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
- /
- Setting up the random effects model in Proc mixed

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

08-24-2011 05:59 PM

My experiment design is as follows.

Each sample in the experiment is run on 2 “kit”s. Each "kit" has 2 “operator”s. Each operator performs the experiment in 2 “lab”s. And,each of these combinations uses a different “part” of the sample-that’s a total of 8 different “part”s from the sample run randomly on different “kit-operator-lab” combinations. (another option is to use 4 "part"s from the same sample and run the same combination on the 2 kits)

All the variables "kit, operator, lab" are random effects and I would like to estimate the variance due to each of these.

It would be very helpful to know how to specify these effects in the "random" statement. Something like this?

proc mixed;

class kit operator room part;

model y=;

random kit operator*lab part(kit);

run;

Thanks!

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

08-25-2011 04:14 PM

I wanted to correct the class statement. It should read:

class kit operator lab part;

Please share your thoughts on what is the best way to setup the model.

Thanks!

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

08-26-2011 10:17 AM

would you be able to show us your data (or at least a subset of it)?

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

08-26-2011 07:18 PM

I am afraid I won't be able to share the data. I realize that I will have to run diagnostics to make sure the model is correct and also modify the mdoel in case there are convergence issues. But, I just want to setup a model that is correct for the design. Thanks

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

08-29-2011 12:31 PM

no problem. i am just having a hard time visualizing your data structure. hopefully some of the other users on here can chime in. can you show some made up dummy data (something that shows your data structure)?