Binomial distribution
In a minute
A counts how many times something happens in a fixed number of independent tries, each with the same chance. You write , find its probabilities on your GDC, and use for the number you expect.
You'll be able to
- Decide whether a binomial distribution fits a situation, write it as , and state its assumptions in context.
- Find with your GDC, and see where it comes from in a small case.
- Turn "at most", "at least", "fewer than", "more than" and "between" into the right GDC calculation.
- Find and interpret the mean and of .
- Use as an , including for batches that meet a condition.
Before you start3 quick questions on what this builds on
Revise first: 4.3 Central tendency and dispersion, 4.5 Probability foundations and expected value, 4.6 Probability diagrams, rules and independence and 4.7 Discrete random variables and expected value
When is a situation binomial?
How many of batteries are faulty? How many of patients recover? Each question counts successes in a fixed number of tries. When every try behaves the same way, a binomial distribution gives the probability of each count.
"Success" is only a label for the outcome you count: it can be a faulty battery. Independent means one trial's result doesn't change the chances for any other, as in 4.6.
In real situations these conditions are assumptions of the model. Questions often ask you to state one, or to judge whether it is reasonable.
Try it first, then reveal one step at a time.
- Each battery picked is a trial: faulty (the success you count) or not. There are trials.
- (a) Each battery is faulty with probability , so .
- (b) Same probability, , for every battery picked. They are picked without replacement, but from thousands, so picks barely change the proportion. Reasonable.
- Independence: one battery being faulty doesn't affect another. This could fail if a machine fault produces a run of bad batteries.
Try it first, then reveal one step at a time.
- (a) She stops when she succeeds, so the number of throws is not fixed.
- (b) There are trials, each rainy or dry. But wet weather comes in spells, so the days are not independent. A binomial model is doubtful.
- (c) The first pick is a girl with probability . If it is, the second is a girl with probability . The probability changes from pick to pick.
- (a) No: the number of trials is not fixed. (b) Doubtful: the days are not independent. (c) No: the probability of a girl changes from pick to pick.
The probability of exactly r successes
Before you hand the work to your GDC, see where a binomial probability comes from. A small case is enough.
Try it first, then reveal one step at a time.
- Let be the number of red lights, so . Write R for red and G for not red, with .
- Two reds can happen in orders: RRG, RGR and GRR. The tree diagram below shows all paths.
- The lights are independent, so multiply along a path: . The other two orders use the same three numbers, so each is too.
- Add the paths: .
Two counts need only one path: , as every trial fails, and .
With more trials the number of orders grows fast: successes in trials can happen in orders. Your GDC's binomial pdf counts them and adds them up for you.
Try it first, then reveal one step at a time.
- Let be the number in the sample who cycle. There are trials, each a yes or no.
- The city is large, so picks hardly change the proportion. Take each time, independently: .
- Write down what you want: . Binomial pdf with , and gives
- Round only at the end: to 3 s.f.
- Interpret: in about of random samples of adults, exactly would cycle to work.
At most, at least and between
Most questions ask about a range of counts, such as "at most ". Your GDC's binomial cdf adds up to , giving in one step.
Everything else comes from and the . only takes whole numbers, so "fewer than " means , and "more than " means .
In the chart below, "at least " shades the bars from up. Its complement is the bars to , so you subtract .
Probability =
Mean np = , variance np(1 − p) =
Try it first, then reveal one step at a time.
- Let be the number of ticket holders who turn up. Assume each turns up with probability , independently of the others. Then .
- (a) Someone has no seat when more than turn up. is a whole number, so that is .
- Use the complement: .
- Binomial cdf with , and gives So .
- (b) On about of such flights, a ticket holder has no seat. Independence is doubtful: families and groups turn up, or not, together.
Try it first, then reveal one step at a time.
- Let be the number returned late, so .
- "Between and inclusive" keeps and . Take , then remove to : .
- Binomial cdf twice:
- So the probability is to 3 s.f. Subtracting would wrongly remove .
Mean, variance and expected number
In 4.5 you found an as . A binomial count is the same idea. With trials and success probability , you expect successes.
is the mean of : the average number of successes if you repeated all trials many times. It need not be a whole number.
The measures how widely the count spreads around its mean. Its square root is the .
Both formulas are in the formula booklet; you don't need to prove them. Since , the variance is smaller than the mean whenever : a quick check.
Try it first, then reveal one step at a time.
- (a) Here and , so .
- Mean: .
- Variance: . Check: it is smaller than the mean.
- If a question asks for the standard deviation, it is .
- (b) Over many shifts of calls, the centre averages complaints per shift.
Try it first, then reveal one step at a time.
- (a) Let be the number of perfect croissants on a tray: .
- "At least " is , so .
- (b) Now each tray is one trial, with success probability
- Expected number of trays , using the unrounded probability.
- Interpret: on average, about of the trays sell at full price each week.
Common mistakes
- Writing . That throws away as well. Use .
- Reading "fewer than " as , or "more than " as . For a whole-number count they are and .
- Using a binomial model when picking without replacement from a small group. The probability changes after each pick.
- Rounding a GDC value before subtracting or multiplying with it. Carry the full value and round only the final answer.
Exam tips
- Find a binomial probability: write and the probability statement, such as , before the GDC value.
- State an assumption: name independence or the constant probability, in context.
- Give a probability exactly or to 3 s.f.; when you round, keep trailing zeros: for , not ; , not .
- Find an expected number: write , or times a probability, with the numbers in, then the value. Don't round it to a whole number.
Key takeaways
Practise this
26 past-paper questions on Binomial distribution in the AI HL bank, each with a mark scheme and a worked explanation.