Did you know? Removing just one name after a draw changes every probability in the next draw. A student may correctly calculate a first chance, copy the same fraction into the second step, and obtain a polished but incorrect result. The hidden issue is not multiplication. It is that the pool of possible outcomes has changed.
Probability without replacement means that an item selected on one draw is not returned before the next draw; therefore the total number of eligible items, and sometimes the number of items of the kind being counted, changes. To calculate a two-stage chance, multiply the first-stage probability by the second-stage probability conditional on what happened first. If several allowed orders produce the event, calculate each separate route and add the non-overlapping routes.
We will use a fictional classroom token-drawing exercise in which a participant cannot be selected twice. It is a mathematics model, not an actual raffle, advertising campaign, school policy or recommendation to organise a paid lottery. The method applies to drawing names, choosing representatives and selecting objects without returning them to a container.
A diagnostic question before any calculations
Put twelve identical folded cards in a clear description: each card represents one distinct eligible participant. After one card is picked, is the second draw out of twelve cards or eleven? A learner who says twelve is treating the picks as though the first card was put back. A learner who says eleven but keeps the number of favoured cards unchanged regardless of which card was selected has understood only half of the rule.
Ask the child to narrate the bag after each draw: “What leaves? What remains? Which label was removed?” That verbal check exposes the mechanism before a formula is memorised. The precise identity of the drawn card matters because the composition of the remaining pool can differ between possible branches.
Worked example: twelve fictional club members, two places
Imagine a school enrichment club making a purely hypothetical demonstration of random selection. There are 12 distinct, equally eligible names: 5 pupils are from a Science activity group, 4 from a Language group and 3 from an Art group. The labels classify names for the exercise only; they are not ability groups, admissions categories or actual school participants. Two names will be drawn one after another without replacement. Assume each remaining name is equally likely at each stage, no name appears twice and no special rule overrides the draw.
The entire starting pool has 5 + 4 + 3 = 12 names. The two selected pupils will be distinct because the first card is removed. Our first question is deliberately ordered: What is the probability that the first selected name is from Science and the second is from Art?
Step one: calculate Science first, then Art
- Before any card is removed, P(Science first) = 5/12.
- If Science was first, one of the five Science cards has left. The remaining bag has 11 cards: 4 Science, 4 Language and 3 Art.
- Conditional on Science first, P(Art second) = 3/11, because all three Art cards remain among eleven eligible cards.
- P(Science first AND Art second) = (5/12) × (3/11) = 15/132 = 5/44 ≈ 11.36%.
The fraction 3/11 is not an arbitrary trick. It describes the actual second-stage state of the pool after the first-stage event. The denominator fell from twelve to eleven. The number of Art cards stayed at three because the first draw removed a Science card. If the first draw had removed an Art card instead, the remaining Art count would have fallen to two.
An independent counting check gives the same result. There are 12 × 11 = 132 ordered pairs of distinct names. Of these, 5 Science names can occupy the first position and 3 Art names the second, giving 5 × 3 = 15 favourable ordered pairs. Thus 15/132 = 5/44. Both methods agree, which is more reassuring than a remembered rule alone.
Step two: distinguish an ordered question from an either-order question
Change the question: What is the chance that the two selected names include one Science member and one Art member, in either order? There are now two routes: Science then Art, or Art then Science. They cannot occur simultaneously for the same ordered pair of two cards, so their probabilities can be added.
- Science then Art: (5/12) × (3/11) = 15/132.
- Art then Science: (3/12) × (5/11) = 15/132.
- Either order: 15/132 + 15/132 = 30/132 = 5/22 ≈ 22.73%.
If a pupil reports 5/44 to the either-order question, the arithmetic may be right but the interpretation is incomplete: only one of the two acceptable sequences was counted. If the child doubles every probability automatically, another error appears whenever the routes are not symmetric or do not remain disjoint. The habit should be “name every allowed route” rather than “always multiply by two”.
Step three: calculate two Science names
For Science followed by Science, the first favourable draw removes one of five Science cards. Only four Science cards remain among eleven. Therefore P(two Science) = (5/12) × (4/11) = 20/132 = 5/33 ≈ 15.15%. The mistaken expression (5/12) × (5/12) would correspond to independent picks from an unchanged pool, not this selection process.
Check the count by ordered pairs again: five choices for the first Science name and four different choices for the second yield 20 favourable ordered pairs out of 132. Or count unordered pairs: choose two of the five Science names (10 pairs) out of all ways to choose two of twelve names (66 pairs), giving 10/66 = 5/33. Two independent representations give the same probability.
Step four: use the complement for “at least one”
A question saying at least one Science name includes one Science and one non-Science, or two Science names. Enumerating all valid sequences is possible, but it is often simpler to calculate the one event that does not qualify: both names are non-Science.
There are 7 non-Science names (4 Language and 3 Art). P(no Science in two draws) = (7/12) × (6/11) = 42/132 = 7/22. Therefore P(at least one Science) = 1 − 7/22 = 15/22 ≈ 68.18%. The complement check is transparent: the event and its opposite together have probability 15/22 + 7/22 = 1.
The result does not predict that 68.18% of some real school’s selections will contain a Science member. It is a probability under this carefully stated fictional model. A few actual draws can deviate strongly from the theoretical proportion by chance, and a biased selection procedure would violate the model’s equal-likelihood assumption.
A probability tree is a map of changing conditions
To draw a small tree, begin with Science 5/12, Language 4/12 and Art 3/12. After a Science first branch, write the second-stage shares as Science 4/11, Language 4/11 and Art 3/11. After a Language first branch, write Science 5/11, Language 3/11 and Art 3/11. After Art first, write Science 5/11, Language 4/11 and Art 2/11. Every set of second-stage branch probabilities must add to 11/11 = 1.
That sum is an immediate diagnostic for mistakes. If a pupil writes 5/11 + 4/11 + 3/11 after Science first, the total would be 12/11, which is impossible for a complete set of mutually exclusive outcomes. The wrong numerator tells us precisely which card was not removed from the mental model.
For foundational background, the OpenStax explanation of independence and sampling distinguishes replacement from no replacement. The OpenStax multiplication rule explains how a conditional second probability combines with a first probability. We use those principles to derive, rather than copy, the unique classroom example here.
What would be different if the first card went back?
If the first card were returned and the pool fully mixed before the next draw, the total would again be 12 and the number of Science cards would again be 5. Then, under independent equal-likelihood selections, P(Science twice) = (5/12) × (5/12) = 25/144 ≈ 17.36%. But one individual could also be selected twice, and the exercise would no longer enforce two distinct names. That operational rule changes the mathematics.
The comparison is useful because it isolates a single assumption. Same original composition, same two selections, but a different between-draw procedure. When students can explain why 5/33 and 25/144 differ, they understand the dependency instead of merely memorising two formulas.
Fairness is more than calculating the right fraction
In a real classroom, teachers should not infer that a mathematically uniform draw is the only fair way to choose representatives. Selection may need consent, availability, safety, accommodation, relevance to the role and transparency about eligibility. If the purpose is to choose an expert presenter, demonstrated readiness might matter more than random selection. If the purpose is to demonstrate probability, make cards equally readable and avoid any way to recognise a name by fold, paper colour or position.
A random procedure with a duplicate name in the container is not equal-likelihood selection across people: someone receives extra chances. The lesson in How to Detect Duplicate Responses in a Classroom Survey supports a related evidence habit: inspect whether units counted as distinct really are distinct. The mathematical probability model should never obscure a procedural defect in how cards were prepared.
Three frequent misconceptions and their repair
- “The denominator stays at twelve.” Repair: count remaining eligible names after the first removal; the second draw uses eleven.
- “The numerator always falls by one.” Repair: it falls for the category that was actually selected, not for every category.
- “If the order does not matter, I can ignore one order.” Repair: count all ordered routes for the event or use consistent unordered-pair counting; never mix the two methods mid-calculation.
Progressive practice from supported calculation to transfer
Guided. Give the learner a smaller bag with 3 blue and 2 green tokens. Draw two without replacement. Ask for blue then green: (3/5) × (2/4) = 6/20 = 3/10. The child must describe the four-token second bag aloud. Then ask for green then blue: (2/5) × (3/4) = 6/20 = 3/10. Either order is 3/10 + 3/10 = 3/5.
Paired. Student A writes a flawed calculation for two blue tokens: (3/5) × (3/5). Student B must diagnose the exact assumption being made and replace it with (3/5) × (2/4) = 3/10. Then swap roles for a flawed either-order answer that omits one branch. This is a discussion of the reason for correction, not just the final fraction.
Independent. Without looking at the worked example, the pupil invents a bag of 4 red, 3 yellow and 2 white tokens and finds the probability of drawing two different colours without replacement. One reliable route is to calculate the complement of same-colour draws: [4×3 + 3×2 + 2×1] ÷ (9×8) = 20/72 = 5/18, so different colours have probability 1 − 5/18 = 13/18 ≈ 72.22%. The child should also justify why the ordered-pair denominator is 9×8 = 72.
How to observe whether the learner is becoming independent
A teacher can check four visible behaviours: the learner describes the remaining sample space correctly, changes the right numerator and denominator, lists the event’s possible orders and checks the result using a second representation. A parent can ask “What happens to the bag if the first card is red?” before allowing a calculator. Correct arithmetic with the wrong bag is a concept error, and repetition alone will not fix it.
Progress is observable when the pupil can transfer the rule from coloured tokens to selecting two different classroom book titles, choosing volunteers from a name list or sampling two defective items from a small batch. Use a new context and new numbers rather than making the child repeat the exact 5/12 example. Do not tell families that completing a probability drill guarantees a particular school grade.
What this model cannot decide
The model assumes each of twelve distinct cards is equally likely, removal is exact, and the second draw occurs only after the first card is set aside. If names are deliberately weighted, a card is accidentally returned, extra tickets exist or a selected person declines and is replaced, the model changes. In a small actual trial, observed frequencies need not equal theoretical probabilities. Nor does a random choice demonstrate which candidate is best suited to a task.
For more study, read the related Core Aim guide to repairing ambiguous multiple-choice questions for another example of checking a question’s conditions. The Mathematics in Bukit Timah guide to independence and fair classroom games connects the arithmetic to wider probability reasoning.
Frequently asked questions
Does “no winner twice” always mean without replacement?
If a name is physically removed or made ineligible for later draws, the simple without-replacement model applies. If a system instead redraws when a duplicate occurs, you must specify that mechanism. Under suitable equal-likelihood conditions the final distribution may be related, but the procedure itself still needs description.
Why do we multiply probabilities along a branch?
A complete ordered branch requires both stage events. The multiplication rule combines the first event’s probability with the probability of the second given that the first happened. When the first event changes the remaining pool, the conditional factor changes.
Can I simply add every branch probability?
Add only the routes that satisfy the requested event and do not overlap as complete outcomes. For “one Science and one Art”, Science→Art and Art→Science are different, non-overlapping ordered sequences. For “at least one Science”, a complement can be shorter and less error-prone.
Can the second draw be more likely to be Science than the first?
Yes, conditional on first drawing a non-Science card. Then all five Science cards remain in a pool of eleven, giving 5/11, which is larger than the initial 5/12. Conditional on first drawing Science, the second Science chance is 4/11 instead. Naming the condition resolves the apparent contradiction.
The learning outcome is better questions, not bigger fractions
The central question is “What changed after the first event?” That simple habit supports probability, experimental design and fair procedures. Explore the eduKate Education Hub, the Core Aim of Education learning hub and the diagnosis-first teaching logic of the immutable Clementi Mathematics reference for related approaches to independent checking. A learner who can reconstruct the bag at every stage can solve unfamiliar selection problems without guessing which memorised formula to use.