What accounts for most of my error?

[#1]

I'm trying to find where the error is coming from in my experiments.  But I have multiple points where errors can occur.  Is there a way to determine which part of my experiment is responsible for the most error?  I have attached an example which has three sources of error:  Experiment, sample and measurement.  All of these contribute to the differences in the value I receive.  I'm hoping for an explanation or maybe a script on how I can set up an analysis looking at the source of error.  Thank you.

[#2 · Accepted solution]

Dear @wyler0 ,

I'm not sure whether I interprete your data right, 

you have 16 measurements,

done with

• 2 samples,
• 2 experiments
• 2 measurements
• and 2 repititions

and you expect the same value for each?

May be you want to have a look at the variability platform, with variance components, see screenshot.

It looks that your repeatability is very poor, it accounts for 75 % of the variations.

See the manual and script for the analysis attached:

Variance Components (jmp.com)

So you have to find other sources or a suitable measurement first.

Variability Chart(
	Y( :Value ),
	X( :Experiment, :Sample, :Measurement ),
	Model( "Main Effect" ),
	Variance Components( 1 ),
	SendToReport(
		Dispatch( {}, "Variability Gauge", OutlineBox, {SetHorizontal( 1 )} )
	)
)

[#3]

Thank you Georg, this is exactly what I was looking for.