Hello all,
I have created a custom design matrix: 4 factors with 4 levels, and 4 factors with 2 levels (all categorical). There are some combinations that I need to disallow; however, I can't seem to get this to work.
I am using Jmp 8.0 for the Mac, and the expression I use to specify my constraint is:
IS == 4 & ID == 2
I am using the ordinal value of the level instead of the name of the level, as the DoE Guide states.
When I then 'Make Table', the constraint is not incorporated. It doesn't work with the value label either, or using any other numaeric and boolean combinations.
I have attached an image of the procedre for clarity.
Any ideas would be greatly appreciated.
Many thanks,
Christina
If I dropped all of the two-way interactions that involve the two factors being dissallowed, namely IS and ID, from the model before attempting to generate the design then I found that no error message was generated and the design that results respects the desired dissallowed combinations.
Lou V
I tried this in JMP 8.02 on the Mac and it worked just fine. It did not give any problems using dissallowed combination of IS==4 and ID==2
Hi - thanks very much for checking on this for me. I tried this design on another mac also running 8.02, and still couldn't get it to work. So it's something I'm doing wrong.
This is my sequence of steps:
I only specify 'disallowed combinations' once the design matrix has been built. If I try to do it before I click the 'Make Design' button, I get the error message: "Could not find a valid starting design. Please check your constraints for consistency."
1. After the design matrix is created, I click on the drop down menu Custom Design.
2. Under 'Disallowed Combinations' I enter my restrictions.
3. I then click the 'Make Table' button.
4. Both the matrix gets re-made, and a table is created, both still featuring the combination I wish to disallow.
I'm not certain where I am going wrong.
Any insight would be greatly appreciated.
Many thanks,
Christina
No problem.
On my Mac JMP 8.0.2
1.) I enter all of your factors, 4 X 4-level categorical ( None, Close, Far, Both) and 4 X 2 Level categorical factors (None and High)
2.) Then under the red triangle I choose disallowed combinations (IS==4 & ID==2) Click "OK"
3.) Click "Make Design"
4.) Click "Make Table"
Verify by Analyze Distribution of IS and ID that the disallowed combinations have been respected.
Not sure why you are getting the error message? I would re-enter the factors and try again from scratch.
Lou V
Here is the verification after Making Design that the dissallowed combination constraint was respected.
Hello again,
Thank you so very much for trying this out for me in more detail! I, too, get the same results as you when specifying main effects.
My problem was when I tried to disallow combinations when I specified 2nd order interactions - I suppose it is not something one can do when investigating higher-order interactions.
I apologize for not stating this in the original query. However, you have done a great deal to help me understand the entire process, and its verification.
Thanks again!
Christina
If I dropped all of the two-way interactions that involve the two factors being dissallowed, namely IS and ID, from the model before attempting to generate the design then I found that no error message was generated and the design that results respects the desired dissallowed combinations.
Lou V
Yes!! That works! Well done. Thank you!
Now: try 2 disallowed combinations with 2nd order interactions. For example:
IS==1 & ID==1 | SS==1 & SD==1
I've tried to drop the interactions like you suggest, but run into the same error. I am probably not dropping the right ones. I've tried different combinations but no luck.
Any thoughts? Although please don't feel the need to continue helping - you have already done so much!
Many thanks,
Christina
Great!
I was able to get 2 dissallowed combinations to work. I pulled out all 2nd order terms with IS, ID, SS & SD included.
And verification
Lou V
Hello!
Excellent work. I can't thank you enough for your help.
Onwards and upwards, now!
Thanks again,
Christina