Quote:
|
Originally Posted by gal
I've tried to read up on this in my old statistics book, but most of my skills in this field are long gone..
I have 4 samples of batch Y and 2 samples of batch X. How can I test whether E(X) > E(Y)? I suspect I don't have enough observations to test this. I looked up some tables on signed-rank tests and the smallest observations they had were 4 to 5. The outcome should follow the normal distribution though so maybe I can use some other test..
Y: 10269, 16806, 14023, 8307
X: 15332, 20149
|
The thing that drives me nuts about these theoretical problems is that the problem is solved with more data. Right now we are talking about confidence levels of a weak correlation that Set X belonging in Set Y. How do we know that set Y doesn't represent a measure of cloud formations and Set X represent some measure of the mating patterns of sea horses? i.e. what exactly are we testing?
I have a very good friend who has a PhD in stats. I will see if he will entertain this.