Language of hypothesis testing

Language of hypothesis testing

Spec O1 · AQA A-Level Maths: Revision notes

Studying this course? Keep your progress in one place.

Key facts

  • H0H_0 is the claim assumed true (one value, H0:p=p0H_0:p=p_0); H1H_1 is what you suspect, and its direction comes from the question.
  • A 1-tail test has H1:p<p0H_1:p<p_0 or p>p0p>p_0; a 2-tail test has H1:p≠p0H_1:p\ne p_0.
  • Work under H0H_0: use X∼B(n,p0)X\sim B(n,p_0).
  • Reject H0H_0 if the pp-value is ≤\le the significance level, or if XX is in the critical region.
  • Conclude in context, cautiously: never say H0H_0 has been proved.

Hypotheses

A hypothesis test uses a sample to decide whether a claim about a parameter is believable. H0H_0 is the claim assumed true; H1H_1 is what you suspect instead.

For a binomial model X∼B(n,p)X\sim B(n,p) the parameter is pp. H0H_0 gives one value, such as H0:p=0.6H_0:p=0.6.

Always define pp in words, for example 'p is the probability that a seed germinates'.

The direction of H1H_1 comes from the wording of the question, never from the sample result.

00.10.20.30.40.50.60.70.80.91H0: p = 0.6

1-tail test

  • H1:p<p0H_1:p<p_0 or H1:p>p0H_1:p>p_0
  • The suspicion has a direction

2-tail test

  • H1:p≠p0H_1:p\ne p_0
  • The suspicion is only that the value has changed

A company claims 60% of customers are satisfied. A manager thinks the figure is lower. Which is the correct alternative hypothesis?

The test statistic

The test statistic is the quantity calculated from the sample, here the number of successes XX. You work under H0H_0 to see how likely the result would be.

The test asks: if H0H_0 is true, how likely is a result like the one observed?

So use the distribution under H0H_0, for example X∼B(20,0.6)X\sim B(20,0.6) if n=20n=20 and H0:p=0.6H_0:p=0.6. Only the value of pp from H0H_0 is used in the calculation.

  1. 1

    State hypotheses

    H₀ and H₁, with p defined

  2. 2

    Assume H₀

    Use its distribution for X

  3. 3

    Compare

    Observed X against the critical region or p-value

  4. 4

    Conclude

    In context

How a hypothesis test runs

n=20n=20 and H0:p=0.6H_0:p=0.6. Which distribution is used for the test statistic?

Critical region

The significance level α\alpha is how unlikely a result must be before you reject H0H_0. The critical region is the set of results that lead to rejection.

The significance level α\alpha is the probability of rejecting H0H_0 when it is true, set in advance (commonly 5% or 1%).

The critical region is the set of values of the test statistic that lead to rejecting H0H_0; its boundary is the critical value. The rest is the acceptance region.

For a 1-tail test all of α\alpha is in one tail. For a 2-tail test at 5%, put at most 2.5%2.5\% in each tail.

00.020.040.060.080.10.120.00030.00160.0065456789kP(X ≤ k)
For X ~ B(20, 0.6): P(X ≤ k) first exceeds 0.05 at k = 8, so the 5% critical region is X ≤ 7

Worked example

X∼B(20,0.6)X\sim B(20,0.6) and H1:p<0.6H_1:p<0.6 at the 5% level. Find the critical region.

In a 2-tail test at the 5% level, what is the maximum probability allowed in each tail?

The p-value

The pp-value is the probability, assuming H0H_0 is true, of a result at least as extreme as the one observed. Reject H0H_0 if it is no more than α\alpha.

For X∼B(15,0.3)X\sim B(15,0.3) and an observed value of 9 with H1:p>0.3H_1:p>0.3, the pp-value is P(X≥9)=0.0152P(X\ge9)=0.0152.

Decision rule: if the pp-value is less than or equal to the significance level, reject H0H_0; otherwise do not reject.

In a 2-tail test, compare the one-tail probability with α/2\alpha/2, or double it and compare with α\alpha.

00.010.020.030.040.050.060.070.080.090.1p-value 0.0152α = 0.05

With α=0.05\alpha=0.05 and a pp-value of 0.01520.0152 in a 1-tail test, what is the decision?

Actual significance and conclusions

With a discrete distribution the critical region rarely uses up exactly α\alpha. Conclusions must be in context and cautious.

The actual significance level is the real probability of falling in the critical region when H0H_0 is true. For X∼B(30,0.35)X\sim B(30,0.35) with critical region X≤5X\le5 or X≥17X\ge17: 0.0233+0.0124=0.03570.0233+0.0124=0.0357, below 5%.

Never say that H0H_0 has been proved. The significance level is also the probability of wrongly rejecting a true H0H_0.

00.010.020.03P(X ≤ 5)P(X ≥ 17)Actual significancePart of the critical regionProbability
B(30, 0.35) with critical region X ≤ 5 or X ≥ 17: the actual significance level is 0.0233 + 0.0124 = 0.0357, below 5%.

Result in the critical region

  • Reject H₀
  • There is sufficient evidence at the 5% level to suggest that the proportion has changed

Result not in the critical region

  • Do not reject H₀
  • There is insufficient evidence to suggest that it has changed

The test statistic is not in the critical region. Which conclusion is best?

Try an exam question

A spinner is claimed to land on red with probability 0.250.25. Sam thinks the probability is higher. In 2020 spins it lands on red 99 times. Test Sam's suspicion at the 5% significance level.

[5 marks]

That's the notes covered.

Carry on to the next subtopic.

Try a question

Which pair of hypotheses should she use?

1 mark, marked against the mark scheme

Revision notes