Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • how do you register interaction variables in mi impute chained?

    My mi impute chained model (which contains both dummy and continuous variables imputed with (regress) and (logit) respectively is failing to converge due to perfect prediction. I have deduced that the problem variables are all the categorical variables that are part of interaction terms. These are a collection of continuous x continuous, continuous x binary and binary x binary variables. I know I should impute these but I am not sure if I should list them after (regress) or (logit)? I tried to both list them after (regress) and (logit) but no success.
    Where do I include these three type of interaction variables in the imputation command?
    Augment both within the parentheses and at the end is not helping either.
    Here is my code with all the variables registered if it is any help (I am using stata 12):

    mi impute chained (regress) age numch numsib ocpr0 jobyear lsal educyear conf nervous spocpr specsec spjobyear lspsal parocpr paeduc numfaminus famtiegravsponsor famtiegrav lwealthwosqm ifoplive age_chimin male_chimin age_male geotravex1_educyear geotravex2_educyear married_chimin married_age parret_numsib parret_age chimin_parret (logit) othersapply disadveth migrant otherpaidtrip pargeotravex1 pargeotravex2 married geotravex1 geotravex2 chimin parret humanities socialscience professions stem business, add(1) rseed(12345) force



Working...
X