topic Re: Subtracting the confidence interval in New SAS User

Subtracting the confidence interval

kngu022 — Sun, 02 Feb 2020 23:07:04 GMT

I would like to subtract the 95% confidence interval out. Could anyone suggest me the way to do it please?

Re: Subtracting the confidence interval

kngu022 — Sun, 02 Feb 2020 23:18:10 GMT

PROC GLM DATA = A3_3;
	CLASS racegp;
	MODEL lbmi = racegp sex age/SOLUTION CLPARM;
	LSMEANS racegp/PDIFF STDERR CL;
	FORMAT racegp racefmt.;
	OUTPUT  OUT = A3_4 LCL=lcl UCL=ucl lsmean= lsm; 
RUN;

I tried above code but the outcome is not right. ANYONE CAN help please?

Re: Subtracting the confidence interval

unison — Mon, 03 Feb 2020 02:09:41 GMT

I would like to subtract the 95% confidence interval out.

Do you mean, remove it from the output (i.e. show the table without the confidence interval showing)?

Please elaborate.

Re: Subtracting the confidence interval

PGStats — Mon, 03 Feb 2020 03:55:04 GMT

Use an ODS output dataset, and do whatever calculation is needed. Add the statement:

ods output parameterestimates=pe;

to your analysis. The parameter estimates data will be written to dataset pe.

Re: Subtracting the confidence interval

kngu022 — Mon, 03 Feb 2020 06:11:50 GMT

Hi Unison,

I mean I want to use the confidence interval only for further analysis. Could you tell me what I should do next?

Re: Subtracting the confidence interval

PaigeMiller — Mon, 03 Feb 2020 12:13:41 GMT

Can you give us a longer and more detailed explanation of what you want, instead of these very brief descriptions that can mean a number of different things?

Re: Subtracting the confidence interval

Reeza — Mon, 03 Feb 2020 16:04:31 GMT

Here's some instructions and explanations on how to capture output that is shown to a data set.
https://blogs.sas.com/content/sastraining/2017/03/31/capturing-output-from-any-procedure-with-an-ods-output-statement/

You need to capture the data into a new data set. Instructions above are available on how to get that data into a data set. Depending on what you're trying to do next, there may be other, better ways but we don't know what you're trying to do overall.