I would like to estimate a translog production frontier function. For that I want to normalize my variables relative to the mean ( i suppose that means subtracting the mean from the variable itself? i understand there are other ways to normalize as well like z=X-Xbar/sd or X-Xmin/Xmax-Xmin as well). When i do so, it creates very small and negative values in both dependent and independent variables. Also after normalization when i log transform, there are even more zero and negative values. My question is, can i still run a multiple regression with negative and zero values (particularly in dependent variable), and does the interpretation of the coefficients change? Can someone please assist. Thanks

More Hayatullah Ahmadzai's questions See All
Similar questions and discussions