Dear all,

I have a question regarding whether I should log-transform my dependent variable or not. I know that there are a lot of threads regarding this topic but I haven´t found a good answer for my case yet. I also posted a similar question on CrossValidated but didn´t get a response there. Thank you very much for considering my case!

I know that if my dependent variable is in log I can interpret the coefficient as a percentage change. However, in my case my dependent variables are ratios such as workers per firm or output per firm. What would the interpretation be if I log these variables (i.e. log(workers/firm))? My independent variable is a dummy for program participation and I am using a DID model. So in effect I am wondering whether dummy = 1 would still be interpreted as being correlated with a change in my dependent variable by approximately 100*beta percent?

Again, thank you very much for your help!

All the best
Leon