----- Original Message -----
From: "Bellessa Frost, Michelle" <[email protected]>
To: <[email protected]>
Sent: Monday, October 20, 2003 12:36 PM
Subject: st: imputing categorical variables
> Hello,
>
> Is there any way to impute a value for data missing from categorical variables
in Stata? The Stata "impute" command uses OLS to estimate missing values,
appropriate only for continuous variables. I need to deal with missing data for
non-continuous variables.
>
> Theoretically, I could use logit and multinomial logit models, with the
predict command, to obtain predicted values for missing cases. But, when there
are missing cases in any of the variables I condition on (the independent
variables in the regression), I get a missing for the predicted variable in that
case.
>
> Has anyone dealt with something like this before?
>
> Thanks--Michelle
>
>
You may want to take a look at -whotdeck- and -hotdeckvar- From the help file
of hotdeckvar:
"Hotdeck imputation is especially useful for discrete variables (e.g 0/1 dummy
variable) where the imputed values shouldn't take any other values. Regression
imputation with impute would result into intermediate values (e.g. 0.56 for 0/1
dummy variables)."
Use -findit hotdeck- to locate and download.
Hope this helps,
Scott
*
* 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/
