1
$\begingroup$

I have an experiment wherein respondents were tested in two time points. However, respondents were tested at t1 and t2 OR t1 and t3 OR t1 and t4. Hence, data is missing at t2,t3,and t4 for 3/4 of respondents. I meet the missing at random assumption, but is there anything else I need to consider before I start up the imputation machine?

$\endgroup$
1
$\begingroup$

Instead of imputing something that seems almost like a planned missingness design (which would fit the MCAR assumption), I would take a look growth curves. These actually adequately handle data that are either missing completely at random or missing at random using maximum likelihood estimation.

You can implement these in R using either a mixed-effects package (lme4 or nlme) or an SEM (OpenMx) package. The links I used discuss growth curves in more detail with those particular packages.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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