I am trying to utilize the 'ice' command for multiple imputation.
My dataset is longitudinal with several time-varying covariates (eg, systolic bp, BMI) collected over 12 years. The data are formatted in long format for use with stset.
I was informed that I needed to transform my data into wide format, with one row per subject (due to independence of observations in the imputation matrix). By reformatting in wide, my time-varying covariates are now individual sequenced variables (eg, BMI1 BMI2...BMI12, BP1 BP2...BP12). I also now have 65 total variables (only 9 of which have complete data). After I have specified the imputation model (using all 65 variables), I get an error message of 'insufficient observations.' I can't seem to find out what this is referring to. Any suggestions? Is there a maximum to the number of variables allowed in the imputation model?
Also, for continuous variables in the imputation model, do I need to check for normality? If so, I assume I will need to transform the continuous variables and use the transformed variables in the imputation model?
Thanks for your help.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
