Hello all,
I'm running a multinomial logit model followed by a linear regression. The outcome variable for the multinomial is innovation_type (four values: product, service, both, neither). I computed the four inverse mills ratios corresponding to the four innovation types as per this post:
https://www.stata.com/statalist/arch.../msg00465.html
In the 2nd stage linear regression, I'm trying to predict the financial outcome (sales) using the innovation type and other control variables. Which three out of the four inverse mills ratios from the multinomial logit should I include in the 2nd stage regression? I'm using outcome 2 (service) as the base for the innovation type in the regression equation. So should I include the IMRs corresponding to the other three innovation types?
Would it be a better model if I use all four inverse mills ratios?
Thank you very much for your help.
I'm running a multinomial logit model followed by a linear regression. The outcome variable for the multinomial is innovation_type (four values: product, service, both, neither). I computed the four inverse mills ratios corresponding to the four innovation types as per this post:
https://www.stata.com/statalist/arch.../msg00465.html
In the 2nd stage linear regression, I'm trying to predict the financial outcome (sales) using the innovation type and other control variables. Which three out of the four inverse mills ratios from the multinomial logit should I include in the 2nd stage regression? I'm using outcome 2 (service) as the base for the innovation type in the regression equation. So should I include the IMRs corresponding to the other three innovation types?
Would it be a better model if I use all four inverse mills ratios?
Thank you very much for your help.
