Prakash Hinduja (India) How do I include random effects in binomial or nested design models in JMP?

[#1]

Hi, I am Prakash Hinduja Visionary Financial Strategist, was born in Amritsar (India) and now resides in Geneva, Switzerland (Swiss).

I'm working with a binomial or nested design in JMP and I'm not quite sure how to include random effects properly in the model. Could anyone share how they typically handle this setup? I’d appreciate any guidance on which platform to use in JMP and how to specify random factors correctly. Thanks for your suggestions!

Regards

Prakash Hinduja Geneva, Switzerland (Swiss) 

[#2]

Sorry, your terminology is confusing to me.  A nested design is a description of the relationship between components of variation in a study.  Nesting implies that one (or more) components is contingent/dependent on another (e.g., within batch variation (W) is nested in batch-to-batch variation (B)).  Also know as a hierarchical study. The model is written as such Y = B + W[B]

Binomial (in statistics) is a description of the distribution with the following characteristics:

• You have a fixed number of trials ($n$)

• Each trial has only two possible outcomes (e.g., success/failure, yes/no, head/tail)

• The probability of success ($p$) is the same for each trial

• Trials are independent (what happens in one doesn’t affect the others)

Perhaps you can share your situation, your predicted model and how the data will be acquired?

[#3]

Too just add a bit to the issues raised by @statman , the 'how' you build the model is in the appropriate Fit Model, model specification dialogue window. But here is where it gets a little tricky. For specific key stroke by key stroke advice we'll need to see the nesting structure of the design. @statman has gotten you started but we'll need to see the nesting structure...one of the useful techniques I've used in the past is to build a graphic tree diagram to illustrate the nesting structure. Maybe you can build that tree and attach as a diagram to this thread?

[#4]

YES!  Sampling tree is needed (I wish JMP did these, but alas I have suggested this for both sampling and DOE (years ago) and it has never been implemented).