Hi Team,
Need your help. I received this question from a professor who is teaching DOE.
4 factors: organic (A), packing(B), aqueous(C) and solvent(D). Professor has considered fractional factorial (resolution 4, no Blocks )with one replicate. (so 16 runs). He has also deselected the +/- in changing rules.
He also notes the negative aliasing because of I=-ABCD (for second fraction) and AB=-CD for instance. Here is the Aliasing Matrix.
Now he looks into Table of Correlations (as mentioned below)
He looks in to correlation between AB and CD. His argument is that the Should it not be -1? i.e. AB is exactly correlated with - of CD. However the table shows +1.
If I was ABCD (first fraction), then AB=CD and correlation between these two would be 1.
Professor is looking at reconciliation and trying to understand JMP here.
your input would be appreciated
Thanks
Murali
Hi Murali,
The table of correlations is a tabulated form of the Color Map on Correlations, which is showing absolute correlation |r|, so it is always positive. In most cases it does not matter whether correlation of effects is positive or negative.
Personally I only ever look at the color map, not the alias matrix. But I'm not in teaching DOE in academia.
This is off-topic but replicating the half factorial seems like a strange choice. With 16 runs you can do the full factorial.
Hope that helps,
Phil
Thanks a lot @Phil_Kay . This helps.
Great point on replicating the fractional factorial. I know this professor who wants to apply matrix algebra in the class and show a lot of derivations in his DOE and Statistics class.
Thanks a lot Phil
Murali
