turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- Mix procedures question

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

07-01-2015 11:06 AM

Hi everyone,

I am having an issue how to account for the extra error term in our mixed statistical model using the SAS mixed procedure. I want to treat it as a split-plot design to come up with an ANOVA table, and one error term should not include the factor L (see equation below).

My guess is something like:

Proc Mixed data=StrainGauges method=type3; class rep F R C L;

model S=F R C FR FC RC FRC L FL RL CL FRL FCL RCL FRCL/ddfm=kr;

Random rep(A|B|C)

However, I am not sure how to write the random statement (e.g. "Random rep(A*B*C)" or "Random rep(A|B|C)"). Can someone help me to write this code?

Thanks in advanced!

Josue

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

07-02-2015 03:30 PM

This is a case where you have several random effects, so the rep(A*B*C) is obviously not what you need (and it is just as obvious that it is the residual error).

A much better approach would be:

random intercept F R C F*R F*C R*C F*R*C/subject=rep;

Also in your model statement, be sure to represent interactions with the '*' operator. As it is now, this would treat FR, FC, RC and FRC as continuous (and undefined) variables.

Steve Denham

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

07-03-2015 09:54 AM

Thank you, Steve. Your comments will be very useful for our analysis!

Josue Bastos