a clear model, but this example shows why independence must be justified.
Students should write “with replacement, each draw has the same composition” or “without replacement, one item has been removed” rather than just reciting multiplication. The explanation reveals which branch probabilities belong in the calculation.
10. One coin twice can still be independent
Toss a fair coin twice under conditions in which the tosses do not affect each other. The ordered possibilities are HH, HT, TH and TT, each with probability 1/4. “Head on the first toss” and “head on the second toss” are independent even though the same coin is used. One toss does not alter the coin’s fairness for the next.
The probability of exactly one head is P(HT) + P(TH) = 1/4 + 1/4 = 1/2. We multiply along each specified sequence, then add across the two mutually exclusive sequences. This combines the two rules in one small problem.
The word exactly matters. “At least one head” includes HH as well, giving 3/4. A student who reports only HT and TH has solved a different problem. Translate verbal conditions into complete events before calculating.
11. Read at least one by using the complement
The event “at least one six in two fair die rolls” is easier through its opposite: “no six in either roll”. Under independent fair rolls, P(no six twice) = (5/6)² = 25/36. Therefore P(at least one six) = 1 − 25/36 = 11/36.
Why not add 1/6 + 1/6 and stop at 1/3? Because the event “first roll six” and “second roll six” can occur together. Their overlap, two sixes, would be counted twice. The correct value 11/36 is smaller than 1/3, exactly as expected after removing that duplication.
The complement method is a tool, not a magic formula. Name the complement carefully and check that it covers every outcome not in the event. “At least one” is often well served by “none”; “at most one” needs a different complement, usually “two or more”.
12. At most, exactly and more than are different instructions
For two fair coin tosses, the count of heads can be 0, 1 or 2. “Exactly one head” means HT or TH: 2/4. “At most one head” means TT, HT or TH: 3/4. “More than one head” means HH alone: 1/4. The arithmetic depends on interpreting the comparison words correctly.
For a die, “greater than 4” gives 5 or 6, while “at least 4” gives 4, 5 or 6. A student who silently replaces one phrase with the other loses a favourable outcome. Mark the boundary number before counting.
The broader lesson is linguistic precision. Probability notation is often shorter than the sentence, but the sentence defines what counts. Read slowly enough to catch inclusion and exclusion words, then work quickly once the event is clear.
13. Probability trees are maps of conditional branches
In a two-step tree, a branch represents an outcome of the next trial under the conditions created by the route already taken. That last phrase is important. On the first branch of a without-replacement problem, the bag’s composition changes; on the next branch, use the updated counts.
Multiply along a complete route because a route requires each event in sequence. Add across alternative complete routes when they cannot occur together in that single realised experiment. A route cannot simultaneously be red-then-blue and blue-then-red. That is why the two route probabilities are added for “one of each”.
Return to Volume 0178 if the tree itself is unfamiliar. This chapter adds the diagnostic layer: identifying mutually exclusive alternatives and independence before constructing the tree.
14. A full without-replacement example
A bag has three green and two yellow counters. Two are drawn without replacement. The probability of green then yellow is 3/5 × 2/4 = 6/20. The probability of yellow then green is 2/5 × 3/4 = 6/20. These two sequences are mutually exclusive paths within a single two-draw experiment.
The probability of one green and one yellow in either order is therefore 6/20 + 6/20 = 3/5. You can verify with unordered pairs. There are 5 choose 2 = 10 equally likely two-counter pairs; 3 × 2 = 6 contain one of each colour, giving 6/10 = 3/5.
Now compare the probability of two green counters: 3/5 × 2/4 = 3/10. The second green is less likely than the first because one green was removed. A common error is to use 3/5 on both branches and get 9/25. Such an answer ignores the described process.
15. Independent events can occur together
On a coin–die experiment, “heads” and “an even die face” can both happen. They are not mutually exclusive. If the coin and die are independent, their intersection has probability 1/2 × 3/6 = 1/4. This example directly challenges the misconception that independent means separate in the sense of unable to coincide.
Independence tells us that observing the coin result does not change the die’s even probability. Mutually exclusive tells us that a single outcome cannot satisfy both events. Both properties concern an event pair, but they answer different questions.
If someone says “heads and an even number are independent because a coin and die are different objects”, refine the reason: independence is justified by the specified fair, unaffected trials. Different objects can still be dependent when one process influences another. The condition, not the noun, is the foundation.
16. Mathematical notation, used sparingly
Let A and B denote events. P(A) is the chance that A occurs. P(A and B) is an intersection; P(A or B) is a union in the inclusive sense unless the question explicitly says otherwise. For mutually exclusive events, P(A and B) = 0 and P(A or B) = P(A) + P(B).
For independent events, P(A and B) = P(A) × P(B). The multiplication is valid because the probability of B is unchanged by learning A occurred. In general sequential problems, a branch probability must reflect the earlier result; do not replace it automatically with an unconditional probability.
Writing one line of mathematical notation is useful only if the symbols are connected to the problem. A candidate who writes P(A) = 3/5 without first identifying A can easily make an undetected reversal. Briefly define the events and then calculate. Clear notation is a service to reasoning.
17. Probability values are bounded
A probability cannot be negative or greater than one. If you add several numbers and obtain 1.3 for an event, something is wrong: overlapping outcomes may have been counted twice, the individual probabilities may have been misread, or the sample space may not describe a valid probability model.
A result of exactly 1 means the event is certain under the model. It does not mean that a measurement was flawless or that a real event is guaranteed despite uncertainty outside the model. Similarly, P(A) = 0 can mean impossible within the stated sample space; it does not always mean an event has never occurred in all conceivable settings.
Use the bounds as an immediate last-line check. For an event that contains two possible routes but not all possible routes, a probability of 1 should make you inspect the event definition again.
18. Real-world sampling must specify what is chosen
Suppose a school club selects one badge from a box. Ten are blue, five red and five green. If each badge is equally likely, P(blue) = 10/20 = 1/2. If badges are arranged so that some are easier to reach, the assumption of uniform selection may be false. The probability model is conditional on the selection mechanism.
Now the club selects one badge from each of two separate boxes. If both selections are independently random, multiplying probabilities is reasonable. But if the club chooses the second box according to the first colour, the second event can depend on the first decision even though the physical boxes are separate.
This is why examination questions specify fairness, randomness, replacement and counts. A student should use the model as stated, then avoid making claims about how actual club members behave unless further evidence is supplied.
19. Why a sample frequency is not automatically a theoretical probability
A fair die has theoretical P(6) = 1/6. If a student rolls it twenty times and obtains five sixes, the observed relative frequency is 5/20 = 1/4. The observations do not force the die’s theoretical probability to become 1/4. A small sample can fluctuate around the model probability.
Conversely, if repeated observations over many trials consistently differ from the model, that is a reason to inspect whether the fairness assumption holds. A sample does not prove the model correct or false by itself; its evidential strength depends on design, size and uncertainty.
Separating theoretical chance from observed frequency will matter in integrated Science tasks where the word probability appears next to trial results. State whether you are calculating from a specified random model or estimating from data.
20. A two-way table can reveal overlap
In a fictional group of twenty pupils, twelve joined the reading club, nine joined the science club and five joined both. The number in at least one club is 12 + 9 − 5 = 16. Four joined neither. If selecting one pupil uniformly, the probability of a reading or science club member is 16/20 = 4/5.
Adding 12/20 + 9/20 produces 21/20, which is impossible. The overlap has caused double counting. In this simple example, listing the shared five members and subtracting them once repairs the model. The exercise reinforces a general habit of asking whether one outcome can satisfy both descriptions.
Do not assume the numbers in a table form a probability sample of all Singapore pupils. They describe this fictional group only. A mathematical probability for a uniformly chosen member of the group and a claim about a wider population are different types of statements.
21. Common wrong methods and the exact repair
Error: multiplying because the question says “or”. The event “1 or 6 on one die” calls for counting alternative mutually exclusive faces, not multiplying 1/6 by 1/6. A single die cannot show both faces, so that product describes an impossible combined event within one roll.
Error: adding because the question says “and”. For independent coin and die events, “heads and six” requires both outcomes, so multiplication gives 1/12. Adding 1/2 and 1/6 would count a much larger set of outcomes than the conjunction asks for.
Error: assuming the second chance stays fixed. In two draws without replacement, the remaining count changes. Make a note of what was removed and adjust the denominator and numerator. Error: ignoring overlap. If an even face can also be greater than three, do not add without removing the shared part.
22. A short method-choice checklist
Before a calculation, write the event in everyday language. Then ask whether it has alternative routes or a sequence of required outcomes. Next ask if the alternatives are mutually exclusive, and whether the probability of a later outcome is affected by earlier results. Only then choose addition, multiplication or a combined tree method.
After calculating, list or sketch outcomes in a simple model to see whether the result is plausible. A tree branch’s probabilities should total one at each decision point when all next possibilities are represented. The probabilities of all complete, mutually exclusive final routes should total one. If they do not, something is missing or inconsistent.
Finally, protect exact fractions until the end. If a question asks for a percentage, convert the final fraction. If it asks for three significant figures, round only when reporting. A correct rule with premature numerical rounding can still lose precision.
23. Thirty original worked probability drills
Each exercise states its sample space or model assumptions. Read the full condition before selecting a rule. Several are paired to show what changes when one assumption changes. These are original learning exercises, not official or predicted K310 examination questions.
Practice 01
Question. One fair die: P(1 or 6)?
Worked answer. 1/3. Why. The faces are mutually exclusive on one roll, so 1/6 + 1/6.
Practice 02
Question. One fair die: P(even)?
Worked answer. 1/2. Why. Three of six faces are even: {2,4,6}.
Practice 03
Question. One fair die: P(even and greater than 3)?
Worked answer. 1/3. Why. The intersection contains only 4 and 6.
Practice 04
Question. One fair die: P(even or greater than 3)?
Worked answer. 2/3. Why. The union is {2,4,5,6}, four of six faces.
Practice 05
Question. Fair coin and die, independent: P(head and 6)?
Worked answer. 1/12. Why. Multiply 1/2 by 1/6.
Practice 06
Question. Fair coin and die, independent: P(head and even)?
Worked answer. 1/4. Why. Multiply 1/2 by 3/6.
Practice 07
Question. Two independent fair coin tosses: P(HH)?
Worked answer. 1/4. Why. Multiply 1/2 by 1/2.
Practice 08
Question. Two independent fair coin tosses: P(exactly one head)?
Worked answer. 1/2. Why. HT and TH each have probability 1/4; add the two routes.
Practice 09
Question. Two independent fair coin tosses: P(at least one head)?
Worked answer. 3/4. Why. Use 1 − P(TT) = 1 − 1/4.
Practice 10
Question. Two independent fair coin tosses: P(no heads)?
Worked answer. 1/4. Why. The sole outcome is TT.
Practice 11
Question. Two independent fair die rolls: P(two sixes)?
Worked answer. 1/36. Why. Multiply 1/6 by 1/6.
Practice 12
Question. Two independent fair die rolls: P(at least one six)?
Worked answer. 11/36. Why. Subtract P(no six twice) = 25/36 from 1.
Practice 13
Question. Bag: 3 red, 2 blue; one uniform draw. P(red or blue)?
Worked answer. 1. Why. Every counter is one of those two colours in this model.
Practice 14
Question. Same bag, replacement and mixing: P(two reds)?
Worked answer. 9/25. Why. Both draws have red probability 3/5; multiply.
Practice 15
Question. Same bag, no replacement: P(two reds)?
Worked answer. 3/10. Why. Multiply 3/5 by the updated 2/4.
Practice 16
Question. Same bag, no replacement: P(red then blue)?
Worked answer. 3/10. Why. Multiply 3/5 by 2/4.
Practice 17
Question. Same bag, no replacement: P(blue then red)?
Worked answer. 3/10. Why. Multiply 2/5 by 3/4.
Practice 18
Question. Same bag, no replacement: P(one of each colour)?
Worked answer. 3/5. Why. Add the mutually exclusive red–blue and blue–red paths.
Practice 19
Question. Same bag, no replacement: P(two blues)?
Worked answer. 1/10. Why. Multiply 2/5 by 1/4.
Practice 20
Question. Same bag, no replacement: P(same colour twice)?
Worked answer. 2/5. Why. Add two reds 3/10 and two blues 1/10.
Practice 21
Question. A bag has 4 green and 1 yellow. Two without replacement. P(two green)?
Worked answer. 3/5. Why. Multiply 4/5 by 3/4 = 12/20.
Practice 22
Question. Same bag, two with replacement and mixing. P(two green)?
Worked answer. 16/25. Why. Multiply 4/5 by 4/5.
Practice 23
Question. Three independent fair coins: P(all tails)?
Worked answer. 1/8. Why. Multiply three factors of 1/2.
Practice 24
Question. Three independent fair coins: P(at least one head)?
Worked answer. 7/8. Why. Use the complement of all tails.
Practice 25
Question. Twenty pupils: 12 in reading club, 9 in science club, 5 in both. How many in at least one?
Worked answer. 16 pupils. Why. 12 + 9 − 5 counts the overlap once.
Practice 26
Question. Same group, uniformly select one pupil. P(neither club)?
Worked answer. 1/5. Why. Four of twenty are in neither club.
Practice 27
Question. One fair die: are ‘1’ and ‘6’ independent?
Worked answer. No. Why. They are mutually exclusive with positive probabilities; seeing one rules out the other.
Practice 28
Question. Fair coin and die: are ‘heads’ and ‘even die face’ mutually exclusive?
Worked answer. No. Why. They can occur together; under stated independent trials their intersection is 1/4.
Practice 29
Question. A fair die rolled twice: P(first 6, second not 6)?
Worked answer. 5/36. Why. Multiply 1/6 by 5/6 for independent rolls.
Practice 30
Question. Two independent fair die rolls: P(exactly one six)?
Worked answer. 5/18. Why. The two exclusive routes total 1/6 × 5/6 + 5/6 × 1/6 = 10/36.
24. A mixed exam-style decision case
A fictional school game has a bag containing two gold counters and three silver counters. A participant draws two counters without replacement. A prize is awarded if both are gold; a smaller prize is awarded if exactly one is gold. First identify all possible route types: gold–gold, gold–silver, silver–gold and silver–silver. These routes are mutually exclusive complete outcomes.
P(both gold) = 2/5 × 1/4 = 1/10. P(exactly one gold) = 2/5 × 3/4 + 3/5 × 2/4 = 6/20 + 6/20 = 3/5. P(two silver) = 3/5 × 2/4 = 3/10. Check the partition: 1/10 + 3/5 + 3/10 = 1.
If the prizes are changed so that a participant wins if at least one counter is gold, the probability is 1 − 3/10 = 7/10. The calculation did not require memorising a new formula; it required recognising which routes are included by the changed instruction. This is the transition from routine tree use to flexible examination reasoning.
25. How to check for errors without redoing the whole question
Check the sample space first. If the die is fair, six outcomes are equally likely. If the collection contains five counters, the first denominator is five. If a counter is not replaced, the second denominator becomes four. These basic checks catch model errors before any complicated working.
Then inspect every or and and. An “or” may require adding disjoint routes, but the events themselves might overlap; a single outcome could be counted twice. An “and” may require multiplying along a sequence, but the second probability might be conditional on the first. Finally, confirm that the answer falls between zero and one.
Estimate if useful. Two successive sixes on independent fair dice must be less likely than one six, so 1/36 makes sense. One of each colour in the 3-red/2-blue without-replacement bag is reasonably common, so 3/5 is plausible. An exact fraction is easier to audit than a prematurely rounded decimal.
26. Connect probability to earlier mathematics
Volume 0178 owns basic tree diagrams and route construction. Volume 0149 explains why final rounding should not disrupt exact calculations. The Mathematics Hub connects these skills to data handling, fractions, algebra and real-world problem representation.
The G3 SEC learner guide provides the broader examination route, and the PSLE Learning Guide supports earlier fraction and chance language. If event descriptions themselves are confusing, the Secondary English Hub helps with condition words such as either, both, at least and exactly.
For the official scope and paper format, consult the 2027 SEAB K310 Mathematics syllabus. The goal is to become so confident about what each event means that addition, multiplication and the appropriate tree branch follow naturally.
27. Finish with the reason, not merely the answer
An examiner should be able to see why you multiplied or added. A short statement such as “With replacement, the two draws have unchanged probabilities” or “The two possible orders are mutually exclusive routes” may explain more than a long row of unexplained arithmetic.
Probability is a language for structured uncertainty. It does not ask a student to guess what will happen next; it asks the student to describe precisely what could happen, under which assumptions, and with what chance. A reliable answer begins by understanding the experiment and ends by checking whether the computed number matches the described event.
