I recently recevied a priliminary dataset and I am planning to run several regressions; however, due to the nature of the data I am not quite sure which estimation method to apply.
I want to regress several independet variables (firm specific characteristics, such as sales, total assets etc.) on
a) Concentration measures (Percentage values)
b) Categorial Variable (1, 2, 3, or 4)
Please correct me if I am wrong, but afaik n both cases an OLS regression would not be adequate.
I learned that I could convert the concenctration measures (a) from an bounded into an unbounded var by the expression log ( x / 100-x). However, I am not sure which estimation method to use after this transformation?
What about the categorial variable? In which case to use Log, in which case probit?
I am grateful for any hint!
Advertise on Talk Stats