Power function, uneven sample number

Question

Hi

I have a data set made up of few samples Group #1 (8 samples) and Group #2 (2 samples).  I am performing a t test to identify the mean differences.  As the sample size is small I would like to understand (1) what the power is currently (2) what number of samples for Group #2 would I need to be 95% confident of the result.  I attached the data set.

Jack

Power example.jmp

Jack_12 · Answer

Hi Mark
When I plugged in 2 as a number it came up with an error, which I assumed was because n=2 is not viable its too small, this was good as I can communicate the test does not work with n=2, it seems that n=4 is minimal number to use in the test

Regards Jack

Mark_Bailey · Answer

Yes. So use n > 2. Aren't you using this retrospective power analysis to prospectively determine the minimum sample size? That number is likely greater than two.

Jack_12 · Answer

Thanks Mark, am I right in thinking that n=2 is literally too small to determine the power function. I plugged in the data and it stated "Some Probability Calculations Signaled Errors", it works with n=3