Programming the statistical procedures from SAS

GEE coefficients interpretation help

Reply
Regular Contributor
Posts: 204

GEE coefficients interpretation help

Dear all. Wish you all well.


Can someone please shed me some light on interpreting the GEE coefficients from Genmod modeled by Gamma log link?


From the output below, it says the response variable y in males is (exp(0.2008)-1)*100=22.23% more than females.  Please correct me if that is incorrect.


And how about the coefficient of the interaction terms?


And have I stated the GEE equation right as?

log(y)=3.8136+0.2008(sex=Male)-0.3503(1=Disagree a lot)-0.2428(2=Disagree a little)-0.3141(3=Neither agree nor disagree)-0.2483(4=Agree a little)+0.1692(year=2010)-0.013(year=2010,1=Disagree a lot)-0.0973(year=2010,2=Disagree a little)-0.0366(year=2010,3=Neither agree nor disagree)-0.0359(year=2010,4=Agree a little)


Your insight is greatly appreciated.

Analysis Of GEE Parameter Estimates
Empirical Standard Error Estimates
ParameterEstimate
Intercept3.8136
sexMale0.2008
sexFemale0
a11=Disagree a lot-0.3503
a12=Disagree a little-0.2428
a13=Neither agree nor disagree-0.3141
a14=Agree a little-0.2483
a15=Agree a lot0
Year20100.1692
Year20110
a1*Year1=Disagree a lot2010-0.013
a1*Year1=Disagree a lot20110
a1*Year2=Disagree a little2010-0.0973
a1*Year2=Disagree a little20110
a1*Year3=Neither agree nor disagree20100.0366
a1*Year3=Neither agree nor disagree20110
a1*Year4=Agree a little2010-0.0359
a1*Year4=Agree a little20110
a1*Year5=Agree a lot20100
a1*Year5=Agree a lot20110
Ask a Question
Discussion stats
  • 0 replies
  • 185 views
  • 0 likes
  • 1 in conversation