proc GENMOD vs GLM

[Polina_Utah]

Could you please explain why I'm getting different results when I ran this two procedures?

title2 "Multiple linear regression using Proc Genmod categorical smoking covariate";

proc genmod data=included2 ;

  weight MEC6YR;

  class smoker RIAGENDR RIDRETH1;

  model LBDVID=smoker RIAGENDR RIDAGEYR RIDRETH1;

run;

title2 "Multiple linear regression using Proc GLM categorical smoking covariate";

proc glm data=included2 ;

  weight MEC6YR;

  class smoker RIAGENDR RIDRETH1;

  model LBDVID=smoker RIAGENDR RIDAGEYR RIDRETH1/ solution;

run;

[1zmm · Accepted solution]

PROC GENMOD uses likelihood ratios rather than the sums of squares used in PROC GLM.  Specify in the MODEL statement of PROC GENMOD the options, WALD, TYPE3, and TYPE1.  The WEIGHT statements of the two procedures also perform different functions (cf., the documentation).