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
- /
- Analytics
- /
- Stat Procs
- /
- Output Standard Deviation Rather than STDERR in GL...

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

05-04-2011 12:46 PM

I'm working on standardizing my coeffient estimates for a report and need to use the standard deviation rather than the standard error. I know I can calculate by hand to get standard deviations but was hoping that a code may exist to output them faster. Thank you very much!

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

05-06-2011 04:59 PM

Your question is not clear. There are different types of standardizations for estimated regression coefficients. The usual one involves the standard deviation for both Y and X (and the regression model parameter estimate). But things are not that clear cut with mixed models (including generalized linear mixed models). I assume you are dealing with mixed models because you put GLIMMIX in the title of your post. With random effects, there is no simple choice for standardization, because there are levels to the variation in Y. You could do the standardization outside of GLIMMIX by simply getting the standard deviations of your variables with proc means (or some other choice), and then just using the regular formula for a standardized coefficient (doing this by hand). But I would consider this to be ad hoc, and I would be cautious in the interpretation.

One can actually get a different standardized estimated fixed effect parameter by adding the STDCOEF option on the MODEL statement. I think this scales the fixed effect parameter estimates strictly by the square root of the sum of squares of X, and does not use any scaling based on Y.

One can actually get a different standardized estimated fixed effect parameter by adding the STDCOEF option on the MODEL statement. I think this scales the fixed effect parameter estimates strictly by the square root of the sum of squares of X, and does not use any scaling based on Y.

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

10-09-2013 08:00 AM

Hello Ivm could you please, give me more information about how STDCOEF works or how to find that information? I get values of 29 and 20 as the stdcoef of my signicant variables, but i don't know how interpreting them beyond the fact tat the one with 29 is "better" than the 20. any suggestions?

thanks

kj

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

10-09-2013 10:16 AM

There is very little information in the GLIMMIX documentation on this. My previous posting from ages ago basically tells you what is given in the documentation. In essence, for the fixed effects, GLIMMIX apparently scales (standardizes) all the X variables (including interactions) internally; this would make the optimization methods more stable (avoiding extremely large or small values). This scaling involves dividing each X by the square root of the sum of squares of the relevant X variables.This is not the usual standardization in linear models where the standard deviation of the Y is also utilized. After model fitting, 'unscaling' is done to get the parameters back on the scale of the original X (this is what is displayed). If you put in the STDCOEF option, you will get the scaled parameter estimates. You can read about this in the help documentation for GLIMMIX. Go to STDCOEF (under MODEL), and then go to NOCENTER option to see the equations. Centering means subtracting mean X (it can be done with or without this).

I don't think you should overly interpret the standardized parameter estimates. These are not the same as those you get in linear models (they are not unitless because the Y units are still embedded in the parameters, based on the documentation). However, since the response variable is the same for all the X predictors, then I suppose that one can roughly compare the magnitude of the standardized parameter estimates. But I would be careful even doing that.