Dear all,
I am working with my resident survey data (n=8356) including 59 items, most of which are ordinal variables (measured by Liket-7 points scale), and others are continuous variables such as age and residency length.
However, all items have some missing values, ranging from 40 to around 200, which follow an arbitrarily missing pattern. This leaves me with 6001 complete observations but more than 2000 incomplete records. And around 1000 records contain only one missing value. So I want to impute them.
I realize that I could use single or multiple imputation method provided by Stata, while it seems that both require variables with complete data to be the "regular variables", but this is not the case in my study. I am not sure how to conduct imputation when all variables have missing values.
Two additional problems further bother me if I choose multiple imputation method in Stata.
1) before running my analysis model, I need to generate fewer composite variables based on these individual items (e.g., taking means of items belonging to the same construct). So how could I get the combined estimates if the imputed data were not prepared for analysis model immediately?
2) My analysis model is actually multilevel ologit model, which seems not supported by -mi- command in Stata at the moment.
Any help would be much appreciated.
I am working with my resident survey data (n=8356) including 59 items, most of which are ordinal variables (measured by Liket-7 points scale), and others are continuous variables such as age and residency length.
However, all items have some missing values, ranging from 40 to around 200, which follow an arbitrarily missing pattern. This leaves me with 6001 complete observations but more than 2000 incomplete records. And around 1000 records contain only one missing value. So I want to impute them.
I realize that I could use single or multiple imputation method provided by Stata, while it seems that both require variables with complete data to be the "regular variables", but this is not the case in my study. I am not sure how to conduct imputation when all variables have missing values.
Two additional problems further bother me if I choose multiple imputation method in Stata.
1) before running my analysis model, I need to generate fewer composite variables based on these individual items (e.g., taking means of items belonging to the same construct). So how could I get the combined estimates if the imputed data were not prepared for analysis model immediately?
2) My analysis model is actually multilevel ologit model, which seems not supported by -mi- command in Stata at the moment.
Any help would be much appreciated.
