Determining Confidence for Different Population Samples - GT

[glok_twen]

hi. i have two samples with known mean and variance. do you know which test i use to determine the confidence that they are drawn from different populations?

also if you know a link that explains how to calculate the test that would be helpful. i have some stats books too in case needed. (but couldn't seem to pick out which the right test was)

thanks,
gt

 

[Dr Transport]

If memory serves me correctly, use the F-test.

 

[glok_twen]

thanks.

best i can tell is that the f-test is used to determine whether two sample variances are indeed drawn from the same population:

http://www.itl.nist.gov/div898/handbook/eda/section3/eda359.htm

and that to see whether the means are drawn from the same population the p-stat is used.

please advise if you think this sounds correct.

 

[ssd]

hi. i have two samples with known mean and variance.

? You imply population mean & variance known? (If you have the samples you can always calculate means and variances... )
Any way, if all the four parameters are unknown, you can test equality of the CVs (coefficient of variations). Testing merely the quality of means or merely the equality of sd's will not serve your purpose.

 

[glok_twen]

what's known

in this case i know only the means and std dev's for 2 samples, but not for the population. i would like to know with a given degree of confidence whether those 2 samples were drawn for the same population.

i've seen conflicting ways to do this - one with a p-score, one with a f-test, and i am still confused with which is correct.

 

[ssd]

in this case i know only the means and std dev's for 2 samples, but not for the population. i would like to know with a given degree of confidence whether those 2 samples were drawn for the same population.

i've seen conflicting ways to do this - one with a p-score, one with a f-test, and i am still confused with which is correct.

You cannot do this unless the sample sizes are known.

 

[glok_twen]

yes, thanks, i inadvertently omitted that - i know the mean, std dev and sample size for two samples. and then the problem stands as originally stated - which test is the correct one to determine a confidence level on whether they were drawn from the same population.