Options for comparing sample size and method from population dataset

[#1]

I’m looking for options to compare sampling size and methodology in an RCBD population dataset. The dataset is from an animal floor pen study (14 animals/pen = EU). Weight and other variables were collected from 100% of EU.
The idea is to look at the dataset and evaluate different scenarios for selecting samples (Randomly from pen, Randomly _# of animals, witching a stand deviation, _# within a standard deviation; and compare the analysis between.
So far this has been achieved by simulating the sampling method (5 times each) with Excel and exporting subsets to JMP for analysis.
I am sure there is a better more streamlined way to achieve this directly in JMP but I’m not sure what to do. Any help is appreciated!

[#2]

This might not be exactly what you're asking for, but Covariate Factors in the Custom Designer can select optimal samples for an experiment. Here's a Mastering JMP Webinar (the whole video is informative, but subsetting rows is specifically discussed starting at around the 48 minute mark).

[#3]

Thank you for the Link, I'll give it a watch and see if it might help my situation!

[#4]

First, welcome to the community.  I must admit, I don't fully understand your situation.  

1. You are using a RCBD, what are the factors you are manipulating within block?  How do you propose to handle the block effect?  Fixed or random effect?  

2. What do you mean by "sampling size and methodology"?

3. It appears you are confounding Pen with the block.  I could also argue your situation is more aligned with sampling vs. DOE.  You could have multiple layers nested depending on hypotheses to identify sources of variation in the study.

4. I'm not sure how simulation would help as I'm not sure how variance is being estimated.

What is needed is to know: What questions you are trying to answer?  What hypotheses do you have?  What are you measuring?  How adequate is the measurement system?  How representative is your study of the true population?

[#5]

Clarification: 

1. RCBD--Feed and Weigh

-- Fixed Effects = Trt  (4 Trts)

--Random Effect = Block (13 Reps/Trt) 

--Response = Live Weight -- additional performance metrics (although these aren't currently of interest) 

2. Collected Data from all 14 Animals From all Pens and analyzed the data using Fit Model-Least Squared Means.

• Now I would like to use the population data and do some "what-if" type scenarios. Say I only collected LW from 1 animal/Pen (randomly and/or within a SD of the Pen Mean) or 6 animals/pen (randomly and/or within a SD of the Pen Mean)
• I want to compare the analysis between these what if scenarios

3. I'm sorry but I'm not entirely sure how to respond to this, but I'll give it a go. 

• I would like to see how the # of animals (sampled)  and method used for sampling (randomly or with a SD of the Pen Mean) impacts the ability to detect differences between treatments; and the amount of variation in output.

4. Simulation (picking diff # of Animal, with the 2 Selection Methods ) were replicated 5 times using excel. 

• The reason we simulated the scenarios, is because the # of animals/pen availible for Random/within SD of Mean avail. for selection was less than the actual pop./pen. 

[#6]

Which version of JMP are you using (standard or Pro)?

[#7]

Standard JMP 17

[#8]

There is a new feature in JMP 17 that might help. See the Design Explorer documentation. This analysis is based on the assumptions of linear regression rather than a Monte Carlo simulation.