Mean Differences and p
In class today, John said he was a bit confused on the connection (in t-tests comparing two group means) between the difference between the sample means and the size of the computed p-value. Perhaps I can say a few things that will help clarify this "connection."
First of all, in order to examine the connection between the mean difference and p, we've got to assume that everything else is held constant. Thus, what I will say next presumes that we do NOT change the amount of data, the number of "tails" in the test, the amount of variability within each sample does not change, or the level of significance.
Now, if the means are close together, the calculated value will be small and the p will be large. In the other hand, if the means are far apart, the calculated value will be large and the p will be small.
Copyright © 2012
Schuyler W. Huck