I would like to test the assumptions for a regression analysis. Since I have missing values, I multiply imputed them with the mice function in R.
I thought it might make more sense to test the assumptions (linearity, normality of residual, homoscedasticity, uncorrelatedness..) after I did multiple imputation. What is the common sense?
Check assumptions before or after imputing?
If after, how can I concretely proceed in R?
Thanks a lot!
Hello Xenis,
The amount of imputed data points would obviously be a factor in determining whether assumption checks could disagree if conducted on complete-case data only vs. complete case + imputed case records. My suggestion would be to report on the full (including imputed values) data set, since it's from that base that your model is being estimated.
Good luck with your work.
Les chiffres sont d'expérience, les hypothèses sont de contours et contexte...mananagement analyse des chiffres d'un Mining ,la traçabilité des composantes principales m3, T, H2O, %Cu,%Co...de la Mine jusqu'à l'usine...Connaître le Processing ,le but, le budget, l'objectif...apte à faire des simulations du temps ...jour,nuit...le rendement, la maintenance, l'électricité...votre problèmatique des chiffres est plus d'informations par flaire d'expérience pour les diverses réorientations,Merci.