Dear research community,

I am about to run a panel regression for 30 countries and a period of 14 years. My dependent variable is OECD Triadic Patent Families which measures the number of patents taken at the EPO, JPO, USPTO. It becomes manifest in fractional count data as a patent will be allocated to multiple countries if there are multiple inventors with different nationalities. As such I treat it as a continuous variable. before testing for appropriate regression models (OLS, fixed-, random effects)/ running different regression models do I have to "inform" stata that the DV has a lower bound at zero? If yes how should I proceed here?

More Frederic Virnich's questions See All
Similar questions and discussions