topic Standardized beta coefficients for categorical variables in Proc Logistic in Statistical Procedures

Standardized beta coefficients for categorical variables in Proc Logistic

abarnett313 — Sat, 02 Dec 2017 00:32:01 GMT

Greetings,

I am trying to obtain standardized beta coefficients for the categorical variables in my model in Proc Logistic. I've used the STB option in the Model statement, but it only produces beta coefficients for the continuous variables. Is there a way to obtain these estimates for the categorical variables, too? (Or do such estimates not exist for categorical variables?) Any help would be greatly appreciated. I'm using SAS Enterprise Guide v 4.3.

Thanks!

Andrew

Re: Standardized beta coefficients for categorical variables in Proc Logistic

stat_sas — Sat, 02 Dec 2017 01:24:45 GMT

Hi,

Why standardized beta coefficients for the categorical variables?

Re: Standardized beta coefficients for categorical variables in Proc Logistic

Reeza — Sat, 02 Dec 2017 01:30:54 GMT

What version of SAS do you have, not EG.

You can run the following to see your SAS/STAT version which is the most important thing:

proc product_status;run;

Note that I could not replicate your issue. I ran the second example in the PROC LOGISTIC documentation with the STB option and I received parameter estimates for all variables, including the categorical variables.

@abarnett313 wrote:

Greetings,

I am trying to obtain standardized beta coefficients for the categorical variables in my model in Proc Logistic. I've used the STB option in the Model statement, but it only produces beta coefficients for the continuous variables. Is there a way to obtain these estimates for the categorical variables, too? (Or do such estimates not exist for categorical variables?) Any help would be greatly appreciated. I'm using SAS Enterprise Guide v 4.3.

Thanks!

Andrew

Re: Standardized beta coefficients for categorical variables in Proc Logistic

abarnett313 — Sun, 03 Dec 2017 19:48:02 GMT

Thanks for responding, and I apologize for not including the SAS version. It's SAS v9.3.

Re: Standardized beta coefficients for categorical variables in Proc Logistic

Reeza — Sun, 03 Dec 2017 20:23:40 GMT

Did you run the documentation example? Did that work for you? Like I mentioned, can't replicate your issue so I suspect it has to do with your data or your code and not an issue with PROC LOGISTIC or SAS.

Re: Standardized beta coefficients for categorical variables in Proc Logistic

abarnett313 — Mon, 04 Dec 2017 17:42:25 GMT

Thanks again for your help. Do you mean this example in the documentation?

https://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_logistic_sect053.htm

Re: Standardized beta coefficients for categorical variables in Proc Logistic

Reeza — Mon, 04 Dec 2017 17:44:57 GMT

Yes, but I highly recommend using the latest SAS documentation and the corresponding program is available as text:

http://documentation.sas.com/?docsetId=statug&docsetTarget=statug_logistic_examples02.htm&docsetVersion=14.2&locale=en

There's a small link at the top of the page to the text program that makes it easier to copy/paste and run the example.

Re: Standardized beta coefficients for categorical variables in Proc Logistic

abarnett313 — Mon, 04 Dec 2017 18:05:55 GMT

Ah, thank you! I used similar statements and didn't receive standardized coefficients for the categorical variables as part of the output. The documentation specifies that the /stb option only provides standardized coefficients for continuous variables. Now that I better understand what standardization means for logistic regression, I see that it doesn't really make sense for categorical variables. Thanks all the same!

Re: Standardized beta coefficients for categorical variables in Proc Logistic

Reeza — Mon, 04 Dec 2017 18:14:48 GMT

That's not what I get, but I agree with the concept of it not being valid.

proc logistic data=Neuralgia;
class Treatment Sex;
model Pain= Treatment Sex Treatment*Sex Age Duration / expb stb;
run;

Note that Treatment and Sex are categorical variables and there are standardized estimates for them.