Reliability of Difference Between Uncorrelated Means (Nᡖ)
The procedure we have just seen can be used only when we have 30 or more subjects in our sample.
We assume that subjects are assigned to each group randomly.
- Nothing but chance determines which group a subject is in.