Re: ANOVA and REG

deleted_user · Posted 09-22-2010 05:43 PM

I am creating a regression model, I find there is some difference between the GLM procedure(ANOVA) and REG procedure, which one should I use to creat a model?
And how to interepret them?
Any suggestion is appreciated.

Paige · Posted 09-23-2010 08:25 AM

In most (all?) comparable situations, the results will be the same.

GLM allows for categorical factors and interactions/polynomial terms.

REG offers regression diagnostics.

By the way, in answer to your question "which one should I use to create a model" ... you don't create a model in software ... the model is independent of software ... it can exist on a blackboard ... you have to specify the terms in the model. Software can estimate the parameters in your models. Message was edited by: Paige