Date:         Fri, 10 Sep 2010 15:43:31 -0400
Reply-To:     Jordan Hoolachan <jihool3670@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Jordan Hoolachan <jihool3670@GMAIL.COM>
Subject:      error while using IVEware's RESTRICT option
Content-Type: text/plain; charset=ISO-8859-1

Dear All,

I am using SAS 9.2 on Windows XP Professional and am running IVEware
from within SAS in order to perform multiple imputation.  I have a
variable "med_mgmt_days1" which is a count variable and is either
missing or a non-negative value.  Additionally, I have another
variable "special_indic1" that has no missing data indicates what type
of value occurs in "med_mgmt_days1."  It has three levels: 0 - the
value in "med_mgmt_days" is not missing; 1 - the value is missing and
I want to impute it; 2 - the value is missing but I don't want to
impute it.

Obviously, I am trying to impute only those values of "med_mgmt_days1"
for which the corresponding value of "special_indic1" is 1.  As such,
I have created the following set up file:

data _null_;
  infile datalines;
  filename setup "impute.set";
  file setup;
  input;
  put _infile_;
datalines4;
  title Multiple imputation;
  datain iwif.iwif_indic;
  dataout iwif.iwif_imputed all;
  continuous <a list of variables>;
  categorical <a list of variables>;
  count med_mgmt_days1 <a list of variables>;
  mixed <a list of variables>;
  transfer special_indic1 <a list of variables>;
  restrict med_mgmt_days1(special_indic1=1);
  bounds age_years(>17);
  iterations 10;
  multiples 10;
  seed 2001;
  minrsqd 0.05;
  run;
;;;;

%impute(name=impute, dir=.);

From the IVE_user guide
(ftp://ftp.isr.umich.edu/pub/src/smp/ive/ive_user.pdf), I would expect
the above code to only impute med_mgmt_days1 when special_indic1=1.
However, I receive the following message

"Warning: No valid cases, imputing restricted value 0"

and IVEware imputes ALL non-missing values of med_mgmt_days1.  At
first I thought that perhaps I had too few cases so I recoded the data
and made sure there were ~30,000 (out of 250,000) observations for
which med_mgmt_days1=. and special_indic1=1.  Unfortunately, that
didn't help either and I'm at a loss as to what is going wrong.

Does anyone with experience using IVEware have any suggestions?

Thanks for your consideration!

-Jordan