topic Re: Degrees of freedom in PROC MIXED and PROC GLIMMIX in SAS Procedures

Degrees of freedom in PROC MIXED and PROC GLIMMIX

ANKH1 — Thu, 18 Oct 2018 16:54:02 GMT

Hello,

I want to report degrees of freedom for the F value from the output of PROC MIXED and PROC GLIMMIX. What should I report since the output only includes degrees of freedom for numerators and denominators?

Thanks in advance for your help,

Re: Degrees of freedom in PROC MIXED and PROC GLIMMIX

PaigeMiller — Thu, 18 Oct 2018 17:38:18 GMT

@ANKH1 wrote:

Hello,

I want to report degrees of freedom for the F value from the output of PROC MIXED and PROC GLIMMIX. What should I report since the output only includes degrees of freedom for numerators and denominators?

I believe this represents a mis-understanding on your part.

There is no such thing as a single "degrees of freedom for the F value". All F-values have both numerator and denominator degrees of freedom.

Re: Degrees of freedom in PROC MIXED and PROC GLIMMIX

ANKH1 — Thu, 18 Oct 2018 17:49:40 GMT

Thanks for responding. But what degrees of freedom do you report with your F value? Given that our model is a two way repeated measures design?

Re: Degrees of freedom in PROC MIXED and PROC GLIMMIX

PaigeMiller — Thu, 18 Oct 2018 21:48:11 GMT

There is no such thing as a single "degrees of freedom for the F value". All F-values have both numerator and denominator degrees of freedom.

You report both the numerator and denominator degrees of freedom

Re: Degrees of freedom in PROC MIXED and PROC GLIMMIX

ANKH1 — Fri, 19 Oct 2018 15:14:19 GMT

Hello,

I have been told that when you report the F value you have to include df. This is an example from SPSS:

For example you report the main effect of SRA as F(2, 1276) = 6.11, p = .002, ηp2 = .009.

SAS output only gives df for numerator and denominator. The first df is referring to the levels of the variable but the second is referring to the df of the error and the SAS output I get doesn't provide it. Do you know how can I get it?

Thanks!

Re: Degrees of freedom in PROC MIXED and PROC GLIMMIX

PaigeMiller — Fri, 19 Oct 2018 16:49:59 GMT

I guess I don't understand the relevance of an SPSS output to PROC MIXED.

Did you actually produce the output from PROC MIXED? This is an example PROC MIXED output which clearly produces numerator and denominator degrees of freedom: https://documentation.sas.com/?docsetId=statug&docsetTarget=statug_mixed_examples01.htm&docsetVersion=14.3&locale=en

Re: Degrees of freedom in PROC MIXED and PROC GLIMMIX

ANKH1 — Fri, 19 Oct 2018 17:10:20 GMT

Yes, I get that output from PROC MIXED but my issue is that reviewers ask to report F (df1, df2), p-value. The first df I get is the df for the numerator but the df2 is for the error which is not given in the SAS output. Do you know how to obtain the df for the error in SAS?

Re: Degrees of freedom in PROC MIXED and PROC GLIMMIX

PaigeMiller — Fri, 19 Oct 2018 17:23:37 GMT

I guess we're not communicating. That example shows both numerator and denominator DF for the F-test. I don't understand why you say it doesn't.

Re: Degrees of freedom in PROC MIXED and PROC GLIMMIX

ANKH1 — Fri, 19 Oct 2018 17:34:56 GMT

Sorry I haven't been clear. The degrees of freedom I am looking for to report are the df from the error not the numerator or denominator.

Re: Degrees of freedom in PROC MIXED and PROC GLIMMIX

PaigeMiller — Mon, 22 Oct 2018 12:37:22 GMT

DF for error is the same (different words, that's all) as DF for denominator

Re: Degrees of freedom in PROC MIXED and PROC GLIMMIX

ANKH1 — Mon, 22 Oct 2018 15:18:35 GMT

Oh, I did not know that. Thanks for the clarification and your help!