Comparing means of unequal distributions

Question

Hi,

I have measurements from two groups with a large variance (for which I have already corrected by a Log10 transformation) and unequal distributions (Bartlett p=0.009).

What is the proper way to compare the means of these two groups?

The data can also be paired.  Is it correct to use the paired t-test on the Log10 data?

Thanks

Shahar · Answer

Thanks a lot, Jim

txnelson · Answer

You will want to used the log transformed data since that transformation results in a normal distribution.  And you would want to use a paired t-test if the data warrants it, to increase the power of the test.

Shahar · Answer

Thanks a lot.
And the groups in general?
Also a t test on Log-transformed data?