I have a bunch of independent variables which are skewed and have negative and zero values. I am seeing a lot of suggestions of using cube root as a transformation. I want to ask what is the harm in using this instead: Sign(x)*log(1+Abs(x))