Response Screening - Magnitude in Y

reallyneedhelp

Hi everyone.

With regards to the response screening, supposed you have a categorical X (factor) and numerical Y (response), thereby fitting Y by X analysis using Oneway, do the magnitude (size of number) matter in Y? I have e^-20 values in each data, but the values (PValue, LogWorth, RSquare, etc) are being listed as null when I do the response screening. As compared to Python, there are values for PValue and RSquare.

Can somebody confirm to me that this is due to the magnitude in Y being too small? Any comments will be appreciated.Thank you!

Dan_Obermiller

The scale should not matter. You can try this easily by performing the analysis in JMP. Now change the scale of your Y by subtracting a constant or choosing the standardize function. Redo the analysis with the modified response. Compare the results.