Programming the statistical procedures from SAS

MIANALYZE AND TESTING JOINT EFFECTS

Reply
N/A
Posts: 1

MIANALYZE AND TESTING JOINT EFFECTS

I am using PROC MIXED to run a 3-level linear regression. I am also using imputed data so I am using PROC MIXED in conjunction with PROC MIANALYZE. I would like to use the TEST statement in PROC MIANALYZE to test a joint effect of two parameter estimates. My code is as follows:

proc mixed data=imputed covtest method=ml;

class PUBID METROID;

model risk=male age race age*race seg seg*race/solution covb;

by _Imputation_;

random intercept / type=un subject=METROID;

random intercept / type=un subject=METROID*PUBID;

ods output solutionF=mixparms CovB=mixcovb;

run;

proc mianalyze parms=mixparms covb(effectvar=rowcol)=mixcovb;

modeleffects intercept male age race age*race seg seg_race;

test1: test seg + seg_race;

run;

Basically, seg_race is the interaction between seg and race and I want to see if the combined estimates for seg and seg_race are significantly different from zero in this model. The model runs but I get an error that states: "WARNING: The TEST statement can be used only when each effect in the modeleffects list is a continuous variable." While my male, race, and seg variables are actually dichotomous, I treat them as continuous (0/1) in both the MIXED procedure and the MIANALYZE procedure. Any suggestions on how I can get this test to run, or a different approach if I want to test whether the combination of parameter estimates are equal to zero?

Thank you!

Katie

Ask a Question
Discussion stats
  • 0 replies
  • 173 views
  • 0 likes
  • 1 in conversation