Glimmix output: same factor got different significance value from "Solutions for Fixed Effects" and "Type III Tests of Fixed Effects"

[lidove]

Dear  All,

I am a beginner of SAS. I used glimmix procedure to model how a binary response variable was affected by several interested factors. The output gave inconsistent results of the significance test of one factor, "gender", from the "Solutions for Fixed Effects" and "Type III Tests of Fixed Effects" (please see the attached results). Could anybody tell me:  1) what's the reason for my results? Is it due to the nature of my dataset which is unbalanced?  2) And under this situation, how do I intepret my results?  which result should I belive, "Solutions for Fixed Effects" or "Type III Tests of Fixed Effects"?

Thanks for the help.

Solutions for Fixed Effects

Effect        Nestlingsex       Estimate    Standard Error        DF          t            P

Intercept                             2.7637         0.3415                 446.8    8.09     <0.0001

……

Gender            female       -0.8945         0.3448                  9237   -2.59     0.0095

Gender             male              0

……

Type III Tests of Fixed Effects

Effect         Num DF   Den DF   F Value       Pr > F

……

Gender          1              9237      0.04           0.8491

……

[SteveDenham]

That seems odd.  Can you share your PROC GLIMMIX code, and the entirety of your Type3 output?  I have a hunch you have some interactions with gender that might be significant, but I'm really just guessing.

Steve Denham

[lidove]

Thank you so much, Steve. The work is to examine what kinds of factors would affect whether a nestling bird would be fed during each feeding event.  The SAS code and type 3 output are as follows:

proc glimmixdata = sasuser.EACHALLOCATION  ;

class  adult  nestling gender  nest position  begging CB  No_feeding_event;

title GLIMMIX Each Feeding preferenc;

model F_N(event='1') = adult  gender  position begging  Tsize  CB adult*gender  adult*position  adult*begging adult*Tsize  adult*CB    gender*position        gender*begging    gender*CB  gender*Tsize position*begging   position*CB  position*Tsize   begging*CB begging*Tsize   Tsize*CB NO_begging_nestlings        /solution  ddfm=kr link = logit   dist = binomial ;

random intercept  no_feeding_event / subject=nest;

run;

Type III Tests of Fixed Effects
Effect	Num DF	Den DF	F Value	Pr > F
Adult	2	6925	1.69	0.1845
Gender	1	9237	0.04	0.8491
Position	1	9237	844.43	<.0001
Begging	1	9237	338.44	<.0001
Tsize	1	9237	2.06	0.1516
CB	1	32.61	0.22	0.6394
Adult*Gender	2	9237	3.66	0.0259
Adult*Position	2	9237	0.17	0.8460
Adult*Begging	2	9237	4.29	0.0138
Tsize*Adult	2	9237	4.48	0.0114
Adult*CB	1	7807	0.18	0.6747
Gender*Position	1	9237	2.61	0.1060
Gender*Begging	1	9237	2.12	0.1454
Gender*CB	1	5463	1.47	0.2253
Tsize*Gender	1	9237	0.01	0.9216
Position*Begging	1	9237	9.41	0.0022
Position*CB	1	9237	1.81	0.1785
Tsize*Position	1	9237	1.43	0.2313
Begging*CB	1	9237	2.19	0.1390
Tsize*Begging	1	9237	0.32	0.5746
Tsize*CB	1	9237	0.23	0.6333
No_begging_nestlings	1	2185	45.46	<.0001

[SteveDenham]

It doesn't depend on the significance of the interaction--the mere inclusion of the interaction in the model will lead to differences between the solution values and the F test values.  I found this to be true even with balanced data.  Thus, I would trust the F test values, for a couple of reasons.  It considers the effect after "factoring out" all the other effects--it is meant to be an orthogonal test to all the other tests.  The theory part of my brain seems to be non-functional this morning, but I know this difference should be apparent to me in the face of including interactions.  The agreement will probably get worse with imbalance, especially with a Kenward-Rogers adjustment.

Steve Denham