topic Proc reg syntax in Statistical Procedures

Proc reg syntax

Mathis1 — Tue, 07 Jul 2020 14:31:32 GMT

Hello,

I don't really see the difference between those two statements :

proc reg data=Have;

model y= x1-x4;

run;

and

proc reg data=Have;

model y= x1 x2 x3 x4;

run;

Any answer ?

Thanks 🙂

Re: Proc reg syntax

Kurt_Bremser — Tue, 07 Jul 2020 14:50:02 GMT

As long as the variables appear in one of these orders in the dataset:

x1 x2 x3 x4
x1 x3 x2 x4

the statements are equal.

Re: Proc reg syntax

PaigeMiller — Tue, 07 Jul 2020 14:53:51 GMT

What differences do you see?

Re: Proc reg syntax

Mathis1 — Tue, 07 Jul 2020 15:15:14 GMT

I stumbled upon a multiple choice question on the internet.

Question is :

Identify the correct sas program for fitting a MLR model with dependent variable y and four predictor variables (x1-x4)/

The propositions are :

I don't quite see why answer D is not a correct answer as well as A.

Re: Proc reg syntax

Kurt_Bremser — Tue, 07 Jul 2020 15:18:28 GMT

Because SOLUTION is not a valid option for the MODEL Statement.

Maxim 1: Read the Documentation.

Re: Proc reg syntax

PaigeMiller — Tue, 07 Jul 2020 15:30:48 GMT

You could have run the code yourself and found out that D doesn't work. Or looked in the documentation.

A is not wrong, it should give the same answers as B, but unnecessary to use a VAR statement in this situation.

Re: Proc reg syntax

PGStats — Tue, 07 Jul 2020 18:23:12 GMT

To add to @PaigeMiller explanation, all variables named in the MODEL statement of proc REG are assumed to be numeric. The VAR statement, so often required in other procedures to distinguish numeric from categorical, isn't required. But proc reg can be used to fit many models interactively, it will compute results after every MODEL statement, not knowing yet if more models are to come. The VAR statement tells the procedure about all the variables that you may consider further along in your models. Knowing that, the procedure only needs to read your data once to accumulate all the required sums of crossproducts.