Dear all,

If i am carrying out multiple imputation as a result of missing data how do i go about it if i have 2 outcomes to consider in my analysis. Do i have to carry out multiple imputation for the outcomes separately? For example, i have 2 outcomes (depression (n=5000) and educational attainment (n=3500)) that have different number of observation as a result of missigness. Some of the covariates used (we use similar covariates for both outcomes) also have missing data which i want to impute. Will it be right to choose the outcome with the highest number of variables and base my imputation on it.

E.g. drop all observation which is missing "depression" and then carry out multiple imputation for educational attainment (another outcome) and other covariates so all of them come to n=5000.