BookmarkSubscribeRSS Feed
🔒 This topic is solved and locked. Need further help from the community? Please sign in and ask a new question.
vitaaquaticus
Fluorite | Level 6

I ran two exactly same models using PROC GLIMMIX and PROC MIXED. I specify both PROCs to use a Gaussian distribution. I looked up the literature and they said that these two models with the same Gaussian distribution should yield similar results? However, I got two different results with respect to residuals (using plots as visual references). Why is this the case? Thank you!!

 

PROC GLIMMIX model

 

ods graphics on;
PROC GLIMMIX DATA = WORK.Data_Final_noAugust_EM plots=all METHOD=RSPL;
	CLASS Year Month Cape Site Transect Quadrat;
	MODEL 'Percent.Cover'n = Month Year|Cape/SOLUTION ddfm = KR CL ALPHA=0.05 INTERCEPT DIST=GAUSSIAN;
   RANDOM Quadrat(Transect) Transect(Site) Site(Cape) /CL ALPHA=0.05 TYPE=VC;
    LSMEANS Year|Cape / PDIFF CL ALPHA=0.05;
RUN;
ods graphics off;

Screen Shot 2020-04-01 at 14.49.12.png

PROC MIXED model

ods graphics on;
PROC MIXED DATA = WORK.Data_Final_noAugust_EM plots(MAXPOINTS=none)=all method=REML;
	CLASS Year Month Cape Site Transect Quadrat;
	MODEL 'Percent.Cover'n = Month Year|Cape/SOLUTION ddfm = KR CL ALPHA=0.05 INTERCEPT;
   RANDOM Quadrat(Transect) Transect(Site) Site(Cape) /CL ALPHA=0.05 TYPE=VC;
    LSMEANS Year|Cape / PDIFF CL ALPHA=0.05;
RUN;
ods graphics off;

Screen Shot 2020-04-01 at 14.50.47.png

1 ACCEPTED SOLUTION

Accepted Solutions
sld
Rhodochrosite | Level 12 sld
Rhodochrosite | Level 12

I'm guessing that the default type of residuals may differ between GLIMMIX and MIXED.

 

The GLIMMIX documentation (for the plot option on the GLIMMIX statement) says:

ALL

requests that all plots appropriate for the analysis be produced. In models with G-side random effects, residual plots are based on conditional residuals (by using the BLUPs of random effects) on the linear (linked) scale. Plots of least squares means differences are produced for LSMEANS statements without options that would contradict such a display.

 

The MIXED documentation (for the plot option on the MIXED statement) just says:

ALL

requests that all plots appropriate for the particular analysis be produced.

 

I suggest that you explicitly specify the type of residual for the plots that you want. If that doesn't sort it out for you, I suggest that you ask tech support.

 

I hope this helps.

 

 

 

View solution in original post

2 REPLIES 2
SteveDenham
Jade | Level 19

That is very odd, and I can't give you a good answer.  Maybe one of the Super FREQs will see this and have an answer. It is not obvious what is going on, but the right hand side of the distribution plot of the residuals shows a distinct second bump in MIXED and the residual v predicted plots look nothing alike.

 

SteveDenham

sld
Rhodochrosite | Level 12 sld
Rhodochrosite | Level 12

I'm guessing that the default type of residuals may differ between GLIMMIX and MIXED.

 

The GLIMMIX documentation (for the plot option on the GLIMMIX statement) says:

ALL

requests that all plots appropriate for the analysis be produced. In models with G-side random effects, residual plots are based on conditional residuals (by using the BLUPs of random effects) on the linear (linked) scale. Plots of least squares means differences are produced for LSMEANS statements without options that would contradict such a display.

 

The MIXED documentation (for the plot option on the MIXED statement) just says:

ALL

requests that all plots appropriate for the particular analysis be produced.

 

I suggest that you explicitly specify the type of residual for the plots that you want. If that doesn't sort it out for you, I suggest that you ask tech support.

 

I hope this helps.

 

 

 

SAS Innovate 2025: Call for Content

Are you ready for the spotlight? We're accepting content ideas for SAS Innovate 2025 to be held May 6-9 in Orlando, FL. The call is open until September 25. Read more here about why you should contribute and what is in it for you!

Submit your idea!

What is ANOVA?

ANOVA, or Analysis Of Variance, is used to compare the averages or means of two or more populations to better understand how they differ. Watch this tutorial for more.

Find more tutorials on the SAS Users YouTube channel.

Discussion stats
  • 2 replies
  • 832 views
  • 2 likes
  • 3 in conversation