R help - Sep 2007 - using transcan for imputation, categorical variable

Dear all,

I am using transcan to impute missing values (single imputation). I have
several dichotomous variables in my dataset, but when I try to impute
the missings sometimes values are imputed that were originally not in
the dataset. So, a variable with 2 values (severe weight loss or
no/limited weight loss) for example coded 0 and 1, shows 3 different
values after imputation (0, 1 and 2).

I have tried two options:

impfile <- transcan (~age + factor(dumgend) +factor(dumcomorb25_i) +
factor(dumwght) + factor(dumsmok)
I(lngtvextra) + eqd2t_i + factor(chemo)+ factor(stage),data=surv.df,
imputed=TRUE )

and this one (with categorical="dumwght")

impfile <- transcan (~age + factor(dumgend) +factor(dumcomorb25_i) +
factor(dumwght) + factor(dumsmok)
I(lngtvextra) + eqd2t_i + factor(chemo) + factor(stage),data=surv.df,
categorical=("dumwght"), imputed=TRUE )

Have can I handle this problem?

Thanks for your help.

Cary


	[[alternative HTML version deleted]]

Cary Dehing-Oberije wrote:
> Dear all,
> 
> I am using transcan to impute missing values (single imputation). I have
> several dichotomous variables in my dataset, but when I try to impute
> the missings sometimes values are imputed that were originally not in
> the dataset. So, a variable with 2 values (severe weight loss or
> no/limited weight loss) for example coded 0 and 1, shows 3 different
> values after imputation (0, 1 and 2).
> 
> I have tried two options:
> 
> impfile <- transcan (~age + factor(dumgend) +factor(dumcomorb25_i) +
> factor(dumwght) + factor(dumsmok)
> I(lngtvextra) + eqd2t_i + factor(chemo)+ factor(stage),data=surv.df,
> imputed=TRUE )
> 
> and this one (with categorical="dumwght")
> 
> impfile <- transcan (~age + factor(dumgend) +factor(dumcomorb25_i) +
> factor(dumwght) + factor(dumsmok)
> I(lngtvextra) + eqd2t_i + factor(chemo) + factor(stage),data=surv.df,
> categorical=("dumwght"), imputed=TRUE )
> 
> Have can I handle this problem?
> 
> Thanks for your help.
> 
> Cary
When a variable is a factor, transcan only imputes actual values. 
Please create a tiny example replicating your problem using simulated 
data, and I'll take a further look.

Frank
> 
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
> 

-- 
Frank E Harrell Jr   Professor and Chair           School of Medicine
                      Department of Biostatistics   Vanderbilt University