My question is in which cases you should think about transforming your variable into a logarithmic one? My dependent variable in the number of patents a long a period of years and one of my explanatory is "R&D BUDGET" (research and development) . Whan im doing a regression i get a low values of R squared and Adj. R squared , and the intuition is that more budget will increase the number of patents . Is this kind of case i should consider transforming my explanatory variable into a logarithmic? Thank you in advance, Nitzan


marked as duplicate by Andy W, Nick Cox, gung - Reinstate Monica, Scortchi - Reinstate Monica, Xi'an Mar 17 '15 at 12:52

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

  • 2
    $\begingroup$ Number of patents is a count variable, so indicates mabe poisson regression, so you could use a GLM with log link. Also, if using linear regression, contemplate a square root transformation, which in Poisson case stabilizes variance. $\endgroup$ – kjetil b halvorsen Mar 17 '15 at 11:58
  • 1
    $\begingroup$ Although the title of the duplicate question focuses on independent variables, the answers are much more general and applicable to this question. $\endgroup$ – Andy W Mar 17 '15 at 11:59
  • 1
    $\begingroup$ Also refer to: stats.stackexchange.com/questions/130262/… $\endgroup$ – Adam Robinsson Mar 17 '15 at 12:12
  • $\begingroup$ To Kjetil i sew that you suggested several of option, it would be great to hear from you knowledge and econometric experience which model do you think that in this research is probably the most appropriate for me to use ? $\endgroup$ – Nitzan Mar 18 '15 at 15:54