Modelling with probability

Modelling with probability

Spec M3 · AQA A-Level Maths: Revision notes

Studying this course? Keep your progress in one place.

Key facts

  • A probability model simplifies a real situation using stated assumptions.
  • Typical assumptions: independent trials, constant probability, a fixed number of trials.
  • State assumptions in context, not just 'independence'.
  • Independent trials: multiply; different arrangements: add. Do not forget the number of arrangements.
  • To critique a model, name the assumption, say why it fails, and say which way the answer moves.

What a model is

A probability model is a simplified description of a real situation, using assumptions so probabilities can be calculated.

Example: a school models each of 600 students as absent with probability 0.060.06, independently. This is a binomial-type model.

State each assumption in context, for example 'whether one student is absent does not affect whether another is absent'.

  1. 1

    Real situation

    e.g. student absence

  2. 2

    State assumptions

    independent, constant probability, fixed number of trials

  3. 3

    Model and predict

    e.g. each of 600 students absent with probability 0.06

  4. 4

    Compare with data

    then refine the model

Then back to step 1

Building and testing a probability model.

Typical assumptions

  • Trials are independent
  • Probability of success is constant
  • Outcomes are equally likely
  • Number of trials is fixed

In context: student absence

  • One absence does not affect another
  • Every student has probability 0.06
  • Each of the 600 students is one trial

Which is the best way to state the independence assumption for the absence model?

Working within a model

Once assumptions are stated, multiply for independent trials and add for different arrangements.

For three buses, each late with probability 0.150.15: none late is 0.853=0.6140.85^3=0.614, exactly one late is 3×0.15×0.852=0.3253\times0.15\times0.85^2=0.325.

Remember the factor of 3 for the number of arrangements.

00.10.20.30.40.50.60.0030123Number of buses lateProbability
Three buses, each late with probability 0.15: the probability of 0, 1, 2 or 3 being late.
  • Independent trialsmultiply
  • Different arrangementsadd

Worked example

A footballer scores each penalty with probability 0.8, independently. Find the probability of scoring exactly two, all three, and at least two of three penalties.

Three buses are each late with probability 0.150.15, independently. What is P(all three late)P(\text{all three late})?

Critiquing assumptions

Take an assumption and ask whether it is realistic, then say which way the answer moves.

Independence fails when events share a cause: buses delayed by the same road closure, students infected in the same outbreak. Constant probability fails when it changes with time, place or group. Replacement matters for small populations.

Positively linked events make extreme outcomes (all three late) more likely than the model says, so the model underestimates them.

00.020.040.060.080.10.120.140.16With replacementWithout replacementMethodProbability
Probability of drawing two red counters from 10 (4 red): replacement changes the answer.

With replacement

  • Probabilities stay the same
  • Events are independent
  • Two red from 10 (4 red): 410×410=0.16\frac4{10}\times\frac4{10}=0.16

Without replacement

  • Probabilities change
  • Events are not independent
  • Two red from 10 (4 red): 410×39=0.133\frac4{10}\times\frac39=0.133

Buses are delayed by the same road closure. How does this affect the probability that all three are late, compared with the independent model?

More realistic assumptions

A refinement changes an assumption to fit reality better, but is harder to calculate with. Decide whether the simple model is good enough.

A refinement is usually a trade-off: for large populations a simple model may be good enough. For 100 counters with 40 red, 40100×3999=0.158\frac{40}{100}\times\frac{39}{99}=0.158, very close to 0.160.16.

00.20.40.60.8Simple modelRefined modelModelProbability
P(at least two scores in three penalties): simple model against refined model

Worked example

A penalty-taker scores with probability 0.8 after a score and 0.6 after a miss. The first penalty is scored with probability 0.8. Find the probability of at least two scores in three penalties.

When is it reasonable to ignore the lack of replacement?

Try an exam question

Four trains each run late with probability 0.10.1, independently. Find the probability that exactly one is late. State one assumption of the model in context, and give a reason it might not hold.

[4 marks]

That's the notes covered.

Carry on to the next subtopic.

Try a question

Which of these is an assumption made by the model?

1 mark, marked against the mark scheme

Revision notes