Missing data

[Paige LaPierre]

Hello!

I am currently figuring out how to deal with with missing data. According to the user guide, they suggest 7 ways to deal with missing data. This is a quote from it:

Impute partial non-response (replace missing values with replacement values). The advantage of imputing partial non-response over reweighting is that all records are kept,which means no data are discarded. Only the missing values in each record are filled in. It is important to quote the imputation rate with your analytical results and give information about the imputation strategy used.

Previous studies using the same dataset that I am have used 0 to replace missing data for dichotomous variables, and mean scores for continuous variables. One uses the Markov chain Monte Carlo (MCMC) method.

I am reaching out for some insight on which one you would recommend. I don't fully comprehend the MCMC, nor how to use it. Moreover, if you could help me out with the code to impute, I would really appreciate it.

Some examples of variables with values that need to be replaced are the following:

• Continuous variables: ppvtr
• Dichotomous: gender, childcare, lowincome, parentschool

I would then use these variables to do a svy regression using bootstrap weights where the code is :

svy: reg ppvtr childcare gender parentschool

I would attatch dataex, but I unfortunately only have access to my data in a secured facility and therefore cannot attatch the output.

Thank you in advance for the help!

[Roman Mostazir]

You should invest some time in Stata's mi suite for multiple imputation manual.