interpreting, reporting t-test and multiple regression using log-transformed variable

#1
Dear all

I wonder if anyone can help with this question. I have to log-transformed (log 10) a number of my variables (skewed distributions). My question is, how do I report the statistical test? e.g. if I perform a t-test, do I need to back-transform my means and SD and the t-statistics? If so, how do I do that? and how do I interpret these back-transformed values?

Similarly, if I use a log-transformed variable as my dependent or independent variable in a regression analysis, how should I interpret/report my result?

many thanks for your help!
Jim
 

noetsi

No cake for spunky
#2
Re: interpreting, reporting t-test and multiple regression using log-transformed vari

There are two alternatives, both are used. One is simply to refer to the results of your logged variables (that is note your results in terms of the logged variables - this makes interpetation difficult at times). The second is to reverse the logging process (I know this can be done, I don't know how in practice).