0
$\begingroup$

I am dealing with a dataset that contains demographic covariates (constant across time) and dummy variables for subjects for six repeated measures in wide format. I.e. a row represents a single measurement for a single subject.

The pattern of missed measurements by subject is random and no more than 5% of all observations are missing. Here, in case a subject misses a single measurement, all but the demographic covariates will be missing.

In my case the number of dummy variables far outnumbers the number of demographic covariates (~500:5) and there is high variability in the five demographic variables.

Imputation of the missing values in the first iteration will be performed according to five covariates that have high variability across the sample. Is the R package mice, Multivariate Imputation by Chained Equations, suitable to fill in such missing values?

I do not want to use Values With Last Observation Carried Forward NA.LOCF, as this will likely lead to bias in my sample.

Improve this question
$\endgroup$
2
  • $\begingroup$ Welcome to the site. I am voting to leave this open, but some are voting to close it because questions about code are off topic. If you remove the reference to R and just discuss the method, those people may cease to object (no guarantees). $\endgroup$
    – Peter Flom
    Commented Jan 16, 2019 at 10:54
  • $\begingroup$ I am attempting to edit the question, but run into a "404 Page not found" message. This might be an issue specific to my browser or user privileges. Before I solve this issue, can I request another user edits the question? $\endgroup$
    – user10712739
    Commented Jan 16, 2019 at 18:03

1 Answer 1

Reset to default
0
$\begingroup$

I solved this issue by changing the orientation of the data to a true wide format--where each repeated measure is represented in a separate column. Another solution is to set all intersects of dummy variables to 0 in the predictor matrix.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.