JC2 · H2 Mathematics · Notes
Construct probability tables from stopping rules and outcomes, recover moments, handle independent observations and evaluate net winnings.
Discrete Random Variables summary and formula sheet: every method on 4 pages
Mr Ian Ang’s summary of the whole discrete random variables chapter on four pages.
Page 1 of 4 Distributions, functions of X and tables without replacement
H2 Math summary notes
Discrete Random Variables formulas and methods, section by section
Page 1 of 4Distributions, functions of X and tables without replacement
Distribution, E(X), Var(X) and the GC: from the table to the mean, variance and standard deviation
- has the distribution in the table. Find , and . First check the table: .
- and .
- , and the standard deviation is . means the average value of when the experiment is repeated infinitely many times.
- On the GC: STAT, 1: Edit, with L1 and L2 . Then STAT, CALC, 1: 1-Var Stats with List: L1 and FreqList: L2, and Calculate.
- Read , then and .
- Warning: the GC gives decimals. If the question asks for an exact value, work by hand.
E of a function of X, and |X - mu| > sigma: list the values of x in the table that satisfy it
- Using the table of section 1, find (a) and (b) .
- (a) .
- GC: set L3 (L1 is 2nd 1), then 1-Var Stats with List L3 and FreqList L2 gives .
- (b) and . means or , so or , so or .
- . List the values of in the table that satisfy the condition.
Building the table, no replacement: multiply the branches, and count the arrangements
- A bag has 3 blue and 5 white counters. Three are taken without replacement and is the number of blue. Then and .
- and .
- Warning: for any order, multiply by the number of arrangements, for B, W, W. The table is for , and the total .
- A bag has 3 white and 2 black counters, removed one at a time until both black are out, and is the number removed. The last one is the 2nd black; arrange the rest. and .
- and .
- The table is for .
Page 2 of 4Unknown constants, linear results and grids
Unknown constants in the table: sum to 1, E(X), E(X squared), and solving with the GC
- takes the values in the table, with and . Find , and .
- From : , so . From : .
- From : . GC on (1), (2), (3): , , .
- Second question: , where is a positive integer. Find . Enter , because we move everything to one side and look for , then use 2nd GRAPH (TABLE).
- The table gives at , at and at . So (only an integer is accepted). Check that the finished table adds up to 1.
Linear results, several observations: E(aX + b), Var(aX + b) and independent observations of X
- and . The adds nothing to the spread: .
- and , always for independent observations of . have the same table as , so .
- Given and , with independent observations of : and .
- and .
- Warning: , but . Two observations are not one observation doubled.
Table of outcomes: two fair observations, and count the cells for each value of x
| cells | ||||
|---|---|---|---|---|
- Two fair four-sided dice (faces 1 to 4) are thrown. is the difference (larger minus smaller). Find the distribution of .
- Draw the grid with across the top and down. The rows of differences are ; ; ; .
- The dice are fair, so every cell is equally likely, with probability . Count the cells for each value of : for .
- So .
- .
Two observations: list the cases, with P(X = x) = kx for x = 1 to 5
- for , and are independent observations of . (a) , so .
- (b) and , so .
- (c) : list the pairs , , and the swaps. This is .
- Warning: "at least 3" includes a difference of exactly 3; "or vice versa" gives the factor 2.
- (d) : the pairs with sum 5 are , , , . So the probability is .
Page 3 of 4Biased events, fair games and the CLT
Table of outcomes: one biased event, where the cells are not equally likely
- A coin shows heads with probability . A spinner shows , , or , each with probability . is twice the spinner score if heads, and the spinner score if tails. Find the distribution of .
- In the grid the spinner values are (probability ), () and (). Heads, probability , gives and tails, probability , gives .
- The cells are not equally likely: . So , , and .
- The table is for . Check that the probabilities add up to 1.
- With the probabilities are .
Expected gains and fair games: gain is winnings minus amount paid, and a fair game has E(gain) = 0
- Gain winnings amount paid. A fair game has ; "on average" means .
- In the black counter game of section 3, a player pays for each counter removed and receives when both black are out. Find for a fair game. What is paid is dollars (not fixed) and what is received is (fixed).
- So , which gives .
- is the winnings of one game: with probability , with probability , with probability . Should a player play? . Since , she is expected to lose a game, so she should not play.
- If the winnings are and she pays for a fair game, then , so .
Many observations: the link to the CLT, for the sum and the mean
- For large (): and , approximately.
- is as in section 9, so and . For 60 plays, find . As is large, .
- So (3 s.f.).
- is the sum of observations of with and . ; find . Here .
- Enter . The table gives at and at , so .
Page 4 of 4Pictures: bar charts, a tree, two dice and the CLT curve
The distribution as a bar chart: heights are probabilities and E(X) is the balance point
- Draw one bar for each value of , with its height equal to : , , and for .
- The heights add up to .
- is the balance point of the bars. Here the balance triangle sits at .
Tree, no replacement: 3 blue and 5 white, three taken out, multiply along the branches
| Outcome | BBB | BBW | BWB | BWW | WBB | WBW | WWB | WWW |
|---|---|---|---|---|---|---|---|---|
| Probability |
- A bag has 3 blue and 5 white counters and three are taken out without replacement. The first branches are for B and for W.
- After B the branches are for B and for W; after W they are for B and for W. At the third draw the branches are after BB, after BW, after WB and after WW.
- Multiply along the branches. The eight outcomes BBB, BBW, BWB, BWW, WBB, WBW, WWB, WWW have probabilities .
- : BWW, WBW, WWB gives .
Two dice: the grid and the bars, with X as the larger minus the smaller
- is the larger score minus the smaller score on two dice, and the grid has 16 equal cells with both scores running from 1 to 4.
- The rows of the grid are ; ; ; .
- Cells of the same number have the same colour. The bars beside the grid have heights at .
- .
A game that loses on average: the bars of the winnings and the balance point
- is , or with probabilities , and , drawn as three bars at those values of .
- The balance triangle marks , just left of .
- : on average she loses.
The CLT curve for the mean: the area to the left of 0.5 under the normal curve
- The mean of 60 plays has and the question asks for .
- The curve is centred at the mean , and is marked on the axis to the right of it.
- Shade the area to the left of .
- .
Worked examples
Discrete Random Variables worked examples
Mr Ian’s worked examples for the chapter, with the full working.
Define the score before finding its distribution
A random variable assigns a number to every possible outcome. List its possible values first, then collect every outcome giving each value. A table is complete only when its probabilities are non-negative and sum to one. A stopping rule changes which paths are allowed.
Stop at the first success: distinguish the last two cases
At most three boxes are opened
One of six boxes contains a prize. Boxes are opened at random without replacement until the prize is found, with at most three boxes opened. Let X be the number of empty boxes opened. Find the distribution of X.
The possible values are 0, 1, 2 and 3. The value 2 means the prize is found third; the value 3 means all three attempts fail.
| x | 0 | 1 | 2 | 3 |
|---|---|---|---|---|
| P(X=x) |
The first three probabilities are equal because the prize’s position in a random ordering is equally likely to be any of the six positions. Positions 4–6 all lead to X = 3.
The standard deviation is . Therefore . X takes integer values: do not add a fraction of the probability at 2.
Variation: stop at the second zero
Five vouchers are numbered 0, 0, 1, 2 and 4. Draw randomly without replacement until the second zero is drawn. Let A be the sum of the numbers drawn. Find its distribution, then find the probability of a score of at least 5 given that at least four vouchers were drawn.
The last voucher must be zero. Before it, there is exactly one zero and a subset of 1, 2 and 4. Their subset sums give the possible scores 0 to 7.
For A = 1, the only allowed paths are (0, 1, 0) and (1, 0, 0). The path (0, 0, 1) is forbidden: the draw would already have stopped.
For A = 3, both 1 and 2 must appear before the final zero. The first three vouchers are 0, 1 and 2 in any of 3! orders. Every such order is allowed: none contains two zeros.
The same counts apply to other subsets of the same size. For A = 0, draw the two zeros first. For A = 7, arrange 0, 1, 2 and 4 before the final zero:
| Score a | Non-zero vouchers drawn | P(A = a) |
|---|---|---|
| 0 | None | |
| 1 | 1 | |
| 2 | 2 | |
| 3 | 1, 2 | |
| 4 | 4 | |
| 5 | 1, 4 | |
| 6 | 2, 4 | |
| 7 | 1, 2, 4 |
The probabilities sum to one. Let L be the number of vouchers drawn. At least four draws means that two or three non-zero vouchers appeared before the second zero, so the condition allows scores 3, 5, 6 and 7.
Scores 5, 6 and 7 also meet the condition. Their total probability is . Write q for the required conditional probability:
Fix the stopping draw before counting the earlier orders. An unrestricted arrangement can include a second zero too early.
Use total probability, mean and second moment
Recover three missing probabilities
X takes values with probabilities . Given and , find a, b and c.
Three unknowns need three equations. The variance gives the second moment:
Subtract the first equation from the simplified third:
All probabilities are valid and sum to one. In , square the values of X; do not square their probabilities.
A squared expression is not a variance
Given and , find .
A variance uses the mean as the centre. Here X − 3 has mean 1, so subtract its squared mean:
Subtracting a constant does not change spread. Equivalently expand the square, find , then subtract .
Build a distribution from equally likely outcomes
The smaller of two die scores
Two independent fair four-sided dice show 1, 2, 3 or 4. Let M be the smaller score, using the common score if they are equal. Find its distribution.
The event requires both scores to be at least m. Subtract the next tail to isolate one value:
| m | 1 | 2 | 3 | 4 |
|---|---|---|---|---|
| P(M=m) |
The 16 ordered die outcomes are equally likely; the four possible values of the minimum are not equally likely.
Independent observations: list the ordered pairs
An absolute difference uses both directions
Let , for . Find k, the mean and variance, then for two independent observations.
Pairs with a positive difference of at least 2 are . The reversed pairs have the same probability by identical distributions and independence:
Write q for the required probability.
A signed difference would give 0.15. The wording “absolute difference” decides whether reversed pairs belong.
Which ordered pairs belong in the event?
6 ordered pairs have |X₁ − X₂| ≥ 2. Their total probability is 0.30.
Condition on the sum of several scores
Three independent games each give score 0, 1 or 2 with probabilities 0.2, 0.5 and 0.3. Given a total score of 3, find the probability that the first score was 2.
The denominator contains and the six permutations of . For a first score of 2, only and remain.
Write , , and q for the required conditional probability.
The total condition destroys the original independence between scores. Use their independence to calculate unconditional path probabilities before taking the conditional ratio.
Winnings and a mean over many games
Translate the score into money before judging the game
Use the distribution from the preceding example. A game pays a gross prize of $3X and costs $8 to enter. Let Y be the player’s net gain. Find the mean and variance of Y, then approximate the probability that the mean net gain in 50 independent games exceeds $1.50.
The expected net gain is $1 per game, so the game favours the player in the long run. A positive expectation does not guarantee a profit in one session. For 50 independent games, by the Central Limit Theorem,
The normal model applies to the mean over many games. It does not make the four-valued score X normally distributed. More examples of totals, means and sample size appear in the Sampling guide.
Before you finish
Check the stopping condition, support and total probability. Use , square multipliers in variances, and multiply joint probabilities only when independence is justified. Give monetary conclusions using net gain.