Subscribe Bookmark RSS Feed

brand new to jmp pro 11 and have a question regarding a hierarchical dataset. I am attempting to determine what each level's contribution is to the total variance. could someone please take a moment to get me started ?

white_535_hotma

Community Trekker

Joined:

Apr 25, 2015

brand new to jmp pro 11 and have a question after reading  a bit of using jmp, discovering jmp, and ch 4 on cluster analysis from the book multivariate methods. i have a hierarchical dataset and am attempting to determine what each level's contribution is to the total variance. could someone please take a moment to get me started ? in great appreciation, jw

3 REPLIES
Peter_Bartell

Joined:

Jun 5, 2014

In a general sense you should be using the Mixed Models platform for your analysis. The exact structure of your model will depend on the nature of the hierarchical nature of your context in which you collected the data. Fixed vs. random effects. Repeated measures? Or other considerations. I suggest as a start reading through the JMP Pro documentation in the online Help regarding Mixed Models and see if you can gather some hints and practices from there.

white_535_hotma

Community Trekker

Joined:

Apr 25, 2015

Peter

cant thank you enough for getting back with me . I will work on this a bit this afternoon and let you know what I have come up with . all the best and thanks again , jw

Peter_Bartell

Joined:

Jun 5, 2014

I was remiss in my earlier reply to also encourage you to PLOT the data in various forms BEFORE doing any modeling work. The Fit Y by X platform, perhaps the Multivariate platform, and if nothing else, if you have a time stamp or order of production/generation of the lowest level of the hierarchy for the specific measurements. Plot the response(s) in time sequence too. You did not mention if your data came from a designed experiment or not? That too will influence how you build your models...say if you had a blocking factor etc. Lots to consider when analyzing data for variance components analysis.