4.5 · SLPaper 1Paper 2

Probability foundations and expected value

33 min read

In a minute

Probability measures how likely an is, on a scale from 00 to 11. When the outcomes are equally likely you count them: P(A)=n(A)n(U)\mathrm{P}(A) = \frac{n(A)}{n(U)}. When you can't count, you estimate from data, and either way you can predict how often the event should happen in many trials.

You'll be able to

  • Use the words trial, outcome, , event and equally likely, and list or tabulate a sample space.
  • Find a theoretical probability by counting equally likely outcomes, including from a two-way table of survey data.
  • Estimate a probability from data with , and tell it apart from a theoretical probability.
  • Use , P(A′)=1−P(A)\mathrm{P}(A') = 1 - \mathrm{P}(A), including for "at least one".
  • Find and interpret the of an event in a given number of trials.
Before you start3 quick questions on what this builds on
Check yourself
Write 940\frac{9}{40} as a decimal.
Check yourself
Find 6%6\% of 350350.
Check yourself
Find 1−0.0641 - 0.064.

Trials, outcomes and sample spaces

You can't say which meal the next customer at a canteen will choose. You can say exactly which meals they could choose. Probability starts with that list.

For one roll of a fair six-sided die, U={1,2,3,4,5,6}U = \{1, 2, 3, 4, 5, 6\} and n(U)=6n(U) = 6. The event AA, "an odd number", is A={1,3,5}A = \{1, 3, 5\}, so n(A)=3n(A) = 3.

When a trial has two stages, write each outcome as a pair. A list works for a few pairs; a table is quicker for more, and you can't miss any.

Worked exampleA school canteen's meal deal is one main (pasta, curry or wrap) and one drink (water or juice). A student picks a main and a drink at random. (a) List the sample space. (b) Write down n(U)n(U).

Try it first, then reveal one step at a time.

  1. Write each outcome as a pair: the main, then the drink. Take each main in turn and go through both drinks.
Worked exampleIn a board game you roll two fair four-sided dice, each numbered 11 to 44. Your score is the larger of the two numbers, or that number if both are the same. (a) Draw a table of the sample space. (b) Let FF be the event "the score is 44". Find n(F)n(F).

Try it first, then reveal one step at a time.

  1. (a) Put the first die down the side and the second along the top. Fill each cell with the larger of its row and column numbers.
Four by four grid of outcomes for two four-sided dice. Each cell shows the larger of the two numbers. The seven cells with score 4, the whole last row and last column, are shaded. Second die First die 1 2 3 4 1 2 3 4 1 2 3 4 2 2 3 4 3 3 3 4 4 4 4 4
The 1616 equally likely outcomes. The 77 shaded cells make up the event "the score is 44".
Check yourself
A fair six-sided die is rolled and a fair coin is tossed. Write down the number of outcomes in the sample space.
Check yourself
Use the table of scores for the two four-sided dice. How many outcomes give a score of 22?

Theoretical and experimental probability

In the board game, 77 of the 1616 equally likely cells give a score of 44. So you expect a score of 44 in 77 out of every 1616 rolls, in the long run. Counting like this gives a theoretical probability.

In the formula booklet
P(A)=n(A)n(U)\mathrm{P}(A) = \frac{n(A)}{n(U)}

This formula is in the formula booklet, so you don't need to memorise it. You do need its condition: it works only when the outcomes in UU are equally likely.

Every probability lies between 00 and 11. An impossible event has probability 00, and a certain event has probability 11. A negative answer, or one above 11, means a slip.

Worked exampleUse the table of scores for the two fair four-sided dice. Find the probability that the score is at least 33.

Try it first, then reveal one step at a time.

  1. Each die is fair, so the 1616 cells are equally likely and n(U)=16n(U) = 16.

Survey results are often given in a two-way table. Choosing one person "at random" makes every person equally likely, so n(U)n(U) is the grand total.

Worked exampleA gym asks 250250 members their age group and when they prefer to train. morningeveningunder 25385725 and over9164\begin{array}{l|cc} & \text{morning} & \text{evening} \\ \hline \text{under } 25 & 38 & 57 \\ 25 \text{ and over} & 91 & 64 \end{array} One member is chosen at random. Find the probability that this member prefers to train in the morning.

Try it first, then reveal one step at a time.

  1. Every member is equally likely to be chosen, so n(U)=250n(U) = 250, the grand total.
Check yourself
Use the gym table. One member is chosen at random. Find the probability that the member is under 2525 and prefers the evening.

Some events can't be counted. A bus is either on time or late, but the two are not equally likely. To estimate the probability, you collect data on many trials.

Learn this
relative frequency=number of times the event happensnumber of trials\text{relative frequency} = \frac{\text{number of times the event happens}}{\text{number of trials}}

A relative frequency is an experimental probability: an estimate from data. A theoretical probability comes from counting equally likely outcomes, with no experiment at all.

Relative frequency of heads after each toss of a fair coin, for 400 simulated tosses. It starts at 1, drops to about 0.23, then settles close to the dashed line at 0.5. 0 0.25 0.5 0.75 1 0 100 200 300 400 Number of tosses Relative frequency of heads
A simulation of 400400 tosses of a fair coin. The relative frequency of heads jumps about at first, then settles near the theoretical 0.50.5.

The more trials you do, the closer the relative frequency tends to get to the true probability. A GDC or spreadsheet can simulate thousands of trials in seconds.

Worked exampleA transport company tracks 12001200 journeys on one bus route. 846846 of them arrive on time. Estimate the probability that a journey on this route arrives on time.

Try it first, then reveal one step at a time.

  1. Arriving on time and arriving late are not equally likely, so you can't count outcomes. Use the data.
Worked exampleA phone game states that each mystery box holds a rare item with probability 0.050.05. Aiko opens 4040 boxes and gets 55 rare items. On a fan forum, players report 20002000 boxes with 108108 rare items. (a) Find each relative frequency of a rare item. (b) State which is the better estimate of the probability, with a reason. (c) Comment on the game's claim.

Try it first, then reveal one step at a time.

  1. (a) Aiko: 540=0.125\frac{5}{40} = 0.125. Forum: 1082000=0.054\frac{108}{2000} = 0.054.
Check yourself
A garden centre plants 360360 seeds of a new variety, and 297297 of them germinate. Estimate the probability that a seed of this variety germinates.
Check yourself
Which of these is a theoretical probability?

Complementary events

Each bus journey either arrives on time or it doesn't. Exactly one of the two happens. So their probabilities must add up to 11.

Every outcome is in exactly one of AA and A′A', so n(A)+n(A′)=n(U)n(A) + n(A') = n(U). Divide by n(U)n(U) to get the rule below.

In the formula booklet
P(A)+P(A′)=1\mathrm{P}(A) + \mathrm{P}(A') = 1

This is in the formula booklet. You will mostly use it rearranged: P(A′)=1−P(A)\mathrm{P}(A') = 1 - \mathrm{P}(A). It holds for estimated probabilities from data too, not just counted ones.

Worked exampleThe probability that a randomly chosen adult in a town holds a passport is 0.670.67. Let HH be the event "holds a passport". Find P(H′)\mathrm{P}(H') and say what it means.

Try it first, then reveal one step at a time.

  1. H′H' is "does not hold a passport". Every adult either holds one or doesn't, so HH and H′H' are complementary.

The complement is most useful for "at least one". "At least one" covers many cases. Its complement, "none", is a single case, and that is much quicker to count.

Worked exampleA bike lock has three dials, each showing a digit from 00 to 99. A code is set at random, with every code equally likely. Find the probability that the code contains at least one 00.

Try it first, then reveal one step at a time.

  1. Each dial has 1010 digits, so there are 10×10×10=100010 \times 10 \times 10 = 1000 equally likely codes. n(U)=1000n(U) = 1000.
Worked exampleA survey of 520520 randomly chosen households in a city finds that 143143 have no car. (a) Estimate the probability that a randomly chosen household in the city has at least one car. (b) Comment on how reliable your estimate is.

Try it first, then reveal one step at a time.

  1. (a) Let CC be "at least one car". Its complement C′C' is "no car", and the survey estimates P(C′)=143520\mathrm{P}(C') = \frac{143}{520}.
Check yourself
A fair four-sided die numbered 11 to 44 is rolled three times. Find the probability of getting at least one 11.
Check yourself
The probability that a randomly chosen phone charger from a large batch is faulty is 0.0260.026. Find the probability that it is not faulty.

Expected number of occurrences

About 0.7050.705 of the bus journeys arrive on time. So in 200200 journeys you would expect about 0.705×200=1410.705 \times 200 = 141 on time. The real count might be 135135 or 148148, but 141141 is the long-run average.

Learn this
expected number of occurrences of A=n×P(A),where n is the number of trials\text{expected number of occurrences of } A = n \times \mathrm{P}(A), \quad \text{where } n \text{ is the number of trials}

This one is not in the booklet, so learn it. The expected number need not be a whole number. It is an average over many repeats, so 22.122.1 is a perfectly good answer.

Worked exampleAfter a vaccine, the probability that a person has a sore arm the next day is 0.0650.065. A clinic vaccinates 340340 people. Find the expected number of people with a sore arm the next day.

Try it first, then reveal one step at a time.

  1. Each person is one trial, so n=340n = 340, and P(sore arm)=0.065\mathrm{P}(\text{sore arm}) = 0.065.
Worked exampleUse the bus company's data from earlier: 846846 of 12001200 journeys arrived on time. Next month the route will run 950950 journeys. Estimate the number of these journeys that will not arrive on time.

Try it first, then reveal one step at a time.

  1. No theoretical probability is given, so use the relative frequency: P(T)≈8461200=0.705\mathrm{P}(T) \approx \frac{846}{1200} = 0.705.
Worked exampleAt a school fair, a game uses the two fair four-sided dice from the board game. A player wins a prize when the score is 44. During the fair, 120120 people play once each. (a) Find the expected number of winners. (b) The organiser buys 5353 prizes. Comment on this decision.

Try it first, then reveal one step at a time.

  1. (a) From the table, P(score=4)=716\mathrm{P}(\text{score} = 4) = \frac{7}{16}.
Check yourself
The probability that an online customer leaves a review is 0.120.12. A shop has 450450 customers this week. Find the expected number who do not leave a review.
Check yourself
In a box of 250250 cables, the expected number of faulty cables is 66. Find the probability that a randomly chosen cable is faulty.

Common mistakes

  • Treating outcomes as equally likely when they are not, such as the scores from two dice. List or tabulate the outcomes that really are equally likely, then count those.
  • Using the wrong total in a two-way table. When anyone in the survey can be chosen, the denominator is the grand total, not a row or column total.
  • Taking the complement of "at least one" as "exactly one", or of "at least 33" as "at most 33". The complements are "none" and "at most 22".
  • Calling a relative frequency the exact probability. It is an estimate, and one from a few trials can be far off.
  • Rounding too early or too far, such as 0.43750.4375 to 0.440.44. Give an exact value or 3 s.f. (0.4380.438).
  • Rounding an expected number to a whole number, such as 22.122.1 to 2222. An expected number is an average, so give the exact value or 3 s.f.

Exam tips

  • Write down a probability from a table: no working is needed, but give an exact fraction or 3 s.f., and check it lies between 00 and 11.
  • List or tabulate a sample space: show every outcome, in pairs for a two-stage trial, so the examiner can see none is missing.
  • Comment on an experimental result: compare the relative frequency with the stated or theoretical probability using both numbers, and say which estimate uses more trials.
  • Find an expected number: write n×P(A)n \times \mathrm{P}(A) with the numbers in, then the value, exact or to 3 s.f., not forced to a whole number.

Key takeaways

Learn thisA trial has outcomes; the sample space UU lists them all; an event is a set of outcomes. For two stages, list pairs or draw a table.
In bookletP(A)=n(A)n(U)when the outcomes in U are equally likely\mathrm{P}(A) = \frac{n(A)}{n(U)} \quad \text{when the outcomes in } U \text{ are equally likely}
Learn thisEvery probability is between 00 and 11. From a two-way table with one person chosen at random from everyone in it, divide by the grand total.
Learn thisRelative frequency =times the event happensnumber of trials= \frac{\text{times the event happens}}{\text{number of trials}} is an experimental estimate of a probability. More trials usually give a better estimate.
In bookletP(A)+P(A′)=1\mathrm{P}(A) + \mathrm{P}(A') = 1
Learn thisFor "at least one", use 1−P(none)1 - \mathrm{P}(\text{none}).
Learn thisExpected number of occurrences =n×P(A)= n \times \mathrm{P}(A) for nn trials. It is a long-run average, so it need not be a whole number.

Practise this

24 past-paper questions on Probability and types of events in the AI HL bank, each with a mark scheme and a worked explanation.

Practise Probability and types of events →