Multiple Imputation

dikshar — Wed, 06 Mar 2019 04:10:32 GMT

I am pretty new to multiple imputations and I truly don't know how to troubleshoot the code. This is what I have so far.

PROC MI data= have
out = impute10
nimpute =10
seed =8842
round=1
simple
minimum= 0 0 0
maximum= 1 1 1;
var catvar1 catvar2 catvar3;
run;

proc logistic descending data = impute10 out =logistimp10;
class catvar1(ref = "1") catvar2(ref = "1") catvar3(ref = "0") catvar4(ref = "1") catvar5(ref = "1") catvar16(ref = "1")/ param = ref;
model result= var1 num var2 catvar4 catvar5 catvar6
bmi catvar1 catvar2 age catvar3/covb;
by _imputation_;
ods output ParameterEstimates=lgsparms CovB=lgscovb;
run;

proc mianalyze parms=lgsparms covb(effectvar=stacking)=lgscovb;
class catvar1 catvar2 catvar3 catvar4 catvar5 catvar6;
modeleffects var1 num var2 catvar4 catvar5 catvar6
bmi catvar1 catvar2 age catvar3;
run;

The issue is with proc mianalyze. I don't know how to handle categorical variables here besides having a class statement. I tried adding parms(classvar=level)=lgsparms but I get an error saying that level1 is not in the parms= dataset.

How should I recode this to properly include categorical variables?

I eventually also want to stratify by a categorical variable, will that be an issue?

Re: Multiple Imputation

Norman21 — Wed, 06 Mar 2019 06:31:08 GMT

You might find this page useful:

https://stats.idre.ucla.edu/sas/seminars/multiple-imputation-in-sas/mi_new_1/

topic Multiple Imputation in New SAS User

Multiple Imputation

Re: Multiple Imputation