turn on suggestions

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

Showing results for

- JMP User Community
- :
- Discussions
- :
- Discussions
- :
- Quick question about screening design

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
- Permalink
- Email to a Friend
- Report Inappropriate Content

Mar 14, 2014 3:27 AM
(1071 views)

When making a simple screening design including only main effects the model still presents (in my case two) 2xinteractions in the Fit Model, Model Specification, Construct Model Effects box. Why is this?

3 REPLIES

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

Mar 14, 2014 6:29 AM
(956 views)

When you use the screening design platform the model shown originates from a "catalog of designs". Lets use a 4 factor example. There are enough degrees of freedom to fit main effects (4) and also 3 interactions (X1*X2)=(X3*X4), (X1*X3)=(X2*X4) and (X1*X4)=(X2*X3). The main effects are clear but the two-way interactions are confounded or aliased with the other two-way interactions. If you desire the model without the interactions just remove the interaction terms from the model dialog. Alternatively you could use the custom design platform directly where only the main effects will be shown for the model if so specified.

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

Mar 17, 2014 12:23 PM
(956 views)

Thank you for your help, truly appreciated.

Am I on the right track if I´m thinking that by the three interactions from your example the confounded interactions wouldn´t say anything of the particular interactions included, but if significant would say that I do have one or more significant 2-way interactions in my model? If I am on the right track, would I get different information depending on how many of the three interactions are significant?

Or can I get any other information from these confounded 2-way interactions?

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

Mar 17, 2014 5:36 PM
(956 views)

It depends. Many time the interactions if active can be "teased-out" of the model due to effect heredity principles. Alternatively if you want to be sure of identifying the correct active interactions an augment to your design that includes the complimentary set of experiments can be run. Domain knowledge and verification runs can also be helpful at helping to clarify.