i would enjoy to have a little more information, if you can share:
0 = say that population a is the one with simple random sample & pop b is the stratified one
1 = what are the two population sizes, Na & Nb ?
2 = what are the two sample sizes, na & nb ?
3 = what was the stratified strategy and what was the simple random strategy adopted?
4 = are the variables upon which the stratification occured also present in pop b?
on of my research focuses is nonparametric, kolmogoroff-smirnoff enhancements. i am thinking that this sort of approach might enable no assumptions regarding parameters, especially variances. before going further, though, the answers to the above two questions seem quite important, and there seem to be many creative and valid strategies that could be embraced.
regression seems interesting, too, and might benefit from weighting or other modifications.
www.twobluecats.comwww.datastatisticsonline.comwww.twobluecats.be
Original Message------
I agree with this answer, but note that it assumes the samples are large enough for the two variances to be well estimated. If not, step 3. should be replaced - perhaps by a permutation test of some sort.
Robert
Robert C. Elston,
3000 Galloway Ridge, Apt G301 Phone: (001) 919-542-0841
Pittsboro NC 27312