GLIMMIX for predicted marginal effect by variables in the random effect

Ruan1 — Sun, 05 Jan 2020 21:47:40 GMT

I am using the GLIMMIX model to do an age-period-cohort analysis. I need a graph shows how the predicted probability of Y_var for each education level changes by COHORT. Just as the picture attached. I am new at SAS. So how can I get the graph? or how can i get the predicted numbers by education by cohort?

********* HERE IS THE CODE---------------------------;
proc glimmix data=NHIS_2 maxopt=250000;
class PERIOD COHORT EDUCATION RACE SEX;
model Y_var (event='1') = AGE AGE_2 SEX RACE EDUCATION EDUCATION*AGE /solution CL
dist=binary;
random intercept SEX RACE /sub=PERIOD solution;
random intercept /sub=COHORT solution;
covtest GLM / WALD;
NLOPTIONS TECHNIQUE=NRRIDG;

RUN;

Re: GLIMMIX for predicted marginal effect by variables in the random effect

Ruan1 — Sun, 05 Jan 2020 21:49:56 GMT

I try the effectplot, but it seems the "x=var" only permits the variable included in the model. But the "COHORT" var is in the RANDOM part. SO):

proc plm restore=model2;
effectplot slicefit(x=age sliceby=education) / clm;
run;

topic GLIMMIX for predicted marginal effect by variables in the random effect in SAS Programming

GLIMMIX for predicted marginal effect by variables in the random effect

Re: GLIMMIX for predicted marginal effect by variables in the random effect