Discrete random variables and expected value
In a minute
A is a number decided by chance, such as the number of cars a household owns. Its gives each value with its probability, and the probabilities add up to . The is the long-run average: it predicts totals, sets fair prices and tells you whether a game is fair.
You'll be able to
- Work with a and its , given as a table or as a formula.
- Use the fact that the probabilities add up to to find an unknown, and find probabilities such as .
- Find the by hand or with a GDC, and interpret it in context.
- Use in problems: decide whether a game is a , find a prize or fee that makes it fair, and predict a total over many trials.
Before you start3 quick questions on what this builds on
Revise first: 4.3 Central tendency and dispersion and 4.5 Probability foundations and expected value
Random variables and their distributions
A town survey picks a household at random and counts its cars. Before the survey you can't know the count. But you can list every value it could take, and how likely each one is.
The symbol means "add up over every value of ". This rule is not in the formula booklet. It is how you find a missing probability or an unknown such as .
Try it first, then reveal one step at a time.
- (a) The probabilities add up to : .
- So and . It lies between and , as a probability must.
- (b) "At least " means , or : .
- Check with the complement: .
- Interpret: of households own two or more cars. In households, expect about of them.
Try it first, then reveal one step at a time.
- (a) Write the two facts as equations. The probabilities add up to : , so .
- "Three times" gives .
- Substitute: , so and . Then .
- Check: , and each value lies between and .
- (b) The model says no learner ever needs more than attempts. Some learners do, so the real distribution has a longer tail.
Distributions given by a formula
A distribution can also be given by a formula, such as for . The set after "for" lists the values can take. The formula gives the probability of each one.
A formula is a short way to write a table. Substitute each value of , write the table out, and then work exactly as before.
Try it first, then reveal one step at a time.
- Substitute each value:
- (a) The probabilities add up to : , so .
- (b) .
- (c) Each week is one trial, so use : weeks.
- Interpret: about weeks a year, on average. It is a long-run average, so it need not be a whole number.
Try it first, then reveal one step at a time.
- Substitute each value: gives , gives , gives and gives .
- (a) Add and set equal to : . The bracket is , so .
- The table is and these add up to .
- (b) At most one means or : .
- Interpret: on about of days the hotel gets no more than one complaint.
The values of don't have to be positive or whole. Only the probabilities are restricted: each one lies between and , and they add up to .
Expected value
Go back to the car survey. In households you'd expect about with no car, with one, with two, with three and with four. That is cars, or cars per household.
Dividing by turns each count back into its probability. So the average is : each value times its probability, added up.
This formula is in the formula booklet. It is the mean of a frequency table from 4.3, with probabilities in place of frequencies.
No household owns cars. That is fine: is an average, so it need not be a value can take. It must lie between the smallest and largest values of .
On a GDC, use one-variable statistics with the values as data and the probabilities as frequencies. The mean is , and the total frequency should be , which checks the table. If your model refuses decimal frequencies and every probability has at most two decimal places, multiply every probability by first: the mean doesn't change, and should then be .
Try it first, then reveal one step at a time.
- (a) Write the sum with every term in: .
- Work it out: . On the GDC: values to in a list, probabilities as frequencies, and with .
- Interpret: households in the town own cars on average.
- (b) cars.
- Assumption: the survey's distribution holds for every household in the town.
Try it first, then reveal one step at a time.
- The probabilities add up to : , so .
- Write with every term in: , so .
- Solve the two equations on the GDC (or subtract them): and .
- Check: , and .
- Each passenger checks bags on average, so passengers check about bags.
Games, fairness and other applications
Many games cost money to play and sometimes pay a prize. Let be a player's gain in one game: prize minus cost. A loss is a negative gain.
is then the player's average gain per game in the long run. If it is positive, players come out ahead on average. If it is negative, the organiser does, which is how stalls, lotteries and casinos make money.
Over games, the expected total gain is . The same idea predicts totals anywhere: cars in a town, bags on a flight, an insurer's payouts.
Try it first, then reveal one step at a time.
- Take the $2 off each prize to get the gains: , and .
- The probabilities are , and . So
- (a) .
- , so the game is not fair. On average a player loses $0.76 per go.
- (b) Each go gives the stall the player's loss, so its expected profit is dollars.
Try it first, then reveal one step at a time.
- Let be the gain. Take the $4 off each prize: The last probability is .
- A fair game needs : .
- Expand: , so and .
- Check another way: a game is fair when the expected prize equals the cost. , the cost of a go.
- Interpret: a prize of $15.50 makes the game fair. The club would set a smaller prize to raise money.
Common mistakes
- Averaging the values of to find , as if they were equally likely. Weight each value by its probability: .
- Setting the sum of the -values equal to , or dropping from some terms. It is the probabilities that add up to .
- Treating as . For whole-number values, "greater than " starts at .
- Using the prizes instead of the gains when deciding whether a game is fair. Take the cost off every prize, including the prize of .
- Concluding "fair" or "not fair" without writing the value of . The value and its comparison with must both be seen.
- Rounding before multiplying by the number of trials. Multiply the full value, then round the final answer.
Exam tips
- Find : write with every term in, then solve. Check each probability lies between and .
- Find : write the sum with its numbers in before the value, even when the GDC does the arithmetic. That line earns the method mark.
- Determine whether a game is fair: define as the gain, find , write its value, and state whether it equals .
- Interpret : say it is a long-run average per trial, in the context's units, such as "a player loses $0.76 per go on average".
Key takeaways
Practise this
10 past-paper questions on Discrete probability distributions in the AI HL bank, each with a mark scheme and a worked explanation.