turn on suggestions

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

Showing results for

Find a Community

- Home
- /
- SAS Programming
- /
- SAS Procedures
- /
- PROC LOGISTIC: INTERPRETING DUMMY VAR SIGNIFICANCE

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
- Highlight
- Email to a Friend
- Report Inappropriate Content

02-10-2009 10:15 AM

Thanks to those who answered my question as to why SAS would set my B's = 0 for 'redundant' qualitative X's. I should have known this.

Now my questions is, how do I determine statistical significance of the parameters with B set to 0 since no p-value is given. Are DEV_WILL and GEN_CORNELL

significant?

Does the significance of the intercept term imply anything about this?

OUTPUT

Note: The following parameters have been set to 0, since the variables are a linear combination of other variables as shown.

DEV_WILL0 = Intercept - DEV_SMITH0 - DEV_JONES0

GEN_CORNELL0 = Intercept - GEN_JACKSON0 - GEN_DAVIS0

Thanks again.

Now my questions is, how do I determine statistical significance of the parameters with B set to 0 since no p-value is given. Are DEV_WILL and GEN_CORNELL

significant?

Does the significance of the intercept term imply anything about this?

OUTPUT

Note: The following parameters have been set to 0, since the variables are a linear combination of other variables as shown.

DEV_WILL0 = Intercept - DEV_SMITH0 - DEV_JONES0

GEN_CORNELL0 = Intercept - GEN_JACKSON0 - GEN_DAVIS0

Thanks again.

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

02-10-2009 11:03 AM

This issue gets into parameterization, which is hard to explain via computer. But it helps to ask yourself what question you're trying to answer, which sounds basic, I know, but it's amazing how often that very clearly and very specifically defining the question helps you find the answer.

You ask how you know if CORNELL is significant. Signficant compared to what? Remember, CORNELL isn't a continuous variable such as AGE, where you get a parameter estimate and a p-value and that tells you what happens as AGE changes. Instead, CORNELL is one general.

The beta and p-value for JACKSON will tell you whether JACKSON is different from CORNELL (or equivalently whether CORNELL is different from JACKSON).

The beta and p-value for DAVIS will tell you whether DAVIS is different from CORNELL (or equivalently whether CORNELL is different from DAVIS).

And you could write a CONTRAST statement to determine whether JACKSON is different from DAVIS.

Those are typically the only kinds of questions I ask in situations like that. But there is a way to answer the question "Is CORNELL different from all generals combined?" and I don't know what it is right now but perhaps that's not what you really need to know anyway.

You ask how you know if CORNELL is significant. Signficant compared to what? Remember, CORNELL isn't a continuous variable such as AGE, where you get a parameter estimate and a p-value and that tells you what happens as AGE changes. Instead, CORNELL is one general.

The beta and p-value for JACKSON will tell you whether JACKSON is different from CORNELL (or equivalently whether CORNELL is different from JACKSON).

The beta and p-value for DAVIS will tell you whether DAVIS is different from CORNELL (or equivalently whether CORNELL is different from DAVIS).

And you could write a CONTRAST statement to determine whether JACKSON is different from DAVIS.

Those are typically the only kinds of questions I ask in situations like that. But there is a way to answer the question "Is CORNELL different from all generals combined?" and I don't know what it is right now but perhaps that's not what you really need to know anyway.