Partial F-test in GLM Proc

MarryEdward — Mon, 02 Dec 2013 17:02:57 GMT

Dear all,

In the following "Proc glm" I am using a panel regression with an interaction term (PIncP*Same4). I want to test for the joint significance of the variable "Same4" with "PIncP*Same4". I used the command "test Same4, PIncP*Same4;" as shown below. It runs but I get no result in the output. I am new to SAS so I am not sure if I am missing something in the proc glm below.

proc glm;

class Off Def OffSeason OffGame Down FieldPos Cold Wind Weather Grass ;

model YardsMinAvgP4 = PIgP Same4 PIncP*Same4 FracPassGameIgP PrevSucc4

YPRGameIgP YPPGameIgP FracSuccGame Down YTG FieldPos PointDiff DiffSpreadvsActual Cold Wind Weather Grass OffHome / solution ;

test Same4 , PIncP*Same4;

run;

Thank You!

Re: Partial F-test in GLM Proc

SteveDenham — Thu, 05 Dec 2013 14:30:43 GMT

The "test" statement in PROC GLM is a method to use some mean square other than the residual mean square as a denominator in an F test, and, so far as I can tell, not a way to do joint testing. If you really wish to do a joint test, you should look at procedures such as GLIMMIX that allow the use of the EFFECT statement or LSMESTIMATE statement.

Steve Denham

topic Partial F-test in GLM Proc in Statistical Procedures

Partial F-test in GLM Proc

Re: Partial F-test in GLM Proc