Warning Messages with PROC MI

hotaylor — Wed, 19 Apr 2017 22:00:36 GMT

Hello!

I have a question regarding the PROC MI procedure. I have received these warnings when I run the PROC MI procedure with the following variables, which includes a mixture of continuous and dummy coded (dichotomous) variables:

PROC MI DATA=OKAY2 NIMPUTE=10 OUT=DOPE.All_Imputed2 SEED=1256;
VAR Loneliness OTHER BLACK Female Age Months_In_HC LessThanHighSchool HighSchool SomeCollegeorHigher
ANXIETY_TOTAL_C2 DEPRESSION_TOTAL_C2 SelfPhysicalHealth SelfMentalHealth GroupMeetLow GroupMeetMed
GroupMeetHigh VolunteerLow VolunteerMed VolunteerHigh ImportTalk2 FullEmotionSupp;
RUN;

Here are the warning messages that I get in the Log:

NOTE: Writing HTML Body file: sashtml.htm
WARNING: A covariance matrix computed in the EM process is singular. The linearly dependent
         variables for the observed data are excluded from the likelihood function. This may
         result in an unexpected change in the likelihood between iterations prior to the final
         convergence.
NOTE: The EM algorithm (MLE) converges in 13 iterations.
NOTE: The EM algorithm (posterior mode) converges in 8 iterations.
WARNING: The initial covariance matrix for MCMC is singular. You can use a PRIOR= option to
         stabilize the inference.
WARNING: The posterior covariance matrix is singular. Imputed values for some variables may be
         fixed.
NOTE: The data set DOPE.ALL_IMPUTED2 has 1480 observations and 103 variables.
NOTE: PROCEDURE MI used (Total process time):
      real time           3.37 seconds
      cpu time            1.06 seconds

Even with these warnings in the PROC MI procedure, and I was able to get an imputed dataset and conduct my regression models using the PROC GLM and PROC MIANALYZE procedures. That being said, I had a few questions regarding these warnings. First, what do these warnings mean? Second, is it okay to run analyses given these warnings? Third, will these warnings significantly affect my results? Fourth, I have a total of 148 respondents in my dataset. Could the relatively small sample size also be affecting the imputation?

When I run analysis using PROC REG, using listwise deletion, I basically get the same results.

I'm using SAS Version 9.4.

Any and all help would be greatly appreciated. Please let me know if you need anymore information.

Thank You!

topic Warning Messages with PROC MI in Statistical Procedures

Warning Messages with PROC MI