topic Re: What parameter to use for DDFM= in proc mixed? in Statistical Procedures

What parameter to use for DDFM= in proc mixed?

Dicarlis — Wed, 10 Apr 2024 13:57:08 GMT

Good morning, I'm currently analyzing a much larger database than I'm used to.
For this analysis I need to use the proc mixed with repeated measures, however when I test the covariance matrices and use the command DDFM=KR or DDFM=kr2 I have two problems:
1- the script does not work due to lack of memory even using options memsize=; independent of how much memory you use (max obs used 20G)
2- The DEN F appear as infinite,
However, I was unable to understand the differences between DDFM= betwithin, contain, kenwardroger, kenwardroger2, residual or satterthwaite.
Here is the model used:
PROC MIXED;
CLASS id_Animal TRT day;
MODEL visit = TRT dday trt*ddai / ddfm= ?? ;
RANDOM id_Animal ;
REPEATED day /TYPE = ?? SUBJECT = id;
RUN;

Re: What parameter to use for DDFM= in proc mixed?

Ksharp — Thu, 11 Apr 2024 03:08:04 GMT

You could use "ddfm= BETWITHIN " and "RANDOM int /subject=subjectid" to save memory .


PROC MIXED;
CLASS id_Animal TRT day;
MODEL visit = TRT dday trt*ddai / ddfm= BETWITHIN  ;
RANDOM int /subject=id_Animal ;
REPEATED day /TYPE =ar(1) SUBJECT = id_Animal;
RUN;

Re: What parameter to use for DDFM= in proc mixed?

SteveDenham — Mon, 06 May 2024 14:38:24 GMT

If you have 20G observations, and have DEN F coming out as Inf, think about what that means - you are out at the point where F tests are indistinguishable from chi squared tests, and the t distribution doesn't differ from the normal until you get out to several decimal places. So, rather than trying DDFM= basically anything, try using the CHISQ option, and use those results rather than the F tests.

If you ran the analysis in GLIMMIX, you could use ddfm=none.

SteveDenham