Comparing populations
What this lesson covers
- The interval is for the difference between groups.
- An interval that excludes zero supports a difference.
- Experiments use randomisation tests.
Reading the difference
- Mean of A minus mean of B.
| Interval | Contains 0? | Conclusion |
|---|---|---|
| 1.2 to 5.8 | No | A is likely higher |
| −1.1 to 3.4 | Yes | No clear difference |
Observational or experimental
- Observational data can suggest a difference between groups.
- A randomised experiment can support a claim about cause.
- A randomisation test asks how often chance alone gives so large a difference.
Common mistakes
- Saying zero in the interval proves no difference.It means the data cannot show one.
- Claiming cause from a survey.Only random allocation supports cause.
Your turn
- 1. The difference interval is 0.4 to 2.6. What can you say?
- 2. The tail proportion of a randomisation test is 0.3. Is chance a plausible explanation?
Show answer
1. Zero is excluded, so the first group is likely higher 2. Yes, as a difference this large is common by chance
What to take away
- Look for zero.
- Design decides what you can claim.
More in AS 91582
Is your child on track?
Take the free placement test