st: condition on imputed data with multiple imputation

["Mustillo, Sarah A" <smustill@purdue.edu>]

I don't think it's necessary to go through all of those steps. With multiple imputation, it's more about the big picture. That is, it doesn't matter so much that the data in the individual cases make sense - it matters more that the means, variances, and relationships are properly represented. 

That said, I wonder if you could handle this by the way you structure your dataset.  Is it long or wide and what type of model are you running? If you are studying the effects of medication on birthweight, could you just eliminate observations after a woman gives birth so that you won't be imputing anything at 38 weeks if the woman gave birth at 36 weeks? Give us more info on your data and your analysis and we can probably provide a better answer to your question...

Sarah




On Jan 18, 2011, at 4:07 PM, "Stas Kolenikov" <skolenik@gmail.com> wrote:

> You can probably impute the time of intake relative to delivery, which
> is a negative variable. Flip the sign to make it positive, take logs,
> and you can impute that with -regress-. Then you can produce a passive
> intake variable out of it. I know that's convoluted, but that might be
> the easiest way to go.
> 
> If you persuade Patrick Royston, he might be able to add support of
> -glm- command, along with the other supported ones. For your case, a
> gamma family with log link might seem like a reasonable option. If you
> are really adventurous, you can try to hack the -ice- code to find
> where an extra command might go, and add it there.
> 
> On Tue, Jan 18, 2011 at 12:59 PM, bichtram huynh <bichtrambe@hotmail.com> wrote:
>> hello everyone,
>> 
>> i'd like to know if it is possible to put limits on imputed data using MICE (ice)?
>> 
>> i want to study the influence of the timing of a medication intake during
>> pregnancy on a birth outcome (for example:birth weight)
>> i have some missing values on the date of the medication intake during pregnancy.
>> So i want to impute  them.
>> 
>> My problem is some women deliver at 36 weeks of gestation, others deliver at 41 weeks...
>> And by imputing data, sometimes i have an imputed data on medication intake
>> at 38 weeks whereas the woman has delivered at 36 weeks.
>> 
>> My question is: is it possible to put a condition so the imputed date of medication intake<term of delivery.
>> thanks for your help,
>> Flo
>> 
>> 
>> 
>> 
>> 
>> 
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/statalist/faq
>> *   http://www.ats.ucla.edu/stat/stata/
>> 
> 
> 
> 
> -- 
> Stas Kolenikov, also found at http://stas.kolenikov.name
> Small print: I use this email account for mailing lists only.
> 
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/