topic Re: Proc Mixed and family wise error in Statistical Procedures

Proc Mixed and family wise error

zz109 — Wed, 04 May 2022 12:28:21 GMT

I used Proc Mixed statements to run multiple path analyses to test the direct and indirect effects of my two IVs and two mediators on three DVs (2x2 within-between mixed experimental design).

One of the p-values associated with these paths was close to .05, and the reviewer raised a concern of experimental-wise type 1 error. I think family-wise type 1 error should be the concern of this case. I wonder whether the Proc Mixed procedures provide any statement or (default) function to adjust p-values to control the family-wise type1 error issue. Thanks a lot for any suggestions and help!

Re: Pros Mixed and family wise error

sbxkoenk — Wed, 04 May 2022 10:57:17 GMT

@zz109 wrote:

I wonder whether the Proc Mixed procedures provide any statement or (default) function to adjust p-values to control the family-wise type1 error issue. Thanks a lot for any suggestions and help!

I do not think so.
The p-values in PROC MIXED are "raw" (i.e. unadjusted).

Maybe you can have a look into MULTTEST Procedure.
SAS® 9.4 and SAS® Viya® 3.5 Programming Documentation | SAS 9.4 / Viya 3.5
SAS/STAT User's Guide
The MULTTEST Procedure
p-Value Adjustments
Familywise Error Rate Controlling Adjustments
https://go.documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/statug/statug_multtest_details11.htm

Cheers,

Koen

Re: Proc Mixed and family wise error

SteveDenham — Mon, 09 May 2022 12:42:00 GMT

Could you post the code you are using? I believe you can get family-wise adjustments of p values for differences between lsmeans using the ADJUST= option in the LSMEANS/LSMESTIMATE/ESTIMATE/CONTRAST statement, but it will depend on how you frame the comparisons. It may also require a GROUP= option for the random or repeated effect.

SteveDenham