Programming the statistical procedures from SAS

Glimmix-Multilevel-Residuals

Reply
New Contributor
Posts: 2

Glimmix-Multilevel-Residuals

Hello everyone,

I would like to check for the normality and homoscedasticity   ( for 3 random effects) in  multilevel (4 levels , 3random intercept) of longitudinal data ( binary outcome) using proc glimmix.   When I run the following code ( plots= (StudentPanel PearsonPanel) gives  the residuals  all together . There is any way to get it for the different levels (random effects)? With boxplot option  (boxplot (subject npanelpos=20) it  gives separated plots  for each random effect but it but I am not sure how to interpret  it  with binary outcome . 

Any help  would be appreciated ,

ods graphics on;

proc glimmix data=PENPIG11  method=RSPL noclprint=20

plots=(StudentPanel  PearsonPanel)

plots=boxplot (subject npanelpos=20);;

class   FarmID CohortID_C PenID_C PigID  Nursc4 Mortpc;

model PigSalmCult (event='Pos')= AvHrT48h_ab Nursc4 Mortpc

                                 Age FarmID   /solution link=logit dist=binary

oddsratio( at age=10 unit age=2  at AvHrT48h_ab=0 unit AvHrT48h_ab=5); ;

nloptions tech=nrridg;;

random intercept/subject= CohortID_C;

random intercept/subject=PenID_C (CohortID_C);

random intercept /subject= PigID (PenID_C CohortID_C);

covtest zeroG / cl(type=plr) parms ;

Title' MT3.10.-AvHrT48h_ab_OR';

format   Nursc4 Mortpc  YNM.  PigSalmCult Cult.  FarmID FarmID. ;

run;

ods graphics off;
Alda

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