Hello statalist,
I have a dataset that contains parasite data that mimics cost-effectivess data. There is a binary variable for infection (0 or 1) and a continuous variable for intensity (for those who are infected).
I have seen a few articles in the literature that discuss a two-step MI (see attached).
In these cases, one would set intensity data to missing for those who are uninfected. Then, values for the binary and continuous variable would be imputed.
Lastly, the two variables would be multipled to give an imputed intensity of 0 when the infection =0 and the imputed continuous intensity when infection=1.
Would anyone know the coding for this? The second MI IMPUTE code would rely on the values of the first imputation. However, I am unsure how to restrict to a variable that was imputed in the second model?
sample code (asc_bin= infection; asc_num=intensity; total n=1010; 48 parasite outcomes missing)
STEP 1: mi impute logit asc_bin group age education school, ad(20) noisily
STEP 2: mi impute pmm asc_num group age education school if asc_bin ==1, ad(20) knn(10) noisily
Thanks,
Layla Sarah
Administrative edit:
Links to articles:
Faria 2014
Vroomen 2015
Burton 2007
I have a dataset that contains parasite data that mimics cost-effectivess data. There is a binary variable for infection (0 or 1) and a continuous variable for intensity (for those who are infected).
I have seen a few articles in the literature that discuss a two-step MI (see attached).
In these cases, one would set intensity data to missing for those who are uninfected. Then, values for the binary and continuous variable would be imputed.
Lastly, the two variables would be multipled to give an imputed intensity of 0 when the infection =0 and the imputed continuous intensity when infection=1.
Would anyone know the coding for this? The second MI IMPUTE code would rely on the values of the first imputation. However, I am unsure how to restrict to a variable that was imputed in the second model?
sample code (asc_bin= infection; asc_num=intensity; total n=1010; 48 parasite outcomes missing)
STEP 1: mi impute logit asc_bin group age education school, ad(20) noisily
STEP 2: mi impute pmm asc_num group age education school if asc_bin ==1, ad(20) knn(10) noisily
Thanks,
Layla Sarah
Administrative edit:
Links to articles:
Faria 2014
Vroomen 2015
Burton 2007