Hypothesis tests for binomial and Poisson
Studying this course? Keep your progress in one place.
Key facts
- Binomial test: under , . Poisson test: under , .
- Scale a Poisson mean to the interval observed.
- Use the critical region method or the probability method; both give the same decision.
- In a two-tailed test compare each tail with half the significance level.
- Conclude in context: "sufficient" or "insufficient evidence".
Setting up the test
The number of successes (binomial) or events (Poisson), with its distribution under .
To test a binomial proportion , the test statistic is the number of successes in trials. Under , .
To test a Poisson mean , is the number of events in a fixed interval. Under , .
Scale the mean to the interval observed: per hour becomes for hours.
- 1
State the hypotheses
and about or
- 2
Give the distribution
Under
- 3
Find the region or probability
Critical region or p-value
- 4
Compare
With the significance level
- 5
Conclude
In context
A shop receives calls at a mean rate of per hour. A test is based on a -hour period. What is the Poisson mean under ?
Binomial tests
Find the critical region, or the probability of a result at least as extreme, and compare with the significance level.
The critical region method finds the largest region in the relevant tail with probability at most the significance level. The probability method finds the probability of a result at least as extreme as observed and compares it with the level.
Tables usually stop at . For , let . For , .
Worked example
and at the level. Find the critical region. If seeds germinate, what is the conclusion?
- 1
is lower-tailed, so find the largest with .
- 2
, but .
- 3
So the critical region is . The observed value is in it, so reject .
For and , which probability do you calculate for an observed value ?
Poisson tests
Define for the interval observed, then find the tail probability.
The hypotheses must define for the interval observed, and the probability uses the tail that points to.
Calls arrive at a mean of per hour. In hours there are calls, and we test at whether the Poisson mean has increased.
Worked example
Calls arrive at a mean of per hour. In hours there are calls. Test at the level whether the rate has increased.
- 1
, , where is the mean number of calls in hours.
- 2
Under , .
- 3
.
- 4
, so reject .
If and , what is ?
Two-tailed tests
The significance level is shared between the two tails.
For or , the significance level is shared, so at each tail has .
Find the tail probability on the same side as the observed value and compare it with , not . Alternatively, double that probability and compare with .
- Two-tailed, 5% leveleach tail 0.025
- Two-tailed, 10% leveleach tail 0.05
In a two-tailed test at the level, what should a one-tail probability be compared with?
Large samples and conclusions
For large use a normal approximation with a continuity correction, then conclude in context.
For large , such as , use under with a continuity correction, checking and exceed . For a critical region, solve an inequality like and take the largest integer .
Always state whether is rejected, then say "sufficient" or "insufficient evidence" in the language of the question. Never say the claim is proved.
Worked example
under . The observed value is and at the level. What is the conclusion?
- 1
Mean and variance , so .
- 2
, with , so .
- 3
Two-tailed, so compare with . Since , do not reject .
Why is a continuity correction needed when approximating by a normal distribution?
Try an exam question
A call centre receives calls at a mean rate of per hour. After a change of staffing, calls are received in one hour. Test, at the level, whether the mean rate has decreased.
[5 marks]
- [1], .
- [1]Under , .
- [1].
- [1], so do not reject .
- [1]There is insufficient evidence that the mean rate of calls has fallen.
That's the notes covered.
Carry on to the next subtopic.
Try a question
Assuming is true, what is ?
1 mark, marked against the mark scheme
Assuming is true, what is ?
1 mark, marked against the mark scheme