turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- JMP User Community
- :
- Discussions
- :
- Discussions
- :
- very stupid newbie question

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Mar 6, 2009 9:01 AM
(997 views)

I apologize for the simplicity of this question, but I'm struggling with a conceptual mapping problem. It has been literally 20 years since I was working with regression software.

In the old days, in building a multiple regression that used "dummy variables," we'd simply specify the dummy (1 if true, 0 if false) and chuck it into the fit command. The results would then include estimates for both the regular variables and the dummy. So e.g., if there are three variables -- PRICE, SQUARE-FOOTAGE, FOR-SALE, where FOR-SALE means the house is currently for sale, and is a 1/0 dummy, you'd dump all three into the calculation, and the equation would be something like Y = a + b*SQUARE-FOOTAGE + c*FOR-SALE + e.

When I try to use the Fit Model to do this, however, I can't seem to get the model to include an estimate for the dummy. It gives me an equation with out the dummy, and then a separate graph/statistics for the dummy. There must be something very simply I'm missing here. I'd be grateful for a hint.

In the old days, in building a multiple regression that used "dummy variables," we'd simply specify the dummy (1 if true, 0 if false) and chuck it into the fit command. The results would then include estimates for both the regular variables and the dummy. So e.g., if there are three variables -- PRICE, SQUARE-FOOTAGE, FOR-SALE, where FOR-SALE means the house is currently for sale, and is a 1/0 dummy, you'd dump all three into the calculation, and the equation would be something like Y = a + b*SQUARE-FOOTAGE + c*FOR-SALE + e.

When I try to use the Fit Model to do this, however, I can't seem to get the model to include an estimate for the dummy. It gives me an equation with out the dummy, and then a separate graph/statistics for the dummy. There must be something very simply I'm missing here. I'd be grateful for a hint.

3 REPLIES

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Apr 14, 2009 3:55 PM
(986 views)

Hi Lessig,

I am new to SAS so I am not sure how helpful I can be but have you tried putting For-Sale into a class statement? I don't think you even need to code it as a dummy variable then.

ie. before your model statement put:

Class For-Sale;

cheers,

Shmea

I am new to SAS so I am not sure how helpful I can be but have you tried putting For-Sale into a class statement? I don't think you even need to code it as a dummy variable then.

ie. before your model statement put:

Class For-Sale;

cheers,

Shmea

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Apr 14, 2009 4:38 PM
(986 views)

Maybe you need to make this dummy variable numeric? It could be of some different type. If you look at the little sign to the left - is it the same blue triangle as for the others?

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Apr 15, 2009 5:29 AM
(986 views)

If your 0/1 variable is set to have modelling type of nominal, you should see a model term for both the 0 case and 1 case. If your 0/1 variables is set to have modeling type of continuous, then you should get the equation as you wrote it.