Multiple imputation, FCS statement : discrim and logistic difference

[Andrea1]

Hi,

I am working with exclusively binary and categorical variables where almost all of them have missing data.

So, I am using the PROC MI with the FCS statement to impute them.

I have several nominal variables to impute but I can't figure out the difference between using the statement logistic(,link=glogit) or discrim(, classeffects=include) knowing that none of my effects are continuous. The results aren't the same for these methods but I can't understand what is the difference and which one should I use ?

Thanks for your help ! Let me know if you need more details

Andréa

[Andrea1]

Someone who would have an idea please ?

[homipilot]

Hi,

You may find the below link useful in understanding the difference.

https://stats.oarc.ucla.edu/sas/seminars/multiple-imputation-in-sas/mi_new_1/Multiple Imputation in ...

Good luck!