Notes/Mathematics/Paper 5/Discrete Random Variables
CAIEA Level9709§5.4

Discrete Random Variables

Probability distribution tables built from grids, trees and selections, with E(X) and Var(X); the binomial distribution B(n, p) with its ranges, mean and variance; the geometric distribution Geo(p) with its ranges and mean; and the second-success and two-stage questions that combine them.

300 min read 14 sub-topics
183
question parts
2021–2025 · 37 papers
13 marks
per paper
≈ 26% of the paper
2.2/3
avg difficulty
moderate
#1
most examined
of 5 topics by marks

A discrete random variable is a number decided by chance that can only take separate, listable values: the number of heads when three coins are thrown, the total on two dice, the number of throws needed to get a 6. This note teaches how to write down its whole distribution, how to summarise it with a mean and a variance, and the two named distributions — binomial and geometric — that model repeated independent trials.

It is the heaviest topic on Paper 5. Across 2021–2025 it carried 490 marks over 183 tagged parts, about 13.2 of the 50 marks on every paper, and it appeared on all 37 papers in that window — usually as two or three separate questions:

topicmarks/paper
Discrete Random Variables13.2
The Normal Distribution11.6
Probability11.2
Permutations and Combinations9.8
Representation of Data9.3

Inside the topic the marks split three ways (a part can carry more than one tag):

sub-topicpartsmarkssections
Binomial distribution B(n, p) and geometric distribution Geo(p)11027807–09, 11–14
Probability distribution tables; E(X) and Var(X)6518001–06
Expectation and variance of the binomial; expectation of the geometric166810, 11

The questions are rarely hard, but they are long and full of small traps: a missing xx-value, "fewer than" read as "at most", a power one too high in a geometric probability, a variance with the square in the wrong place. Almost every mark comes from writing the method down in full — the mark schemes award one mark for a correct term and another for the complete unsimplified expression before the final answer. That is how the worked examples below are laid out.

Reading the notes in this topic.

  • Mark-scheme codes. M = a method mark (for a correct method, even with a slip in the numbers); A = an accuracy mark for a correct answer, which needs the M mark before it; B = a mark for a correct stated value or fact on its own. "M1" means one method mark.
  • Past-paper labels. "9709/52 M/J 2024 Q1(c)" means paper 5, variant 2, from the May/June 2024 series, question 1(c). F/M = February/March and O/N = October/November. Papers up to 2019 were called S1 and numbered 61, 62, 63.
  • Section numbers. "§07" means section 07 of this note. Other topics are named in words, e.g. "the Probability topic (syllabus 5.3)".

The "use a suitable approximation" parts that often follow a binomial question — replacing B(n,p)\mathrm{B}(n, p) with a normal distribution — belong to the Normal Distribution topic (syllabus 5.5) and are taught in that note; only the npnp and np(1−p)np(1-p) they start from are taught here (§10).

Before you start you should be able to
  • Counting with nCr^{n}C_r and counting arrangements (Permutations and Combinations topic, syllabus 5.2)

  • The multiplication law for independent events, adding mutually exclusive events, the complement rule P(A′)=1−P(A)P(A') = 1 - P(A), conditional probability, tree diagrams (Probability topic, syllabus 5.3)

  • Solving linear and simple simultaneous equations

  • Using logarithms to solve an<ca^n < c (P1/P2 laws of logarithms)

  • Working confidently with fractions and decimals to at least 4 significant figures

By the end of this page you can
  • Draw up a probability distribution table from equally likely outcomes (grids for dice and spinners), from independent events with different probabilities, and from selections without replacement, checking that the probabilities sum to 1

  • Find an unknown constant kk (and a second constant) when P(X=x)P(X=x) is given by a formula

  • Calculate E(X)\mathrm{E}(X) and Var(X)\mathrm{Var}(X) from a table, and use a table to find probabilities of events, including conditional probabilities

  • Find two unknown probabilities in a table from ∑P(X=x)=1\sum P(X=x) = 1 and a given E(X)\mathrm{E}(X) or probability condition

  • Recognise when B(n,p)\mathrm{B}(n, p) is a suitable model, state its conditions in context, and use P(X=r)=nCr pr(1−p)n−rP(X=r) = {}^{n}C_r\,p^r(1-p)^{n-r}

  • Find "at least / fewer than / between" binomial probabilities, choosing the shorter of the direct sum and the complement

  • Find the smallest nn for which P(at least one success)P(\text{at least one success}) exceeds a target, using logarithms

  • Use E(X)=np\mathrm{E}(X) = np and Var(X)=np(1−p)\mathrm{Var}(X) = np(1-p), including working back to nn or pp

  • Recognise when Geo(p)\mathrm{Geo}(p) is a suitable model, use P(X=r)=p(1−p)r−1P(X=r) = p(1-p)^{r-1} and E(X)=1p\mathrm{E}(X) = \frac1p

  • Find geometric range probabilities using P(X>k)=(1−p)kP(X > k) = (1-p)^k

  • Find the probability that the second or third success happens on a given trial

  • Solve two-stage questions in which a probability found earlier becomes pp for a new binomial or geometric model

01

Random variables and distribution tables from equally likely outcomes

Syllabus requirement · §5.4

“

draw up a probability distribution table relating to a given situation involving a discrete random variable XX, and calculate E(X)\mathrm{E}(X) and Var(X)\mathrm{Var}(X).

”

A random variable is a number whose value is decided by chance. "The number of heads when three coins are thrown", "the total score on two dice" and "the number of throws needed to get a 6" are all random variables: you cannot say in advance what the number will be, but you can say how likely each possible value is.

Two pieces of notation run through the whole topic:

  • a capital letter such as XX names the random variable itself — "the number of heads";
  • a lower-case letter such as xx stands for one particular value it can take — x=0x = 0, x=1x = 1, and so on. So P(X=2)P(X = 2) is read "the probability that the number of heads is 2".

XX is discrete when its possible values can be listed one by one (usually whole numbers: 0,1,2,…0, 1, 2, \ldots), with gaps between them. Heights and times, which can take any value in a range, are continuous and belong to the Normal Distribution topic (syllabus 5.5).

A probability distribution table lists every value XX can take, with the probability of each:

xxx1x_1x2x_2⋯\cdotsxnx_n
P(X=x)P(X=x)p1p_1p2p_2⋯\cdotspnp_n

Every such table obeys two rules:

0⩽P(X=x)⩽1  for every x,∑P(X=x)=10 \leqslant P(X=x) \leqslant 1 \ \text{ for every } x, \qquad \sum P(X=x) = 1

The symbol ∑\sum (capital Greek sigma, also printed Σ\Sigma) means "add up over every value of xx in the table". So the second rule says the probabilities in the bottom row always add up to exactly 1, because XX must take one of its listed values. It is the most useful check in the topic: after building any table, add the bottom row.

What "draw up the probability distribution table" earns marks for

Mark schemes give the first mark for a table with the correct list of xx-values, and the rest for the probabilities, each linked to the right xx-value. So always list every possible value first (including values such as 00 that are easy to forget), then fill in the probabilities. Fractions do not need to be simplified; decimals must be exact or correct to at least 3 significant figures.

Drawing up a distribution table
  1. 1

    List the possible values of XX. Read the rule that defines XX and write down every number it can produce.

    This is the top row of the table, and it earns the first mark on its own.

  2. 2

    Work out how each value can happen. For equally likely outcomes, list them all (a grid is best for two dice or spinners) and count; for unequal probabilities, see §02; for selections without replacement, see §03.

    Each value of X is an event, and its probability comes from the outcomes that make up that event.

  3. 3

    Write each probability in the table under its value of xx.

    Keep fractions over the same denominator where you can — it makes the next step easy.

  4. 4

    Check that the bottom row adds to 1.

    If it doesn't, an outcome has been missed or counted twice. Fix it before using the table for anything else.

Equally likely outcomes: use a grid

When two fair dice or spinners are used together, every pair of results is equally likely. With a spinner of aa sides and one of bb sides there are a×ba \times b equally likely pairs, so

P(X=x)=number of pairs that give xa×bP(X = x) = \frac{\text{number of pairs that give } x}{a \times b}

The reliable way to count is a grid: one spinner along the top, the other down the side, and in each cell the value of XX that pair produces. Then count the cells holding each value.

Two spinners, X = the total score

A fair spinner AA has three sides numbered 1,2,31, 2, 3. A fair spinner BB has two sides numbered 1,21, 2. Both spinners are spun and XX is the sum of the two scores. Draw up the probability distribution table for XX.

Show full working
Each cell = score on A + score on Bspinner Bspinner A12123233445counts (6 cells)X = 2: 1 cell → 1/6X = 3: 2 cells → 2/6X = 4: 2 cells → 2/6X = 5: 1 cell → 1/6the shaded 3 and 4 happen two ways each; 2 and 5 only one

Spinner A down the side, spinner B along the top; each cell holds the total for that pair. Counting equal cells gives the table.

  1. 1

    Count the equally likely pairs. Spinner AA has 33 equally likely scores and spinner BB has 22, so there are 3×2=63 \times 2 = 6 equally likely pairs.

    The spinners are fair and independent, so every (A, B) pair has the same probability, 1/6.

  2. 2

    Fill in the grid of totals. Row A=1A=1: 1+1=21+1=2, 1+2=31+2=3. Row A=2A=2: 2+1=32+1=3, 2+2=42+2=4. Row A=3A=3: 3+1=43+1=4, 3+2=53+2=5.

    Writing the value of X in every cell is what stops a pair being missed.

  3. 3

    List the possible values of XX. The totals in the grid are 2,3,4,52, 3, 4, 5.

    The smallest total is 1 + 1 and the largest is 3 + 2, and every whole number between appears.

  4. 4

    Count the cells for each value. X=2X=2: 11 cell. X=3X=3: 22 cells. X=4X=4: 22 cells. X=5X=5: 11 cell.

    3 and 4 each happen in two different ways, so they are twice as likely as 2 or 5.

  5. 5

    Divide each count by 66. P(X=2)=16,P(X=3)=26,P(X=4)=26,P(X=5)=16P(X=2)=\frac16, \quad P(X=3)=\frac26, \quad P(X=4)=\frac26, \quad P(X=5)=\frac16

    Leaving them over 6 keeps the check in the next step simple.

  6. 6

    Check the sum. 1+2+2+16=66=1 ✓\frac{1+2+2+1}{6} = \frac66 = 1 \ \checkmark

    The counts must add to the number of cells in the grid, 6.

Answer
xx22334455
P(X=x)P(X=x)16\frac1626\frac2626\frac2616\frac16

The same grid works for any rule — sum, product, difference, 'the higher score' — only the number written in each cell changes.

A rule with two cases, on two dice

9709/55 M/J 2025 Q1(a)3 marks

Two fair 6-sided dice with faces labelled 1,2,3,4,5,61, 2, 3, 4, 5, 6 are thrown. The two scores are noted. The random variable XX is defined as follows.

  • If the two scores are equal, X=0X = 0
  • If the scores are not equal, XX is the larger score minus the smaller score

Draw up the probability distribution table for XX.

Show full working
Each cell's value is |first score − second score|second diefirst die123456123456123451123421123321124321154321000000countsX = 0: 6/36X = 1: 10/36X = 2: 8/36X = 3: 6/36X = 4: 4/36X = 5: 2/36X = 0 is the 6-cell diagonal; then 10, 8, 6, 4, 2 cells

The 36 equally likely pairs, each labelled with X. The diagonal (equal scores) gives X = 0 in 6 cells; the bands on either side then give 10, 8, 6, 4 and 2 cells for X = 1 to 5.

  1. 1

    Count the equally likely pairs. Each die has 66 faces, so there are 6×6=366 \times 6 = 36 equally likely ordered pairs.

    (2, 5) and (5, 2) are different outcomes — the first die and the second die are different objects — so both appear in the grid.

  2. 2

    Read the rule as one formula. For equal scores X=0X=0, and "larger minus smaller" is also 00 when the scores are equal. So in every cell, X=∣first−second∣X = |\text{first} - \text{second}|, the difference between the two scores ignoring its sign.

    Spotting that the two cases join up into one rule makes the grid quick to fill.

  3. 3

    List the possible values of XX. The difference can be 0,1,2,3,40, 1, 2, 3, 4 or 55.

    The biggest possible difference is 6 − 1 = 5.

  4. 4

    Count the cells for each value from the grid. X=0X=0: the diagonal, 66 cells. X=1X=1: 55 cells above the diagonal and 55 below, 1010 cells. X=2X=2: 4+4=84+4=8. X=3X=3: 3+3=63+3=6. X=4X=4: 2+2=42+2=4. X=5X=5: 1+1=21+1=2.

    Each band of equal differences runs parallel to the diagonal, one cell shorter on each side every time you move out.

  5. 5

    Divide each count by 3636. P(0)=636, P(1)=1036, P(2)=836, P(3)=636, P(4)=436, P(5)=236P(0)=\frac{6}{36},\ P(1)=\frac{10}{36},\ P(2)=\frac{8}{36},\ P(3)=\frac{6}{36},\ P(4)=\frac{4}{36},\ P(5)=\frac{2}{36}

    The mark scheme accepts these unsimplified, or as 1/6, 5/18, 2/9, 1/6, 1/9, 1/18.

  6. 6

    Check the sum. 6+10+8+6+4+236=3636=1 ✓\frac{6+10+8+6+4+2}{36} = \frac{36}{36} = 1 \ \checkmark

    The counts add to 36, the number of cells.

Answer
xx001122334455
P(X=x)P(X=x)636\frac{6}{36}1036\frac{10}{36}836\frac{8}{36}636\frac{6}{36}436\frac{4}{36}236\frac{2}{36}

Part (b) of this question goes on to find E(X) and Var(X) from this table — it is worked in §05.

Dice and spinners with repeated numbers

Some questions use a die with faces such as 1,2,2,3,3,31, 2, 2, 3, 3, 3. It is still a fair die with six equally likely faces — the two faces marked 22 are different faces that happen to show the same number. Give the grid one row (or column) per face, not per number, so the grid for two such dice is still 6×66 \times 6. Then P(score 3)=36P(\text{score } 3) = \frac36, not 13\frac13.

Shortcut: faces showing the same number can be grouped into one block of the grid, as long as the block keeps all its cells. For two dice with faces 1,1,1,2,2,21, 1, 1, 2, 2, 2 and 1,1,2,2,3,31, 1, 2, 2, 3, 3, the pair "red 2, blue 3" is a block of 3×2=63 \times 2 = 6 of the 3636 cells, so it has probability 636\frac{6}{36}.

Common mistakes
  • Counting (2,5)(2, 5) and (5,2)(5, 2) as one outcome, so the grid for two dice has only 2121 cells

    Use ordered pairs: two dice give 3636 equally likely outcomes

    The 21 'unordered' outcomes are not equally likely — a double can happen one way, a non-double two ways.

  • Leaving a value out of the top row because it looks special, e.g. X=0X=0

    List every value the rule can produce before filling in probabilities

    The first mark is for the correct set of x-values, so a missing value costs marks straight away.

  • Giving a die with faces 1,1,2,2,3,31, 1, 2, 2, 3, 3 only three rows in the grid

    One row per face: six rows, each with probability 16\frac16 — or one block of two rows per number

    Three single rows would treat the die as having three faces with the wrong cell counts; each number must keep the 2 faces it really has.

Your turn

Draw the grid every time, even when you think you can see the counts.

  1. 19709/52 M/J 2021 Q4(a)3 marks

    A fair spinner has sides numbered 1,2,21, 2, 2. Another fair spinner has sides numbered −2,0,1-2, 0, 1. Each spinner is spun. The number on the side on which a spinner comes to rest is noted. The random variable XX is the sum of the numbers for the two spinners.

    Draw up the probability distribution table for XX.

    Stuck? Show hint

    A 3 by 3 grid — give the first spinner one row per side, so the number 2 gets two rows.

    Show solution
    1. 1

      Count the pairs. 3×3=93 \times 3 = 9 equally likely pairs.

      Each spinner has three equally likely sides.

    2. 2

      Row for the side marked 1: 1+(−2)=−11+(-2)=-1,  1+0=1\ 1+0=1,  1+1=2\ 1+1=2.

      Add the first spinner's number to each of the second spinner's numbers.

    3. 3

      Row for the first side marked 2: 2+(−2)=02+(-2)=0,  2+0=2\ 2+0=2,  2+1=3\ 2+1=3.

    4. 4

      Row for the second side marked 2: the same totals again, 0,2,30, 2, 3.

      Two different sides show 2, so this row appears twice — that is what makes totals involving 2 more likely.

    5. 5

      Count each value. X=−1X=-1: 11. X=0X=0: 22. X=1X=1: 11. X=2X=2: 33. X=3X=3: 22.

      The counts add to 9, the number of cells.

    6. 6

      Divide by 9. P(−1)=19P(-1)=\frac19, P(0)=29P(0)=\frac29, P(1)=19P(1)=\frac19, P(2)=39P(2)=\frac39, P(3)=29P(3)=\frac29.

      Check: 1 + 2 + 1 + 3 + 2 = 9, so the probabilities add to 1.

    Answer
    xx−1-100112233
    P(X=x)P(X=x)19\frac1929\frac2919\frac1939\frac3929\frac29
  2. 29709/53 O/N 2025 Q4(a)3 marks

    A fair red spinner has 4 sides, numbered 1,2,3,41, 2, 3, 4. A fair blue spinner has 4 sides, numbered 0,1,2,30, 1, 2, 3. When a spinner is spun, the score is the number on the side on which it lands. The two spinners are spun at the same time.

    The random variable XX denotes the higher of the two scores obtained. If the two scores are equal, then the value of XX is 00.

    Draw up the probability distribution table for XX.

    Stuck? Show hint

    A 4 by 4 grid of 16 cells. Mark the cells where the two scores are equal first — those are all X=0X=0.

    Show solution
    1. 1

      Count the pairs. 4×4=164 \times 4 = 16 equally likely pairs.

    2. 2

      Equal scores give X=0X=0. The scores can be equal only at 1,11,1, 2,22,2 and 3,33,3 (red has no 0, blue has no 4): 33 cells.

      Check the equal pairs before anything else, because the rule overrides 'the higher score' there.

    3. 3

      Red score 1. Blue 00 gives higher =1=1. (Blue 1 is equal. Blue 2 and 3 give higher 22 and 33.)

      Work across each row of the grid in turn.

    4. 4

      Red score 2. Blue 0,10, 1 give higher =2=2; blue 22 is equal; blue 33 gives 33.

    5. 5

      Red score 3. Blue 0,1,20, 1, 2 give higher =3=3; blue 33 is equal.

    6. 6

      Red score 4. Blue 0,1,2,30, 1, 2, 3 all give higher =4=4.

    7. 7

      Count each value. X=0X=0: 33. X=1X=1: 11 (red 1, blue 0). X=2X=2: red 1 blue 2, and red 2 blue 0 or 1: 33. X=3X=3: red 1 blue 3, red 2 blue 3, red 3 blue 0, 1, 2: 55. X=4X=4: 44.

      Check: 3 + 1 + 3 + 5 + 4 = 16.

    Answer
    xx0011223344
    P(X=x)P(X=x)316\frac{3}{16}116\frac{1}{16}316\frac{3}{16}516\frac{5}{16}416\frac{4}{16}
  3. 39709/53 O/N 2024 Q2(a)3 marks

    A red fair six-sided dice has faces labelled 1,1,1,2,2,21, 1, 1, 2, 2, 2. A blue fair six-sided dice has faces labelled 1,1,2,2,3,31, 1, 2, 2, 3, 3. Both dice are thrown. The random variable XX is the product of the scores on the two dice.

    Draw up the probability distribution table for XX.

    Stuck? Show hint

    Use the block shortcut from the callout: the 36-cell grid splits into blocks, one for each (red number, blue number) pair.

    Show solution
    1. 1

      Size of each block. Each red number is on 33 faces and each blue number is on 22 faces, so every (red number, blue number) block holds 3×2=63 \times 2 = 6 of the 3636 cells.

      Grouping equal faces is fine because each block keeps all of its cells.

    2. 2

      List the six blocks and their products. 1×1=11\times1=1,  1×2=2\ 1\times2=2,  1×3=3\ 1\times3=3,  2×1=2\ 2\times1=2,  2×2=4\ 2\times2=4,  2×3=6\ 2\times3=6.

      Two rows of red numbers times three columns of blue numbers gives six blocks.

    3. 3

      Count the cells for each product. X=2X=2 appears in two blocks: 6+6=126 + 6 = 12 cells. X=1,3,4,6X = 1, 3, 4, 6 each appear in one block: 66 cells.

      Products that come from different blocks are added together.

    4. 4

      Divide by 36. P(1)=636P(1)=\frac{6}{36}, P(2)=1236P(2)=\frac{12}{36}, P(3)=636P(3)=\frac{6}{36}, P(4)=636P(4)=\frac{6}{36}, P(6)=636P(6)=\frac{6}{36}.

      Check: 6 + 12 + 6 + 6 + 6 = 36.

    Answer
    xx1122334466
    P(X=x)P(X=x)636\frac{6}{36}1236\frac{12}{36}636\frac{6}{36}636\frac{6}{36}636\frac{6}{36}
  4. 49709/52 O/N 2024 Q3(a)3 marks

    A fair coin and an ordinary fair six-sided dice are thrown at the same time. The random variable XX is defined as follows.

    • If the coin shows a tail, XX is twice the score on the dice.
    • If the coin shows a head, XX is the score on the dice if the score is even and XX is 00 otherwise.

    Draw up the probability distribution table for XX.

    Stuck? Show hint

    There are 2×6=122 \times 6 = 12 equally likely (coin, dice) pairs. Make a 2-row grid: one row for tail, one for head.

    Show solution
    1. 1

      Count the pairs. 2×6=122 \times 6 = 12 equally likely outcomes.

    2. 2

      Tail row (twice the score): 2,4,6,8,10,122, 4, 6, 8, 10, 12.

      Each of these six outcomes has probability 1/12.

    3. 3

      Head row (even score kept, odd score gives 0): dice 1,2,3,4,5,61,2,3,4,5,6 give 0,2,0,4,0,60, 2, 0, 4, 0, 6.

    4. 4

      Count each value across both rows. X=0X=0: 33. X=2X=2: 22 (tail-1, head-2). X=4X=4: 22. X=6X=6: 22. X=8,10,12X=8, 10, 12: 11 each.

      Check: 3 + 2 + 2 + 2 + 1 + 1 + 1 = 12.

    Answer
    xx002244668810101212
    P(X=x)P(X=x)312\frac{3}{12}212\frac{2}{12}212\frac{2}{12}212\frac{2}{12}112\frac{1}{12}112\frac{1}{12}112\frac{1}{12}
Practise probability distribution tablesReal past-paper questions · Probability distribution tables; expectation E(X) and variance Var(X)

The rest of this note

Checking your access…

Can you do all of these?

  • List every possible value of X before any probability — the first mark is for the correct set of x-values

  • Use a grid of ordered pairs for two dice or spinners; give repeated faces their own rows

  • With biased coins, work out each scenario separately; only multiply by a number of orders when every order has the same probability

  • Without replacement: count selections with nCr for a handful, or multiply changing fractions for 'until' — never use the geometric formula

  • Add the bottom row of every table: it must be exactly 1

  • P(X = x) = kx²: substitute each x (a negative x squared is positive), sum to 1, solve for k, then give numbers

  • Var(X) = E(X²) − [E(X)]²; square x (not p) in E(X²); keep fractions exact

  • Two unknown probabilities: one equation from sum to 1, one from E(X) or the given condition

  • Binomial needs fixed n, two outcomes, constant p and independence; state conditions in context

  • Translate 'at least / more than / at most / fewer than' into a list of x-values, then take the shorter of the list and its complement

  • Write the unsimplified binomial terms before the decimals, keep 5–6 s.f. in each term, round only the answer

  • Smallest n: 1 − (1 − p)ⁿ > target; dividing by a negative log flips the inequality; round up

  • E(X) = np and Var(X) = np(1 − p); if both n and p are unknown, divide the variance by the mean

  • Geometric: no counting factor, r − 1 failures, E(X) = 1/p; for ranges ask 'how many trials must all fail?'

  • r-th success on trial n: choose r − 1 positions from the first n − 1 trials only, then multiply by p for the last trial

  • Two-stage questions: decide what one trial is, carry the earlier p unrounded

Now do the questions
183 real Paper 5 parts from 2021–2025, sorted by difficulty, with mark schemes