Dear colleagues
I'm thinking of a method to isolate the effect of a particular variable (X) on say Y and then use that it in an entirely different regression and another dependent variable.
For example, assume I have the following regressions:
yi=β0+β1xi+β2zi+ϵi (1)
and let's say that the negative effect of x on y is established (just for the sake of argument), what is the best way to isolate the reduction of Y that is attributed to x and use it in another regression (2)?
ji=β0+β1(reduction of Y attributed to x in regression 1)i+β2pi+ϵi (2)
I thought about running the regression with and without x and then compare the residuals but not entirely convinced.Or should I save β1 and use it?
Any Ideas?
Thank you
0 Response to predicting effect
Post a Comment