Hello all,
The mean of each sample was computed as an estimate of the population
mean uk (sum of xk,m / n = Xk = uk). However, the mean generated from
the different samples have different levels of N (observations). I am
trying to test whether these mean are statistically significant. I
have n, the average mean absolute error (Avg. MAE), the standard
deviation, and the standard error for each measurement. I know how to
test the differences when the sample is balanced but I am not quite
sure how to do it when the sample is not balanced. For example, I
want to compare model 1 with model 4 but model one has an N of 177,
while model four has an N of 49.
The null hypotheses tested are:
H0: uk <= 0
The alternative in each case was uk > 0.
This is what I am trying to end up with table-wise:
Model k Model 4
Avg.MAE StdDev Avg.MAE StdDev MeanDiff StdEr TestStat(pval)
Model 1
Model 2
Model 3
Any thoughts would be greatly appreciated.
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
