topic ANOVA in Statistical Procedures

ANOVA

cjohnson — Wed, 14 Oct 2009 14:21:51 GMT

As I understand it, ANOVA with categorical variables is equivalent to Linear Regression with dummy variables. But, when I run ANOVA, I do not get coefficients. Can anyone tell me why this is, or if there is an option to add these?

Thanks,
Chris

Re: ANOVA

Doc_Duke — Wed, 14 Oct 2009 14:45:16 GMT

Not that I know of. However, GLM will provide the coefficients.

Re: ANOVA

cjohnson — Wed, 14 Oct 2009 14:52:28 GMT

I don't get the coeff when I run glm either. If I use glm with continuous indep and dep variables, I get the coeff, but not when I use catagorical indep and continuous dep.

Re: ANOVA

Susan — Wed, 14 Oct 2009 15:12:13 GMT

Perhaps the SOLUTION option on the MODEL statement (available in the GLM, MIXED, and GLIMMIX procedures) provides what you're looking for.

If not, then it would be helpful if you posted your code.

--Susan

Re: ANOVA

deleted_user — Wed, 14 Oct 2009 16:15:56 GMT

In the model statement /solution or /s will give all the coefficients for fixed effects both class and otherwise

Re: ANOVA

cjohnson — Wed, 14 Oct 2009 17:07:41 GMT

Thanks very much! The solution option did the trick.

PROC GLM DATA=CLAIMS;
CLASS INJURY;
MODEL INCURRED = INJURY /SOLUTION;
MEANS INJURY / HOVTEST;
RUN;

Re: ANOVA

deleted_user — Wed, 14 Oct 2009 18:54:16 GMT

I see Susan recommended the same "solution"