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
- :
- What's the difference between scaled estimates and standardized betas in multipl...

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
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Jan 17, 2010 9:44 AM
(2888 views)

Could anyone give me a simple explanation of the difference between JMP's scaled estimates and standardized betas, and when one should use one vs. the other in reporting the results of a multiple regression analysis? I have been using scaled estimates for some time when reporting regression results, but most journal articles I see use standardized betas. They both seem to be doing the same thing, but JMP's scaled estimates are larger than std betas when I compare them with the same data. Thanks in advance for any advice!

1 ACCEPTED SOLUTION

Accepted Solutions

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Jan 18, 2010 10:12 AM
(3209 views)

From the jmp guide:

"the Scaled Estimates command on the Effect Screening menu gives coefficients corresponding to factors that are scaled to have a mean of zero and a range of two. If the factor is symmetrically distributed in the data then the scaled factor will have a range from –1 to 1. This corresponds to the scaling used in the design of experiments (DOE) tradition. Thus, for a simple regressor, the scaled estimate is half the predicted response change as the regression factor travels its whole range."

By standardized betas (beta coefficients) they are normalized based on the scales of the factors (whose range could be much larger than 2)...so how they compare is dependent on the actual scales. They are both intending to do the same thing and that is to normalize the coefficients so they are easily compared.

"the Scaled Estimates command on the Effect Screening menu gives coefficients corresponding to factors that are scaled to have a mean of zero and a range of two. If the factor is symmetrically distributed in the data then the scaled factor will have a range from –1 to 1. This corresponds to the scaling used in the design of experiments (DOE) tradition. Thus, for a simple regressor, the scaled estimate is half the predicted response change as the regression factor travels its whole range."

By standardized betas (beta coefficients) they are normalized based on the scales of the factors (whose range could be much larger than 2)...so how they compare is dependent on the actual scales. They are both intending to do the same thing and that is to normalize the coefficients so they are easily compared.

3 REPLIES

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Jan 18, 2010 10:12 AM
(3210 views)

"the Scaled Estimates command on the Effect Screening menu gives coefficients corresponding to factors that are scaled to have a mean of zero and a range of two. If the factor is symmetrically distributed in the data then the scaled factor will have a range from –1 to 1. This corresponds to the scaling used in the design of experiments (DOE) tradition. Thus, for a simple regressor, the scaled estimate is half the predicted response change as the regression factor travels its whole range."

By standardized betas (beta coefficients) they are normalized based on the scales of the factors (whose range could be much larger than 2)...so how they compare is dependent on the actual scales. They are both intending to do the same thing and that is to normalize the coefficients so they are easily compared.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Jan 19, 2010 7:40 PM
(2725 views)

Thanks. I had read that description in the jmp guide, and gathered that it was something similar to what is done to standardize regression coefficients, but was surprised that the two approaches gave such different coefficient values. Does this imply that when dealing with independent variables with wide variations in range it might be better to use standardized beta coefficients than scaled estimates? Or are they both equally valid ways of comparing the effect of different independent variables in a regression?

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Jan 25, 2010 12:04 PM
(2725 views)

It depends on if you are just interested in analysis or in writing an equation which will need to be interpreted.