Carrying Out a Test for the Difference Between Two Population Means
Your hypotheses are ready. Now measure the evidence: calculate a two-sample statistic, find the correct -value and explain what the results say about the population means.
By the end of this lesson, you should be able to:
- Calculate the standard error and unpooled two-sample statistic.
- Use the appropriate degrees of freedom and alternative to find a -value.
- Interpret that -value assuming equal population means.
- Compare with and justify a conclusion in context.
Before you start: Review Topic 4.9: setting up a two-sample -test. You need independent groups with a quantitative response and justified conditions.
First time learning this? Follow the delivery-time calculation, then compare a lower-tail test, a two-sided test and a randomized experiment.
Here to revise? Use the complete test checklist, then attempt the practice questions before opening the solutions.
The concept in 60 seconds
The null says the population means are equal. The test asks how unusual your observed sample mean difference would be under that assumption.
Difference → standardized evidence → probability → conclusion. Divide the observed difference by its standard error to get . Use the appropriate distribution to find the -value. Compare that -value with the planned significance level .
A small -value makes the observed result hard to explain by sampling variation under equal population means. A large -value means the evidence is not strong enough to reject equality at the chosen ; it does not prove the means are equal.
The direction of the alternative comes from the original question. You cannot switch to the favorable tail after seeing the data.
Finish the delivery-time test
Continue the study from Topic 4.9. Independently selected SRSs compare service A’s deliveries with service B’s deliveries during a defined period. The question is whether A has a longer population mean delivery time. Use .
| Service | Sample mean | Sample | |
|---|---|---|---|
| A | |||
| B |
Define and as the population mean delivery times for these services during that period. Test against .
The independent SRSs support random sampling and group independence. The checks are and . Both sample sizes are at least . We can use an unpooled two-sample -test.
Technology gives . The greater-than alternative uses the right-tail probability: .