search

ijyliu / ECMA-31330-Project

Econometrics and Machine Learning Group Project
2 stars 1 forks source link

Add non-standardized regular OLS coefficient to Regression Tables #14

Closed ijyliu closed 3 years ago

ijyliu commented 3 years ago

This will be interpretable, whereas our standardized/rescaled coefficients generally won't be. It could give a sense of the effects size/scale we are looking at.

Also, would it be smart to convert the coefficients obtained after standardization back into the original units by multiplying by the standard deviation? Does this make sense?

ijyliu commented 3 years ago

Now, may just not standardize X and Y in the empirical application.

ijyliu commented 3 years ago

Eh, not sure adding this is needed if we don't make causal claims and are only focusing on the relative magnitude of the coefficients.

ijyliu commented 3 years ago

Also I guess we don't really want people looking at the OLS coefficients since they aren't the most correct/unbiased ones, so it wouldn't make sense to show this interpretation for those only.