topic Proc reg: Obs*R2 in SAS Procedures

Proc reg: Obs*R2

Golf — Tue, 26 May 2020 05:25:04 GMT

Hello,

Based on the result of the proc reg below.

proc reg   data = example outest=parameters; 
model  y =  A B C D E F;
run;

How can I use SAS to calculate Obs*Rsquare, instead of using calculator?

Thank you

Re: Proc reg: Obs*R2

ed_sas_member — Tue, 26 May 2020 07:24:31 GMT

Hi @Golf

PROC REG assigns a name to each table it creates in the results.

You can retrieve them through the ODS OUTPUT: https://documentation.sas.com/?docsetId=statug&docsetTarget=statug_reg_details52.htm&docsetVersion=15.1&locale=en

-> To retrieve the value of the R square (i.e. the % of variability explained by the model), the table name is FITSTATISTICS. Here is the code to be submitted before the PROC REG:

ods output fitstatistics = _fitstatistics;

What do yo mean by "Obs*Rsquare"?

Re: Proc reg: Obs*R2

Golf — Tue, 26 May 2020 10:54:14 GMT

Dear Ed,
Thanks for you response.
Obs*Rsquare is the number of observation multiply to R-square.
If I use "ods output fitstatistics = _fitstatistics", how can I know which symbols are used for the number of observation and the value of R-Square?
Best,

Re: Proc reg: Obs*R2

PaigeMiller — Tue, 26 May 2020 11:18:40 GMT

You probably need

ods output fitstatistics=fitstatistics nobs=nobs;

as the number of observations used in the regression is in the NOBS table.

how can I know which symbols are used for the number of observation and the value of R-Square?

You can know which variables contain the values you want by actually looking at these data sets that are created.

I'm skeptical that Obs*Rsquare has any particular meaning in a regression; why do you want this value?

Re: Proc reg: Obs*R2

Golf — Tue, 26 May 2020 13:12:09 GMT

I need obs*R-square to test overidentification based on the method of Sargant test.
Thanks.