topic Re: Some advice on which PROC to use? in SAS Procedures

Some advice on which PROC to use?

GreggB — Tue, 20 Jan 2015 16:10:00 GMT

I want to create a model like so : dependent var = independent var1 + indep var2 +...indep var_n

My dependent var is dichotomous. (Not met = 0, met = 1)

My independent vars are categorical. I will have sex (M,F), meal_status (F,R), Race (W, B, I, H, P, O) eng_prof (1-9) and school_ID.

I want SAS to examine how much explanation each indep var contributes toward the model. (I have some experience using logistic regression. I'm just not sure in this case.)

I also recall a test that SAS will do that introduces one indep variable at a time to determine if it should be included in the model but I don't remember the particulars.

Any advice/comments would be a great help.

Re: Some advice on which PROC to use?

Reeza — Tue, 20 Jan 2015 17:32:53 GMT

The documentation has pretty good examples and details.

I'd consider looking into proc CATMOD and LOGISTIC.

Re: Some advice on which PROC to use?

ballardw — Tue, 20 Jan 2015 17:49:18 GMT

I also recall a test that SAS will do that introduces one indep variable at a time to determine if it should be included in the model but I don't remember the particulars

In Proc Logisitic (and a few others) you are likely thinking of the SELECTION option for the model statement.

Re: Some advice on which PROC to use?

Ksharp — Wed, 21 Jan 2015 06:59:33 GMT

independent variables are all category variables ?

Use proc CATMOD

and post is at SAS Statistical Procedures to get better answer.

Xia Keshan

Re: Some advice on which PROC to use?

FriedEgg — Wed, 21 Jan 2015 07:17:27 GMT

Use selection with LOGISTIC or create the design matrix that you want to see for the different effects in CATMOD (reference the documentation or ask if more help is needed).

http://support.sas.com/documentation/cdl/en/statug/63347/HTML/default/viewer.htm#statug_catmod_sect017.htm