Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Richard Williams <richardwilliams.ndu@gmail.com> |
To | statalist@hsphsun2.harvard.edu, <statalist@hsphsun2.harvard.edu> |
Subject | Re: st: Issues re: mi impute |
Date | Thu, 07 Jun 2012 19:44:51 -0500 |
At 04:26 PM 6/7/2012, Goin, Dana wrote:
I want to use -mi impute- but am having difficulties figuring out how to adapt it to what I need. I'm using Stata 12.1 for Windows. The dataset I'm using has poor information on family income -- it's categorical and missing many values, and I need it to be continuous and complete. I want to use another dataset (that has the same independent variables as the first dataset, but also has a continuous income variable) to fit an OLS model. I want to use this model to impute continuous income on my first dataset, but I can't figure out how to use previous estimation results for imputation. Any help/ideas would be much appreciated, thanks!
I wonder if you aren't expecting too many miracles from -mi impute-. Many of the cases are completely missing income, and the rest have only a rough approximation of it. Some wild guesses on how to proceed:
* In the good data, create a collapsed income using the same categories as in the bad data. In the bad data, create a continuous income var that is missing for all cases.
* If income is dependent, use intreg. You can see if you believe the intreg results by running the same intreg model in both data sets. You can impute the missing values for the collapsed income, but you often don't gain much by imputing the dependent variable.
* If income is independent, consider breaking it into dichotomies and using those in the model. You can do tests to see if the dichotomies can be combined into a single variable.
* Getting back to your original idea -- combine the 2 data sets after you have computed the income vars as I described above. Use mi impute chained on the continuous and categorical income variables, including the categorical income var in the imputation model for the continuous income. Use mi estimate selecting only the cases from the crummy data set. This gets you imputed continuous income values that take into account whatever useful info the collapsed income var contains.
Again, I think you just don't have the data to do what you want. But I guess you can try and see what happens.
------------------------------------------- Richard Williams, Notre Dame Dept of Sociology OFFICE: (574)631-6668, (574)631-6463 HOME: (574)289-5227 EMAIL: Richard.A.Williams.5@ND.Edu WWW: http://www.nd.edu/~rwilliam * * 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/