Robinson Tuition
A-level Further Statistics 1 · Edexcel 9FM0 · free Further Maths worksheet

Size and power of a test

Size, power, power functions, comparing two tests, choosing a sample size.

12 original questions, 88 marks. Type your answers below for an instant mark and a score (algebra in any equivalent form, e.g. 2sqrt(3) or 2√3), or print the worksheet, mark it with the answers sheet, and use the solutions to see every step written out by hand.

Worksheet PDFWorked solutions PDFAnswers PDF

The questions

1

Owen thinks a coin is biased towards heads. He tosses it 12 times and tests H0: p = 0.5 against H1: p > 0.5, where p is the probability of a head. He will reject H0 if he gets 10 or more heads.

  1. Find the size of the test.
  2. Given that p = 0.7, find the power of the test.
4 marks
Show answer
a) 0.0193 b) 0.2528
2

The number of goals scored by a football team in a match has a Poisson distribution with mean 3. A new striker joins the team and the manager tests whether the mean number of goals per match has increased, using the number of goals in the striker's first match.

  1. Find the critical region for a test of H0: λ = 3 against H1: λ > 3 at the 5% level of significance.
  2. Write down the size of the test.
  3. Given that the mean is now 5 goals per match, find the power of the test.
5 marks
Show answer
a) X ≥ 7 b) 0.0335 c) 0.2378
3

A charity finds that 5% of people it phones agree to make a donation. A volunteer, Jamal, uses a new script and thinks this probability, p, has increased. He phones people one at a time until one agrees, and records the number of calls, X, including the successful one. He will reject H0: p = 0.05 in favour of H1: p > 0.05 if X ≤ c.

  1. Using a 10% level of significance, find the largest possible value of c and the size of the test.
  2. Write down the power function of the test in terms of p.
  3. Find the power of the test when p = 0.2
5 marks
Show answer
a) c = 2, size 0.0975 b) 1 - (1 - p)2 c) 0.36
4

A council says that 30% of households in a town recycle their food waste. A campaign group believes the proportion, p, is lower. They ask 20 randomly chosen households and test H0: p = 0.3 against H1: p < 0.3 at the 5% level of significance.

  1. Find the critical region for the test.
  2. Show that the power function of the test is (1 − p)18(1 + 18p + 171p2).
  3. Find the power of the test when p = 0.1
6 marks
Show answer
a) X ≤ 2 b) (1 - p)18[(1 - p)2 + 20p(1 - p) + 190p2] = (1 - p)18(1 + 18p + 171p2) c) 0.6769
5

The number of breakdowns per month of a printing machine has a Poisson distribution with mean λ. Over many years λ = 5. After a service, the engineer tests H0: λ = 5 against H1: λ < 5 at the 5% level of significance, using the number of breakdowns in one month.

  1. Find the critical region for the test.
  2. Show that the power function of the test is e−λ(1 + λ).
  3. Copy and complete the table, giving values to 4 decimal places.
    λ12345
    Power0.73580.19910.0404
  4. Sketch the power function on the axes, for 0 ≤ λ ≤ 5
  5. State the size of the test.
01234500.20.40.60.81λPower
7 marks
Show answer
a) X ≤ 1 b) P(X ≤ 1) = e(-λ) + λe(-λ) = e(-λ)(1 + λ) c) 0.4060 and 0.0916 d) A smooth decreasing curve from (0, 1) through the plotted values to (5, 0.0404) e) 0.0404
6

The time, in minutes, that customers wait on a phone line is normally distributed with mean μ and standard deviation 4. The company claims that μ = 50. A customer group thinks that μ is greater than 50. They record the waiting times of a random sample of 16 calls and test H0: μ = 50 against H1: μ > 50 at the 5% level of significance.

  1. Find the critical region for the sample mean X.
  2. Write down the size of the test.
  3. Find the power of the test when μ = 52.5
  4. State, with a reason, how the power at μ = 52.5 would change if a larger sample were used with the same significance level.
6 marks
Show answer
a) X̄ > 51.64 b) 0.05 c) 0.8038 d) Increase: 4/√n is smaller, so X̄ is more concentrated about 52.5 and more likely to be in the critical region
7

The probability that a seedling of a type of tomato plant survives being planted out is 0.3. A gardener, Priya, thinks this probability, p, is lower in her garden. She considers two tests of H0: p = 0.3 against H1: p < 0.3, each at the 5% level of significance.

Test A: plant 25 seedlings and count the number, X, that survive.
Test B: plant seedlings one at a time until one survives, and record the number planted, Y.

  1. Find the critical region and the size of Test A.
  2. Find the critical region and the size of Test B.
  3. Find the power of each test when p = 0.1
  4. State, giving your reasons, which test you would recommend when p = 0.1
9 marks
Show answer
a) X ≤ 3, size 0.0332 b) Y ≥ 10, size 0.0404 c) Test A 0.7636, Test B 0.3874 d) Test A: smaller size (less chance of a Type I error) and much higher power (less chance of a Type II error)
8

The volume of shampoo in a bottle is normally distributed with mean μ ml and standard deviation 12 ml. A quality manager will test H0: μ = 200 against H1: μ < 200 at the 1% level of significance, using the mean X of a random sample of n bottles. The critical region is X < c.

  1. Find an expression for c in terms of n.
  2. The manager wants the power of the test to be at least 0.9 when μ = 195. Find the smallest value of n that achieves this. Show your working clearly.
  3. For this value of n, find the value of c.
8 marks
Show answer
a) c = 200 - 2.3263 × 12/√n = 200 - 27.916/√n b) 75 c) 196.78
9

A manufacturer requires that no more than 10% of the light bulbs it makes are faulty. An inspector, Tomasz, tests H0: p = 0.1 against H1: p > 0.1, where p is the proportion of faulty bulbs. D is the number of faulty bulbs in a random sample of 20

Test P: Tomasz takes a sample of 20.

  • If D ≥ 5 he rejects H0.
  • If D ≤ 3 he does not reject H0.
  • If D = 4 he takes a second sample of 20, and rejects H0 only if this second sample contains 3 or more faulty bulbs.
  1. Find the size of Test P.

Given that the proportion of faulty bulbs is actually 0.25

  1. find the power of Test P,
  2. find the expected number of bulbs inspected using Test P.

Test Q uses a single sample of 30 bulbs and rejects H0 if it contains 6 or more faulty bulbs. Test Q has size 0.0732 and, when p = 0.25, power 0.7974

  1. Suggest, giving your reasons, which test Tomasz should use.
9 marks
Show answer
a) 0.0722 b) 0.7575 c) 23.8 d) Test P: similar size and only slightly lower power, but fewer bulbs inspected on average (23.8 against 30)
10

The number of faults reported per day on a telephone network has a Poisson distribution with mean 2. An engineer wants to test whether the mean number of faults per day has increased, at the 5% level of significance.

Test A uses the number of faults, X, on one day.
Test B uses the number of faults, Y, over two days.

  1. Find the critical region and the size of each test.
  2. Write down the power function of Test A in terms of λ, the mean number of faults per day.
  3. Find the power of each test when λ = 3.5
  4. State, giving a reason, which test you would recommend.
9 marks
Show answer
a) A: X ≥ 6, size 0.0166; B: Y ≥ 9, size 0.0214 b) 1 - e(-λ)(1 + λ + λ²/2 + λ³/6 + λ⁴/24 + λ⁵/120) c) A: 0.1424, B: 0.2709 d) Test B: its size is still below 0.05 and its power is much higher
11

A spinner has five equal sections numbered 1, 2, 3, 4 and 5. Callum spins it 60 times and calculates the mean score, X. He tests the hypotheses

H0: the spinner is fair    H1: the spinner is not fair

and will reject H0 if X < 2.7 or X > 3.3

  1. Show that, if the spinner is fair, the score on one spin has mean 3 and variance 2
  2. Using a suitable approximation, show that the size of the test is 0.100 to 3 significant figures.
  3. Given that the spinner is actually biased so that the score on one spin has mean 3.4 and variance 1.9, find the power of the test.
  4. Describe what would happen to the power of this test if Callum spun the spinner more times, using the same critical region. Give a reason for your answer.
10 marks
Show answer
a) Mean (5 + 1)/2 = 3, variance (25 - 1)/12 = 2 b) Size = 2 × P(Z < -1.6432) = 0.100 c) 0.7130 d) Power increases: Var(X̄) = σ²/n falls, so X̄ is more concentrated about 3.4 and more likely to exceed 3.3
12

A machine makes ceramic tiles. The manager claims that the probability, p, that a tile is cracked is 0.1. A buyer believes that p is greater than 0.1 and wants to test H0: p = 0.1 against H1: p > 0.1 at the 5% level of significance.

  1. The buyer first considers inspecting tiles until she finds one cracked tile, rejecting H0 if this happens soon enough. Show that no critical region exists for this test at the 5% level.

Instead, she inspects tiles until she has found 2 cracked tiles. Y is the number of tiles inspected.

  1. Find the critical region and the size of this test.
  2. Show that the power function of the test is p2(3 − 2p).
  3. Sketch the power function for 0 ≤ p ≤ 1 on the axes.
  4. Show that the power of the test is 0.5 when p = 0.5
00.20.40.60.8100.20.40.60.81pPower
10 marks
Show answer
a) P(X ≤ 1) = 0.1 > 0.05, so even X = 1 is not significant: no critical region b) Y ≤ 3, size 0.028 c) P(Y ≤ 3) = p² + 2p²(1 - p) = p²(3 - 2p) d) Increasing curve from (0, 0) to (1, 1), through (0.5, 0.5) e) 0.5²(3 - 2 × 0.5) = 0.5

Your score

Answer the questions above and your score appears here.

Every question here is original and written by Robinson Tuition. It follows the Edexcel A-level Further Mathematics specification and suits AQA and OCR Further Maths too; this is not an exam-board paper.

All A-level Further Statistics 1 topics